# Beeble

> Beeble is an AI VFX platform for filmmakers and video professionals — production-grade tools (SwitchX video transformation, Canvas node-graph compositing, VFX pass generation, relighting, background removal, SDR to HDR) and top generative video models in one cloud workspace, plus Beeble Studio for fully on-device work.

This is the full-content companion to https://beeble.ai/llms.txt: the beeble.ai landing pages as an index, followed by the full text of every published blog, research, showcase, and compare page. The product runs at https://app.beeble.ai; the product docs and developer API publish their own llms-full.txt files at https://docs.beeble.ai/llms-full.txt and https://developer.beeble.ai/docs/llms-full.txt.

## Site Pages

- [Beeble — AI VFX for Filmmakers](https://beeble.ai/): AI VFX for filmmakers — everything your shot needs to become film. Production-grade tools and every top model, with full creative control.
- [SwitchX](https://beeble.ai/features/switchx): Swap backgrounds, wardrobe, and scenes while shape, motion, and expression stay locked. SwitchX relights and restyles your footage to blend seamlessly.
- [Reframe](https://beeble.ai/features/reframe): Expand any image or video into a new aspect ratio with SwitchFrame video outpainting. The original stays intact, frame for frame. Up to 20 seconds, 4K, 10-bit color.
- [Canvas](https://beeble.ai/features/canvas): Wire SwitchX, leading external models, and compositing into one node graph. Generate, composite, and branch variations at any step.
- [SDR to HDR](https://beeble.ai/features/sdr-to-hdr): Bring in an 8-bit clip and get back a real 16-bit HDR EXR sequence — blown highlights rebuilt, crushed shadows cleaned. Powered by SwitchHDR.
- [Background Remover](https://beeble.ai/features/background-remover): Remove backgrounds from video, images, and image sequences with no green screen. Preview and refine the alpha matte, then export transparent RGBA footage for any editor.
- [AI Rotoscoping](https://beeble.ai/features/ai-rotoscoping): AI rotoscoping instead of manual spline work — pull a clean matte from any footage in seconds, refine tough frames by hand in Edit Alpha, and export mattes for Nuke, After Effects, or DaVinci Resolve.
- [VFX Pass Generator](https://beeble.ai/features/vfx-pass-generator): Generate production-ready render passes from any video with SwitchLight 3.0 — PBR maps, normal and depth maps, and clean AI-rotoscoped alpha mattes. Exported per pass for Nuke, Blender, Unreal Engine, or any DCC.
- [3D Relight](https://beeble.ai/features/3d-relight): Relight real footage with HDRIs, point lights, and video-driven lighting in a real-time browser 3D scene. Physically based results with no shaders, no calibration, no installs.
- [Upscaler](https://beeble.ai/features/upscaler): Upscale images to 512MP and video to 4K with Topaz Gigapixel AI, Topaz Video AI, and Magnific. Batch up to 20 files and interpolate frames to 60fps.
- [Image Generator](https://beeble.ai/features/image-generator): Generate with Nano Banana 2.1, GPT Image 2.5, FLUX 3, Seedream, and more in one workspace. Text to image or image to image, up to 4K.
- [Video Generator](https://beeble.ai/features/video-generator): Generate AI video with Seedance 2.5, Kling 3 Omni, Gemini Omni Flash 1.1, MiniMax H3, Wan 3.0, and more. Text to video, image to video, reference to video, up to 4K — then finish the shot in Canvas.
- [3D Scene Generator](https://beeble.ai/features/3d-scene-generator): Generate explorable 3D scenes with World Labs Marble from a text prompt, images, a video, or a panorama. Walk through, capture up to 4K stills, export the Gaussian splat for Unreal and Blender.
- [Seedance 2.0](https://beeble.ai/models/seedance-2-0): Generate with Seedance 2.0 on Beeble. Every tier: Standard, Fast, and Mini. Native 4K, 15-second clips, 12 mixed references, and a full compositing graph around it.
- [Seedance 2.5](https://beeble.ai/models/seedance-2-5): Seedance 2.5 on Beeble — 30 seconds of cinema in one pass, native 4K, sound included.
- [Kling 3.0](https://beeble.ai/models/kling-3-0): Generate with Kling 3.0 and Kling 3 Omni on Beeble. Text, image, reference, or video in — native 4K with sound, 3 to 15 second clips, and video-to-video editing in the same graph.
- [MatAnyone 2](https://beeble.ai/models/matanyone): Run MatAnyone 2 video matting online — no ComfyUI, no Colab, no GPU. Stable alpha mattes from any footage at up to 4K/60fps, exported as RGBA for any editor.
- [CorridorKey](https://beeble.ai/models/corridor-key): Corridor's AI green screen keyer, licensed on Beeble — true foreground reconstruction with zero spill, in the browser with no plugin and no GPU.
- [Beeble Cloud Pricing](https://beeble.ai/pricing-cloud): Canvas, SwitchX, Reframe, Background Remover, VFX Pass Generator, 3D Relight, SDR to HDR, Upscaler, and built-in image and video generators. Frame-accurate output, flexible credits, full commercial use.
- [Teams & Enterprise Pricing](https://beeble.ai/pricing-cloud/organization): Scale Beeble across your studio with Team and Enterprise plans. Shared credit pools, central billing, seat and role management. Enterprise adds unlimited seats, SSO, audit logs, and SOC 2.
- [Beeble Enterprise](https://beeble.ai/enterprise): Unlimited seats, flexible credits, AI model permission controls, and dedicated Slack support. Every AI VFX tool behind one admin dashboard, billed by invoice, PO & wire.
- [Beeble Studio](https://beeble.ai/beeble-studio): A local VFX pass generator and relighting tool — 4K output, local AI models, and total privacy built in. Fully on-device.
- [Creative Partner Program](https://beeble.ai/creator-program): Monthly credits, early access to new models, and a direct line to the team. Built for creators making real work.
- [Affiliate Program](https://beeble.ai/affiliate-program): Earn 20% recurring commission on every paid referral, for up to 12 months. Free to join, built for creators and studios.
- [Downloads](https://beeble.ai/resources/downloads): Download Beeble Studio for local VFX pass generation and relighting on Windows and Linux, and Beeble Camera for relight-ready, camera-tracked footage on iPhone.
- [Brand Kit](https://beeble.ai/brand): Official Beeble logo files, brand color and usage rules. Download the full brand kit or individual SVG and PNG logos.
- [Updates](https://beeble.ai/updates): Release notes and changelog — every model, tool, and workflow update shipped to Beeble.


# Preparing the next generation of filmmakers: Inside ETC
Source: https://beeble.ai/blog/preparing-next-generation-filmmakers-inside-etc-usc
Published: 2026-09-04

Inside USC's Entertainment Technology Center, Erik Weaver and Tom Thudiyanplackal discuss AI filmmaking, education, and production-ready workflows.

Technology has always shaped the way movies are made. From photochemical film to digital cinematography, virtual production, and now artificial intelligence, every generation of filmmakers has adapted to new tools while building on the same foundations of storytelling and cinematic language.

For the [Entertainment Technology Center](https://www.etcenter.org/) (ETC) at the University of Southern California's School of Cinematic Arts, exploring the practical applications of future technologies is central to its mission. Working alongside studios, industry partners, and educators, ETC evaluates how emerging production tools integrate into real-world filmmaking workflows while helping prepare the next generation of storytellers for an industry in constant evolution.

Beeble has been incorporated into several ETC education and research initiatives. We spoke with **Erik Weaver, Principal of AI & Emergent Media**, and **Tom Thudiyanplackal, EP, Producer and Industry Mentor**, about why experimentation matters, how filmmaking education is changing, and why storytelling remains the constant.

## How does ETC evaluate emerging technologies?

**Erik Weaver:** Every major technological shift changes the tools filmmakers use, but our role remains the same. At ETC, we explore the practical applications of future technologies through studio-grade productions and educational programs, helping the industry understand what works today while preparing filmmakers for what's next.

Beeble became part of that exploration because it gives students and researchers a practical way to experiment with AI production workflows while allowing us to evaluate how those workflows fit into professional filmmaking.

## You've said AI is changing the language of production itself. What do you mean by that?

**Tom Thudiyanplackal:** Throughout cinema history, production methods have evolved around the limitations of technology. Film had one set of limitations, digital another. AI introduces something fundamentally different.

Instead of simply capturing a fixed number of takes, filmmakers are increasingly capturing data that gives them the freedom to iterate throughout production and post-production. That changes how we think about production itself.

At the same time, different technologies are becoming specialised for different tasks. Cameras remain essential for capturing performances, but there are now better tools for world-building, environment capture, actor likeness, and other production elements. The challenge becomes understanding how all those tools work together to create the final image.

![Students exploring AI filmmaking workflows at ETC](https://desktop.beeble.ai/public/landing/cms/images/blog/preparing-next-generation-filmmakers-inside-etc-usc/students-exploring-ai-filmmaking-at-etc.webp)

## How is this changing filmmaking education?

**Tom:** Visual effects at this scale would traditionally have been inaccessible to students. We've seen students arrive with ambitious ideas that might once have been considered unrealistic for a class project, but today's tools are allowing them to pursue those creative visions.

We've already seen students develop projects that previously would have been considered unrealistic for a class. Instead of rewriting ideas to match production limitations, they're able to explore more ambitious creative concepts while learning how professional workflows operate.

**Erik:** One of our students questioned why they had even gone to film school after enrolling in an AI filmmaking class. At the end of the course they came back energised. They realised that without understanding storytelling, visual language, and filmmaking fundamentals, the AI tools weren't enough.

Technology can accelerate production, but it doesn't replace the craft of filmmaking. Understanding composition, pacing, performance, and emotion remains just as important as ever.

![USC students learning filmmaking with Beeble production tools](https://desktop.beeble.ai/public/landing/cms/images/blog/preparing-next-generation-filmmakers-inside-etc-usc/etc-filmmaking-education-with-beeble.webp)

## How are the studios you work with approaching AI?

**Tom:** The conversation has shifted from whether AI is used to how it's managed responsibly.

Studios care deeply about provenance, ownership, and understanding exactly how AI fits into a production pipeline. They want to know where assets come from, how they're generated, and how intellectual property is protected.

That's why we're interested in production-ready workflows rather than isolated demonstrations. The goal is to understand how these tools integrate into real filmmaking.

## How does that research extend beyond the classroom?

**Tom:** One example is *Pathways*, a research filmmaking project that allowed us to work with cinematographer [Roberto Schaefer](https://www.imdb.com/name/nm0769656/) while exploring AI-assisted production workflows.

The project investigated how lighting decisions made during production could better support AI later in the pipeline. Rather than treating AI as something that happens only in post-production, we looked at how cinematography, lighting, and virtual production could all be planned together from the start.

It also became a fascinating case study because the project spanned several generations of Beeble's technology, allowing us to see how quickly the tools evolved while we were still building the workflow.

## What advice would you give filmmakers exploring new technologies?

**Tom:** Stay true to your story, not to your tools. Technology will continue to evolve, sometimes faster than a production schedule. Commit to the workflow you've chosen for a project, but always maintain a separate space for experimentation and research. That's how you'll continue learning without constantly restarting your productions.

**Erik:** The fundamentals don't change. Storytelling, cinematography, and visual language are still what audiences respond to. AI simply gives filmmakers another way to realise those ideas.

As filmmaking evolves, ETC's work demonstrates the value of evaluating new technologies through practical application rather than theory. By bringing together research, education and production, the institution is helping shape how the next generation of filmmakers will navigate an industry where the tools continue to evolve, but the principles of great storytelling remain unchanged.

Ready to try Beeble?


# Relight realistic car plates with Beeble Canvas
Source: https://beeble.ai/blog/canvas-tutorial-car-plates
Published: 2026-08-28

A Beeble Canvas tutorial: replace a green screen car plate with a driving background, then relight the whole shot for golden hour with SwitchX.

Car scenes shot against a green screen give filmmakers control over the background, but replacing the plate is only part of creating a convincing final shot. The lighting, reflections, and atmosphere inside and around the vehicle also need to feel like they belong to the new environment.

In this Beeble Canvas tutorial, we use green screen footage of two actors in a car, replace the background with a driving plate, and relight the entire shot for a golden hour look.

Using Video Matte, Canvas compositing, Nano Banana 2, and SwitchX, the workflow creates new lighting and reflections across the windshield, car hood, and subjects, helping the foreground and background feel as if they were captured together.

## Watch the tutorial

*Practice footage provided by ActionVFX.*

## The challenge: Making a car plate feel like part of the shot

Replacing a green screen behind a car is relatively straightforward. Making the resulting composite feel convincing is more difficult. The lighting on the actors and the vehicle needs to match the outdoor surroundings. Reflections on the windshield and hood should respond to the environment, particularly if the car is in motion. If the background suggests warm evening sunlight while the foreground retains the lighting of a green screen stage, the two elements immediately feel disconnected.

Traditionally, productions might address this during the shoot with interactive lighting or an LED volume, or recreate the effect through a more involved compositing workflow.

Here, we first build the composite in Canvas, then use a generated reference image to guide SwitchX in relighting the entire shot.

## Step 1: Generate a video matte

We begin by isolating the actors and car from the green screen.

Inside the Video Matte node, we choose **Select Mode** and open the **Smart Select** tool. **Using Point Select**, we identify the subjects and vehicle while excluding the green screen background.

![The Video Matte node in the Canvas graph with the Smart Select tool open](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/video-matte-smart-select.webp)

![Point Select in the Alpha Editor marking the two actors and the vehicle interior](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/video-matte-point-select.webp)

Once the selection is ready, Video Matte propagates the mask across the clip, yielding a foreground plate that can be composited into a new environment.

The important distinction here is that we are not generating a new car or subjects. We are extracting the original foreground so it can remain the foundation of the final shot.

![The Video Matte results tab showing the extracted foreground plate over a transparent background](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/video-matte-foreground-plate.webp)

## Step 2: Combine the foreground and driving plate

With the matte complete, we bring in the provided driving plate and connect the foreground and background elements to a Compositor node.

The Compositor works as a layer-based editor, allowing us to position the rotoscoped car footage over the driving plate and establish the basic composition before relighting.

![The driving plate clip connected into the Canvas graph alongside the matted car footage](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/compositor-driving-plate.webp)

![The Compositor node combining the foreground and driving plate on two layers](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/compositor-layers.webp)

For some shots, we also add a Blur node to the background plate before it reaches the Compositor. This creates a stronger sense of depth of field and helps separate the subjects and vehicle from the environment behind them.

![A Blur node softening the driving plate before it reaches the Compositor](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/blur-node-depth.webp)

At this stage, the elements are in the right place, but they still carry different lighting characteristics. The next steps are about making them feel like a single image.

## Step 3: Create a golden hour reference image

We now extract a frame from the composite and send it to an **Image Generator** node using Nano Banana 2.

The prompt focuses specifically on the desired lighting: a golden hour treatment with detailed instructions for how the light and reflections should behave across the car.

The background plate is already established, so we want to preserve its details while exploring how the completed shot should look under different lighting conditions.

The resulting image becomes the visual target for SwitchX.

![The full Canvas graph with Extract Frame, Image Generator, and SwitchX branches for each shot](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/image-generator-graph.webp)

![A Text node holding the shared golden hour prompt that defines the target look for all six shots](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/golden-hour-prompt.webp)

A strong reference is particularly important at this stage because it conveys more than a simple description like "golden hour." It establishes how that lighting should interact with the actors, windshield, hood, and other reflective surfaces in the frame.

![The Image Generator node showing the generated golden hour reference image](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/golden-hour-reference.webp)

## Step 4: Relight the shot with SwitchX

With the reference image complete, we connect the composite to SwitchX.

For this workflow, **we leave the alpha input empty.** Without an alpha, SwitchX preserves the details across the entire frame and uses the reference primarily to influence lighting and color, equivalent to using **Fill Mode** in SwitchX.

We also connect a **Text** node containing a simple description of the reference image, providing SwitchX with additional context for the scene. SwitchX then carries the golden hour treatment from the reference into the moving footage.

The effect goes beyond a conventional color adjustment. New reflections appear across the windshield and hood, while the actors and vehicle respond to the warmer environment. The result helps connect the foreground car plate with the driving footage behind it, making the composite feel like a more cohesive shot.

![The SwitchX node with no alpha connected, relighting the composite from the reference image](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/switchx-relight-result.webp)

## Step 5: Assemble the final sequence in Canvas

The same workflow can then be repeated across the remaining shots. For this tutorial, six shots are processed using the same overall approach and golden-hour direction, helping to maintain consistency across the sequence.

Once each shot is complete, we connect them to a Compositor node and arrange them on its layer-based timeline. This makes it possible to play back and edit the finished sequence directly inside Canvas without leaving the project.

![The six-shot Canvas graph with source, matte, plate, composite, reference, and SwitchX columns feeding one sequence](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/six-shot-graph.webp)

![The Compositor node playing back all six relit shots on separate layers](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/sequence-compositor.webp)

![The Compositor timeline with the six relit shots arranged in order](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-car-plates/compositor-timeline.webp)

## Credits usage breakdown

### Shot 1

- Video Matte = 5 Credits
- Image Generator (Nano Banana 2) = 4 Credits
- SwitchX = 50 Credits

### Shot 2

- Video Matte = 4 Credits
- Image Generator (Nano Banana 2) = 4 Credits
- SwitchX = 40 Credits

### Shot 3

- Video Matte = 3 Credits
- Image Generator (Nano Banana 2) = 4 Credits
- SwitchX = 30 Credits

### Shot 4

- Video Matte = 4 Credits
- Image Generator (Nano Banana 2) = 4 Credits
- SwitchX = 40 Credits

### Shot 5

- Video Matte = 3 Credits
- Image Generator (Nano Banana 2) = 4 Credits
- SwitchX = 30 Credits

### Shot 6

- Video Matte = 4 Credits
- Image Generator (Nano Banana 2) = 4 Credits
- SwitchX = 40 Credits

### Total

- 277 Credits = approx $9.23

## Why this workflow matters

Car scenes are a good example of why background replacement alone is rarely enough.

Vehicles contain reflective glass, glossy bodywork, and complex surfaces that respond visibly to their surroundings. When a background changes, those interactions need to change too if the shot is going to feel believable.

This Canvas workflow brings those stages together. Video Matte handles the foreground extraction, the Compositor establishes the new environment, an Image Generator creates a clear visual target, and SwitchX carries that lighting direction back into the moving footage.

For filmmakers and VFX artists working with green screen car footage, it provides a way to explore different environments and lighting conditions while preserving the original performances, vehicle, and photography.

All assets used in the tutorial are available on Beeble as a Canvas project template, so you can open the workflow and experiment with it directly.

### Ready to relight your own car plates in Canvas?


# Let's bring Beeble to SXSW 2027
Source: https://beeble.ai/blog/sxsw-2027-vote
Published: 2026-08-18

Help us bring Beeble to SXSW 2027. Vote for our PanelPicker session on creative control in the age of AI before voting closes on August 23.

**How do filmmakers stay in control in the age of AI?**

We've submitted a session for SXSW 2027 exploring one of the biggest questions facing filmmakers as AI becomes part of production and post: not simply what these technologies can create, but who gets to make the creative decisions.

In **Creative control in the age of AI**, Beeble CEO Hoon Kim and Head of Production and Partnerships Conrad Curtis will explore how AI can extend creative control without replacing the people behind the camera.

Through real-world production examples, they'll examine why cinematography becomes more valuable in AI workflows, how creative decisions can extend beyond capture, and practical approaches to preserving artistic authorship.

**Want to see the session at SXSW 2027?**

PanelPicker community voting is open until **August 23 at 11:59pm PT**.


# Controlling the frame: what we learned at SIGGRAPH 2026
Source: https://beeble.ai/blog/siggraph-2026-recap
Published: 2026-07-24

Inside our packed Birds of a Feather session at SIGGRAPH 2026 — the four levels of creative control professionals expect from AI, and productions using them.

Last week at SIGGRAPH 2026, we were delighted to see a packed room for our Birds of a Feather session, "Controlling the Frame: AI VFX, Generative Video, and the Future of Compositing".

The room was over capacity before we'd even started. People sat along the walls and stood at the back for the full 69 minutes, and when we opened the floor for questions, we had to cut the queue off after 31 minutes with hands still up. It's the best problem a speaker can have, and we're still thinking about the conversations that followed.

What played out in that room wasn't simply a discussion about AI; it was a conversation about something far more important: creative control.

That became the central theme of the session, led by Beeble CEO Hoon Kim and Production and Partnerships Lead Conrad Curtis, bringing together filmmakers, VFX artists, researchers, and developers.

![A full room at the Beeble Birds of a Feather session during SIGGRAPH 2026](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/siggraph-2026-bof-room.webp)

## Three years in a hallway

Hoon opened with the origin story: the idea for Beeble started three years ago, in a hallway not far from where we were standing to give the talk. It's a small detail, but it set the tone for the hour that followed — this wasn't a pitch built in a boardroom, it was built by people who'd spent years in production trying to solve their own problem.

## AI is changing compositing

Much of today's conversation around generative AI focuses on speed. But professional productions have never been driven by speed alone. They are driven by repeatability, collaboration, consistency, and the ability to refine creative decisions over hundreds or thousands of shots.

That means AI has to become part of an existing production pipeline. Traditional compositing offers precise control but can be time-intensive. Fully generative systems offer incredible speed but often sacrifice predictability. The opportunity lies in combining the strengths of both approaches.

One slide summarised this simply: It's always both.

![Session slide contrasting traditional compositing with fully generative systems](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/its-always-both-slide.webp)

## Professionals want control

Throughout the session, we introduced what we see as the four levels of creative control that artists increasingly expect from AI systems.

### 1. Pixel control

Can you decide exactly what changes, and what stays untouched?

Professional artists don't want AI repainting an entire frame every time they make a change. They need reliable subject preservation so performances, costumes, and subtle details remain intact while environments or lighting evolve.

### 2. Frame control

Can a single reference image direct the look of an entire shot?

Reference imagery should become creative direction, allowing artists to establish mood, production design and lighting without losing the original performance.

### 3. Shot control

Can the generated world stay coherent as the camera moves?

Maintaining consistency through motion remains one of the hardest problems in generative video. Camera tracking, depth understanding, and temporal consistency become essential.

### 4. Sequence control

Can the world remain consistent across an entire production?

Artists need environments, lighting and production design that remain stable from shot to shot while still allowing creative iteration.

Professionals want all four.

![Session slide showing the four levels of creative control: pixel, frame, shot, and sequence](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/four-levels-of-control.webp)

## Real productions are already showing what's possible

Rather than focusing on technology demonstrations, the session highlighted four very different productions that illustrate how controllable AI is already entering professional workflows.

### Hellbound: preserving the craft

One of the strongest examples came from the Korean feature Hellbound.

Instead of relying entirely on AI-generated environments that felt short of resembling true 18th century Joseon Dynasty architecture, the production team traveled to real historical locations to capture photographic references and assets. Those real-world images then guided AI generation for reference images, ensuring historical accuracy while preserving the authenticity that purely synthetic imagery struggled to achieve.

Beeble was also used in real-time on set to quickly test shots, while honoring and amplifying the intent of hair and makeup artists, wardrobe stylists, gaffers, and crew.

Real locations, real photography, and real performances remained the foundation. It was the only showcase of the session to earn applause on its own — the room responding not to the generation, but to the fact that the performance had been preserved.

![Hellbound location photography used as reference for AI generation](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/hellbound-reference-1.webp)

![Hellbound reference imagery guiding a Joseon Dynasty environment](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/hellbound-reference-2.webp)

### The fire station piece: playing inside the frame

Conrad walked through a separate piece shot at a fire station, using it to show how far pixel control now extends: a subject given a costume change and, as he put it, "a little extra workout" — entirely inside SwitchX, on the same plate, without re-shooting. It's a small example, but it's the clearest illustration of the pixel control principle in practice: change the details, keep the performance.

## XGRIDS: the scan becomes ground truth

Another showcase explored combining Gaussian Splat captures with generative video. Instead of generating environments from scratch, spatial scans become reliable "ground truth", allowing artists to preserve accurate geometry while using AI to modify appearance.

This points towards a future where spatial capture and generative AI complement one another rather than compete.

![XGRIDS Gaussian Splat capture used as spatial ground truth](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/xgrids-gaussian-splat-1.webp)

![Gaussian Splat geometry preserved while AI modifies scene appearance](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/xgrids-gaussian-splat-2.webp)

![Spatial scan combined with generative video in the XGRIDS workflow](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/xgrids-gaussian-splat-3.webp)

## NoiseLab: generating worlds that stay under control

The session also explored emerging workflows where entirely new environments are created using AI while remaining editable inside traditional 3D tools.

Instead of losing creative flexibility, artists retain camera control, composition, and scene management throughout production.

![NoiseLab AI-generated environment remaining editable in a 3D scene](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/noiselab-generated-world-1.webp)

![Camera control and composition retained over a generated world](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/noiselab-generated-world-2.webp)

![Scene management for an AI-generated environment inside traditional 3D tools](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/noiselab-generated-world-3.webp)

## ETC's Pathways: AI inside a professional compositing pipeline

Entertainment Technology Center's *Pathways* project showed the tools being used in a compositing pipeline.

- AI became another stage within a professional compositing workflow in Nuke.
- SwitchLight generated physically based rendering passes.
- SwitchX handled controllable video transformations.

Traditional compositing techniques, including frequency separation, relighting, and finishing, completed the pipeline.

The takeaway was clear: AI works best when integrated into established production practices.

![ETC Pathways node graph with AI as a stage inside a Nuke comp](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/etc-pathways-nuke-1.webp)

![SwitchLight physically based rendering passes inside the Pathways pipeline](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/etc-pathways-nuke-2.webp)

![SwitchX controllable video transformation in the Pathways comp](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/etc-pathways-nuke-3.webp)

![Frequency separation, relighting, and finishing completing the Pathways pipeline](https://desktop.beeble.ai/public/landing/cms/images/blog/siggraph-2026-recap/etc-pathways-nuke-4.webp)

## Better tools, not bigger models

Towards the end of the discussion, the conversation shifted away from today's models and towards where AI VFX is heading. We shared several ideas that resonated strongly with the audience:

- Domain experts building domain-specific AI models.
- Smaller, more capable creative teams.
- AI systems that understand the broader context of an entire production.
- Above all, greater creative control.

## AI is a pass, not a final

Hoon's own summary of the argument, delivered near the end of the talk: AI becomes another pass. Not a replacement for the compositor's pass, the colorist's pass, or the artist's pass — one more stage in a pipeline that still ends with a human making the final call.

The final result still comes from artists making creative decisions, refining shots, integrating multiple techniques, and shaping every frame with intent.

That philosophy sits at the heart of how we think about controllable AI VFX. And on where the technology itself is headed, Hoon left the room with a line worth repeating: this is the worst these tools will ever be. Everything from here looks small in the rear-view mirror.

## Thank you, SIGGRAPH.

Thank you to everyone who joined us for the discussion, asked thoughtful questions, and shared their own experiences. The full room reinforced something we've believed for a long time: The future of AI in filmmaking will be defined by how much creative control artists retain.

And that's a conversation we're excited to continue.


# Conrad Curtis on AI filmmaking at ACM SIGGRAPH
Source: https://beeble.ai/blog/siggraph-2026-conrad
Published: 2026-07-01

Beeble's Conrad Curtis at ACM SIGGRAPH on how AI relighting, spatial capture, and virtual production are reshaping modern filmmaking workflows.

This week, Beeble's Head of Production and Partnerships, Conrad Curtis, joined filmmakers and technologists at ACM SIGGRAPH Los Angeles to discuss one of the fastest-changing areas in production: how AI, spatial capture, and virtual production are reshaping filmmaking workflows.

Drawing on more than 15 years in physical production, Conrad explored how emerging tools can fit into existing production pipelines, giving filmmakers more flexibility before, during, and after a shoot.

## Building workflows that solve real production challenges

As a DGA Assistant Director, founder of VPX Lab, and Production and Partnerships Lead at Beeble, Conrad has spent the last several years testing practical production technologies alongside working filmmakers.

His presentation focused on a simple idea: every new technology should solve a real production problem.

The session demonstrated how AI, 3D scanning, or virtual production can work together to create more efficient, controllable workflows while preserving creative intent.

## How quickly the technology has evolved

One of the presentation's themes was just how rapidly production technology has changed.

Only a few years ago, creating accurate digital replicas required days of LiDAR scanning, manual point cloud registration, and extensive optimization before assets could be used in production.

Today, portable spatial capture devices can generate high-quality scans in minutes, making location digitization practical for a much wider range of productions.

## Spatial capture meets AI VFX

The presentation also explored how spatial scans can complement AI-assisted visual effects.

Using examples from Beeble SwitchX, Conrad demonstrated workflows where filmmakers combine:

- Spatial scans for accurate scene geometry.
- AI-generated physically based rendering (PBR) passes.
- Semantic masking and AI roto.
- Reference-based generation for environment replacement and relighting.

Instead of rebuilding an entire scene from scratch, these workflows allow filmmakers to preserve the original performance while introducing new environments, lighting conditions, or visual effects with greater control.

## Keeping filmmakers in control

A recurring message throughout the talk was that AI should increase creative control, not reduce it.

Whether using SwitchX to relight footage or combining spatial scans with generative tools, the goal is to give cinematographers, VFX artists, and directors more options after capture while maintaining continuity and respecting the creative decisions made on set.

As AI capabilities continue to evolve, the conversation is shifting away from simply generating images toward building production-ready workflows that integrate with existing filmmaking practices.

For Beeble, that means developing tools that fit naturally into modern production pipelines, helping artists spend less time on technical bottlenecks and more time refining the creative result.

We'd like to thank ACM SIGGRAPH Los Angeles for hosting the event and everyone who joined the discussion.

Interested in exploring AI relighting and reference-based VFX workflows for your own productions?


# Turn an empty stadium into a packed crowd with AI VFX
Source: https://beeble.ai/blog/canvas-tutorial-crowd-vfx
Published: 2026-06-25

See how Beeble Canvas turns an empty stadium into a packed, cheering crowd using generative AI, compositing, and SwitchX color matching.

![Empty stadium transformed into a packed crowd scene with Beeble Canvas](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/crowd-vfx-hero.webp)

Creating a convincing stadium crowd has traditionally required significant resources. Productions often rely on hundreds of extras, crowd replication techniques, green screen shoots, or extensive visual effects to create the feeling of a live event.

In this Beeble Canvas tutorial, we transform an empty stadium into a packed venue using generative AI, compositing, and SwitchX. The result is a shot that feels like it was captured during a live match.

## Watch the tutorial

*Practice footage provided by ActionVFX.*

## Starting with an empty stadium

Our goal is to add a large, active crowd to the stands while preserving the original player, camera work, and overall visual style.

The workflow is built around three stages: generating a crowd-filled background, isolating the player, and using SwitchX to match the final composite back to the original footage.

## Step 1: Generate the crowd background

We begin by trimming the shot to 238 frames and extracting the first frame. This frame gives us a clean view of the empty stadium before the player enters the scene.

![Clean first frame of the empty stadium before the player enters](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/empty-stadium-first-frame.webp)

![Canvas Image Generator node configured to fill the stands with a crowd](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/image-generator-node.webp)

Using Nano Banana with an Image Generator node, we generate multiple versions of the frame with the stadium filled with active crowds.

After reviewing the results, we select the strongest crowd-filled image and use it as the foundation for the next step.

![Multiple Nano Banana crowd-filled variations of the stadium frame](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/nano-banana-crowd-options.webp)

To bring the scene to life, we extend the image into video using Image-to-Video models. We test three different options: Seedance 2.0, Happy Horse, and Kling 3.0. Across all three models, we use a prompt that asks for a static, fixed-angle tripod shot, a massive cheering crowd filling the stadium, no players introduced into the scene, and a photorealistic 4K look.

![Image-to-Video model options tested: Seedance 2.0, Happy Horse, and Kling 3.0](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/image-to-video-models.webp)

Each model produces a slightly different interpretation. Seedance and Happy Horse both deliver strong results, but Kling 3.0 produces the most convincing crowd movement and becomes our final background plate.

![Kling 3.0 crowd movement chosen as the final background plate](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/kling-crowd-background.webp)

![The empty stadium transformed into a packed, energy-filled venue](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/crowd-filled-stadium.webp)

At this point, we have transformed an empty stadium into a packed, energy-filled venue.

## Step 2: Isolate and composite the football player

With the crowd background ready, the next task is to bring the football player back into the shot. Because the player is not visible in the opening frame, Auto Mode in Video Matte is not the right choice. Instead, we use Select Mode and open the Smart Select tool.

![Select Mode and the Smart Select tool open in Video Matte](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/smart-select-tool.webp)

We move to frame 99, where the player is clearly visible, and simply type "man" to select the football player. Video Matte then propagates that selection throughout the entire clip, isolating the player even though the footage was never shot on a green screen.

![Selecting the football player by typing "man" at frame 99](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/player-selection-frame-99.webp)

![Video Matte propagating the player selection across the entire clip](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/isolated-player-matte.webp)

To help the composite feel more natural, we add a small amount of blur to the player layer before merging it with the generated stadium background.

![A small blur added to the isolated player layer before merging](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/player-blur-adjustment.webp)

Using a Merge node, we place the isolated player over the crowd-filled stadium.

![Merge node placing the isolated player over the crowd-filled stadium](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/merge-node-composite.webp)

The scene immediately begins to feel larger in scale, while preserving the original performance and camera movement.

## Step 3: Match the final shot with SwitchX

At this stage, the composite is working, but the color and tone still feel disconnected from the original footage.

The generated stadium background appears cooler, bluer, and more contrasty than the source video. To bring everything back into the visual language of the original shot, we render the final composite through SwitchX.

![Composite where the generated background looks cooler and bluer than the source](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/switchx-color-mismatch.webp)

We select a reference frame from the original footage where both the stadium and the player are visible and send it to SwitchX.

Importantly, we leave the alpha mask empty.

![Reference frame sent to SwitchX with the alpha mask left empty](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/switchx-reference-frame.webp)

Without an alpha mask, SwitchX preserves everything in the frame and focuses on relighting and recoloring the image. This behaves similarly to Fill Mode in the dedicated SwitchX tab, allowing the generated crowd and original player to inherit the color and tone of the source footage.

![SwitchX relighting the composite to match the original footage color and tone](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-crowd-vfx/switchx-final-match.webp)

The result is a crowd-filled stadium that feels naturally integrated into the original scene, rather than added in post-production.

## Why this workflow matters

Crowd scenes have historically been one of the most resource-intensive elements of sports, concerts, and large-scale event productions.

Canvas offers a different approach. By combining generative AI with masking, compositing, and color matching, we can dramatically increase the perceived scale of a scene while maintaining creative control over the final image.

Rather than replacing traditional VFX techniques, this workflow combines generative tools with established post-production practices to create results that remain grounded in the original footage.

For filmmakers, commercial producers, and VFX artists, it provides a practical way to create larger, more cinematic scenes without expanding the production footprint.

## Credits usage breakdown

### Crowd VFX

- Video Matte = 8 Credits
- Image Generator = 4 Credits

### Video Generator

- Seedance = 189 Credits
- Happy Horse = 100 Credits
- Kling 3 = 60 Credits

### Final Composite

- SwitchX = 80 Credits

### Total

- 441 Credits = $14.7

### Ready to create larger crowd scenes in Canvas?


# Add realistic fire VFX while preserving original footage
Source: https://beeble.ai/blog/canvas-tutorial-fire-vfx
Published: 2026-06-24

A Beeble Canvas tutorial: turn an unlit torch into believable fire that relights the scene with SwitchX while preserving the original footage.

![Realistic fire VFX added to a torch in a cave shot with Beeble Canvas](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-fire-vfx/fire-vfx-hero.webp)

Adding fire to a shot is easy. Adding fire that feels naturally integrated into the scene, while preserving the actor, environment, and cinematography, is much harder.

In a new Beeble Canvas tutorial, founder Hoon Kim demonstrates a complete workflow for turning an unlit torch into a convincing fire source while maintaining every important detail in the original footage. The fire illuminates the actor's face, influences the surrounding cave environment, and feels like it was captured on set.

The tutorial showcases how Canvas combines generative AI, masking, and SwitchX's relighting capabilities into a controllable visual effects workflow.

## Watch the tutorial

## The challenge: Fire changes everything

When fire is introduced into a scene, it doesn't simply occupy space. It becomes a light source.

That means the surrounding environment should react accordingly. Faces, clothing, walls, and shadows all need to respond to the new lighting conditions. Traditional workflows often require multiple compositing passes and significant manual work to achieve a believable result.

In this example, the source footage contains an actor holding an unlit torch inside a cave. The goal is to transform the torch into a blazing flame while preserving the actor, the cave, and the original cinematography.

## Step 1: Exploring concepts with AI

The workflow begins with a single frame extracted from the footage. Rather than manually writing prompts, the tutorial uses Gemini to generate multiple prompt variations describing realistic torch fire. The prompts are structured so they can be automatically separated and processed inside Canvas using a Text Iterator node.

This approach turns prompt creation into a rapid ideation process, allowing artists to explore several creative directions before committing to a final look.

![Gemini-generated prompt variations feeding a Text Iterator node in Canvas](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-fire-vfx/step-1-prompt-variations.webp)

## Step 2: Generating reference images

Those prompts are then passed into an Image Generator node using Nano Banana 2.

Canvas generates multiple reference images, each presenting a different interpretation of the desired fire effect. By viewing all options simultaneously, artists can quickly compare results and select the reference image that best matches their creative vision.

The workflow establishes a clear visual target that guides the next stage of the process.

![Multiple AI-generated fire reference images compared side by side in Canvas](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-fire-vfx/step-2-reference-images.webp)

## Step 3: Building a targeted mask

One of the most important parts of the workflow is masking. The mask tells SwitchX exactly which areas should remain untouched and where it is free to introduce new visual elements.

In this example, multiple masks are combined to protect the actor, clothing, and environment while leaving room for the fire to be generated naturally.

A particularly important detail occurs where the torch overlaps the actor's face. The face is excluded from the protected region, ensuring the generated fire can extend into the correct spatial area without creating unwanted artifacts.

The result is a highly controlled generation zone that preserves the integrity of the original shot.

![Combined masks protecting the actor, clothing, and cave environment](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-fire-vfx/step-3-mask-build.webp)

![Mask detail excluding the actor's face so fire can extend into the correct area](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-fire-vfx/step-3-mask-face.webp)

## Step 4: Generating fire with SwitchX

Once the source footage, reference image, and mask are connected, the project is ready for SwitchX. This is where the transformation happens.

SwitchX introduces the fire effect while maintaining the visual information protected by the mask. More importantly, it relights the surrounding scene to account for the new flame source.

The actor's face receives warm highlights. The cave environment reacts to the firelight. Clothing and surrounding surfaces gain realistic illumination.

Instead of looking like an overlay, the fire becomes part of the scene. The same workflow can then be applied across additional camera angles, making it practical for multi-shot sequences.

![SwitchX relighting the cave scene around the newly generated fire](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-fire-vfx/step-4-switchx-result.webp)

## Step 5: Matching the final grade

The final stage focuses on color consistency. Because the tutorial includes both a close-up and a wider shot, the grade is adjusted to ensure the fire feels consistent between angles. Small adjustments to warmth, red tones, and overall darkness help unify the sequence and strengthen the illusion that both shots belong to the same scene.

![Final color grade matching fire warmth across the close-up and wide shots](https://desktop.beeble.ai/public/landing/cms/images/blog/canvas-tutorial-fire-vfx/step-5-final-grade.webp)

All assets and the Canvas project template used in this tutorial are available to try right now on Beeble.

Practice footage provided by ActionVFX.

## Credits usage breakdown

### Shot 1

- LLM = 1 Credit
- Image Generator = 20 Credits
- Video Matte = 8 Credits
- SwitchX = 80 Credits

### Shot 2

- LLM = 1 Credit
- Image Generator = 20 Credits
- Video Matte = 7 Credits
- SwitchX = 70 Credits

### Total

- 207 Credits = $6.9

### Ready to add this technique to your project?


# Join Beeble for a VES discussion on AI relighting
Source: https://beeble.ai/blog/ves-ai-relighting-session
Published: 2026-06-24

Beeble joins the Visual Effects Society on July 14 for a virtual session on AI relighting across VFX, post-production, and virtual production.

![Beeble × VES virtual session on AI relighting in VFX workflows](https://desktop.beeble.ai/public/landing/cms/images/blog/ves-ai-relighting-session/ves-ai-relighting-hero.webp)

On July 14, Beeble will join the Visual Effects Society (VES) for a virtual educational session exploring the growing role of AI relighting across visual effects, post-production, and virtual production workflows.

Presented by **VES Education and Outreach Committee Chair Ben Lock** and **Beeble founder and CEO Hoon Kim**, the session will examine how AI-assisted relighting can help artists generate useful image-based passes, make lighting adjustments after capture, and integrate live-action footage into new environments or lighting conditions.

You can expect a discussion around practical questions facing VFX professionals today:

- Where does AI relighting fit into existing pipelines?
- What level of artist control is required?
- How production-ready are current tools?
- What should studios and artists consider when evaluating AI-assisted workflows?

The session will include a live workflow demonstration, a product overview, and a moderated discussion on artist control, creative intent, and the responsible adoption of AI technologies in professional production environments.

Whether you're a compositor, lighter, VFX supervisor, virtual production artist, or simply interested in emerging AI workflows, we invite you to join the conversation.

**Date:** July 14, 2026 — 5 PM Pacific Time

**Format:** Virtual Event via Zoom

Curious how AI relighting works in practice? Explore [3D Relight](https://beeble.ai/features/3d-relight), Beeble's browser-based relighting editor.


# Controlling the frame: Beeble at SIGGRAPH 2026
Source: https://beeble.ai/blog/beeble-siggraph-2026-controlling-the-frame
Published: 2026-06-18

Join Beeble's Birds of a Feather session at SIGGRAPH 2026 to explore controllable AI VFX, generative video, and the future of compositing.

We'd like to invite you to participate in our Birds of a Feather (BOF) session at SIGGRAPH 2026, bringing together artists, researchers, and technologists to discuss the evolving role of AI in visual effects and post-production.

**[Controlling the Frame: AI VFX, Generative Video, and the Future of Compositing](https://s2026.conference-schedule.org/presentation/?id=bof_129&sess=sess305)**

📅 **Monday, July 20, 2026**

🕑 **2:00 PM – 3:00 PM PDT**

📍 **Room 513**

As AI tools become increasingly capable of transforming images after capture, through relighting, background replacement, look development, and generative scene modification, the conversation is shifting beyond what these systems can do to a more fundamental question:

**How much creative control do artists retain?**

The session will explore the emergence of controllable AI VFX and hybrid compositing workflows, where traditional visual effects techniques intersect with learned systems. Examining how professionals are navigating questions of lighting control, consistency, creative intent, and authorship.

Topics will include:

- AI-assisted relighting and look development.
- Generative video workflows in production and post-production.
- Hybrid compositing pipelines combining traditional and AI-driven approaches.
- Maintaining consistency across shots and sequences.
- Preserving creative intent throughout the image-making process.
- The evolving relationship between artists and intelligent tools.

![Controlling the Frame session details at SIGGRAPH 2026](https://desktop.beeble.ai/public/landing/cms/images/blog/beeble-siggraph-2026-controlling-the-frame/siggraph-2026-session.webp)

The discussion will be led by:

- **Hoon Kim**, CEO and Co-Founder, Beeble AI
- **Conrad Curtis**, Founder, VPX Lab and Production & Partnerships Lead, Beeble AI

Birds of a Feather sessions are designed to encourage open discussion and community participation, making them an ideal forum for exploring the opportunities and challenges surrounding AI's growing role in visual effects.

Whether you're a compositor, VFX supervisor, cinematographer, researcher, pipeline TD, virtual production specialist, or AI developer, we invite you to join the conversation.

We look forward to seeing you at SIGGRAPH 2026!

Until then, experience the next generation of controllable AI VFX with Canvas, our node-based compositing environment built for artists who require creative control.


# Beeble Editor is now 3D Relight
Source: https://beeble.ai/blog/beeble-editor-is-now-3d-relight
Published: 2026-06-17

Beeble Editor is now 3D Relight — same product, same workflow, only the name changed. A browser-based tool for real-time relighting and compositing.

![Beeble Editor is now 3D Relight — relight and composite VFX/PBR passes in a physically based 3D scene](https://desktop.beeble.ai/public/landing/cms/images/blog/beeble-editor-is-now-3d-relight/3d-relight-interface.png)

Beeble Editor is now **3D Relight**. Same product. Same workflow. Only the name changed.

## TL;DR

If you used Beeble Editor before, you already know 3D Relight.

## Why the rename

"Beeble Editor" was too generic and didn't explain what the tool does. **3D Relight** makes it instantly clear: relight and composite VFX/PBR passes in a physically based 3D scene.

[3D Relight](https://beeble.ai/features/3d-relight) is a browser-based tool for real-time relighting and compositing of VFX/PBR passes.


# OpenArt VFX: AI video transformation, powered by SwitchX
Source: https://beeble.ai/blog/openart-vfx-powered-by-switchx
Published: 2026-06-13

OpenArt VFX, built by OpenArt and powered by Beeble's SwitchX, brings background replacement, cinematic relighting, and localized VFX to real footage.

***OpenArt VFX combines background replacement, cinematic relighting, and localized video transformation into one streamlined editing experience — built by OpenArt and powered by Beeble's SwitchX.***

AI video generation unlocks entirely new ways to create visual content. However, many creators, filmmakers, and production teams are not interested in generating new footage; they want to edit what's already been shot.

How do you change environments in post? Refine the mood of a scene? Apply selective visual effects without rebuilding an entire shot inside a traditional post-production pipeline?

That is the idea behind OpenArt VFX, a new AI-powered video transformation suite built by OpenArt and powered by Beeble's [SwitchX](https://beeble.ai/features/switchx) technology.

Designed around real production footage and creator workflows, OpenArt VFX combines background replacement, cinematic relighting, and localized video transformation into a streamlined editing experience that allows creators to iterate faster while maintaining high-quality visual consistency.

## AI video is evolving beyond generation

Working with existing footage demands workflows built around:

- Transforming instead of replacing.
- Iterating instead of restarting.
- Preserving performances and camera motion.
- Creating multiple visual directions from a single production.
- Reducing production overhead without sacrificing creative control.

OpenArt VFX was designed around those needs. The platform enables creators to refine environments, lighting, and selective regions while maintaining the integrity of the original shot.

## Three transformation workflows, one upload

OpenArt VFX combines three core workflows into a unified experience built for both accessibility and controllability.

### Replace Background

Isolate the primary subject in a shot and transform the surrounding environment.

Creators can work from presets, upload reference imagery, or generate entirely new scenes using OpenArt's AI image tools. A simple indoor setup can become a cinematic studio environment, a stylized cityscape, or a branded campaign backdrop.

The workflow is designed to reduce the complexity traditionally associated with background extraction and compositing while maintaining natural subject separation and motion consistency.

### Relight and Mood

Transform the visual atmosphere of an entire shot while preserving original movement and performance.

Daylight scenes can shift toward golden hour. Neutral footage can become stylized noir. Standard talking-head content can take on a more cinematic editorial aesthetic. The workflow emphasizes rapid visual iteration and creative exploration while maintaining frame-to-frame coherence.

### VFX

Selective frame-level transformation workflows. Paint over specific areas you want to keep, then guide the result using prompts, presets, or reference imagery. The system applies transformations across the shot while preserving consistency throughout the sequence.

## Get your results in three simple steps

Upload your video, pick your mode, then generate and compare.

![OpenArt VFX in three steps: upload your video, pick your mode, then generate and compare](https://desktop.beeble.ai/public/landing/cms/images/blog/openart-vfx-powered-by-switchx/steps.png)

## Powered by Beeble's SwitchX

At the core of OpenArt VFX is SwitchX, Beeble's video-to-video transformation model developed for high-fidelity video editing and controllable AI-driven workflows.

SwitchX enables:

- High-quality subject isolation.
- Cinematic relighting.
- Localized frame-level edits.
- Environment transformation.
- Temporally consistent video outputs.

> "OpenArt VFX represents a major step toward making professional-grade video transformation accessible to every creator," said **Coco Mao, CEO and Co-Founder of OpenArt**. "Many AI tools can generate video, but very few are designed for footage already shot on a real camera. SwitchX made professional-quality editing and advanced VFX work seamless and accessible."

For Beeble, the collaboration reflects a broader vision for how AI integrates into production and post-production workflows.

> "We built SwitchX to remove the technical barriers between imagination and production," said **Hoon Kim, CEO and Co-Founder of Beeble**. "Seeing OpenArt bring these capabilities to millions of creators is exactly the kind of creative future we envisioned."

## Bridging creative accessibility and production-grade AI

What makes OpenArt VFX particularly compelling is not simply the use of AI video technology, but how that technology has been packaged into practical creative workflows.

The collaboration between OpenArt and Beeble focuses on making advanced video transformation more accessible without losing the controllability, consistency, and visual quality required for professional results.

As AI tools continue to mature, workflows are increasingly shifting away from isolated generation systems toward integrated creative pipelines that help creators iterate faster, scale production more efficiently, and explore visual ideas with fewer technical barriers.

OpenArt VFX represents one example of that broader evolution: professional-grade AI video transformation designed around the realities of modern content creation and production.

## Available now

OpenArt VFX is available now as part of the OpenArt platform, powered by [SwitchX](https://beeble.ai/features/switchx) by Beeble.


# Releasing Beeble's Team Plan for studios
Source: https://beeble.ai/blog/beeble-team-plan-for-studios
Published: 2026-06-06

Beeble Team Plan brings shared credits, central billing, shared Canvas files, and flexible seat management to studios and production teams.

![Beeble Team Plan — a shared workspace for studios and production teams](https://desktop.beeble.ai/public/landing/cms/images/blog/beeble-team-plan-for-studios/team-plan-hero.png)

***Beeble Team Plan brings shared credits, central billing, shared Canvas files, and flexible seat management into a single workspace built for studios and production teams.***

Since day one, Beeble has empowered individual creators to push the boundaries of VFX. Today, we're excited to bring that same creative power to entire studios with the launch of Beeble Team Plan.

Designed for teams that need more than individual accounts, Team Plan provides a shared workspace that makes it easier to manage resources, collaborate on projects, and scale production workflows.

From boutique VFX houses and animation studios to content agencies and virtual production teams, Team Plan helps organizations spend less time managing tools and more time creating exceptional work.

## Key features

### Shared credit pool

Give your entire team access to a single credit balance. Artists can use the resources they need when they need them, while studio leaders gain greater visibility and control over usage across projects and departments.

### Central billing

Managing multiple subscriptions across a growing team can quickly become a headache. Team Plan simplifies administration with a single billing account and payment method, giving finance and operations teams a clearer view of costs and usage.

### Shared team files

Collaboration is at the heart of every production. Team members can share Canvas files directly within the workspace, making it easy to hand off projects, review work, iterate together, and maintain consistency across productions. Everyone stays aligned, wherever they are in the workflow.

### Flexible seat and role management

As productions evolve, so do team requirements. Team Plan makes it easy to add or remove seats as needed and assign Admin or Member roles, ensuring the right people have the right level of access at every stage of production.

### Built for collaborative production

Whether you're managing a small creative team or a growing studio, Team Plan provides the tools needed to streamline operations, improve collaboration, and scale your use of Beeble.

We're excited to see what your team creates next!


# Beeble at AI on the Lot 2026
Source: https://beeble.ai/blog/beeble-at-ai-on-the-lot-2026
Published: 2026-05-22

We're excited to share that Beeble will be attending AI on the Lot this May 27–28 in Culver City.

**We're excited to share that we'll be attending AI on the Lot this May 27–28 in Culver City.**

AI on the Lot has become one of the most important gatherings at the intersection of generative AI, filmmaking, media, and creative technology, bringing together filmmakers, studios, startups, technologists, artists, and investors who are actively shaping the future of storytelling.

![Beeble at AI on the Lot 2026 in Culver City](https://desktop.beeble.ai/public/landing/cms/images/blog/beeble-at-ai-on-the-lot-2026/ai-on-the-lot-banner.png)

Representing our team at the event will be **Cary Lin**, **Conrad Curtis**, and **Alex Goena** from our marketing, partnerships, and customer teams.

We're looking forward to connecting with creators, partners, and teams exploring what's next for AI-powered media workflows, storytelling tools, and creative production.

If you'll be there too, let's connect.

See you in Culver City!


# Launching Canvas: AI compositing for production
Source: https://beeble.ai/blog/launching-canvas-ai-compositing
Published: 2026-05-22

Canvas is our new node-based AI compositor — bringing AI generation, compositing tools, and workflow automation into one professional production environment.

![Beeble Canvas — node-based AI compositor for professional production](https://desktop.beeble.ai/public/landing/cms/images/blog/launching-canvas-ai-compositing/canvas-hero.png)

***Canvas brings together AI models, compositing tools, and workflow automation in one node-based environment, giving artists and production teams a more controllable, scalable way to create and refine visual content.***

For the last few years, AI video has advanced at an incredible pace. New models are appearing every month, producing increasingly impressive results. Yet anyone working in professional VFX, post-production, or virtual production knows that generating an image or a video is only a small part of the challenge. The real work begins when those assets need to be integrated into a production pipeline, refined through multiple rounds of feedback, matched across sequences, and delivered at a professional standard.

That reality is what inspired us to build Canvas, our new node-based AI compositor designed specifically for professional production workflows. Built for filmmakers, post-production teams, content creators, and video editors, our tool combines AI generation, compositing tools, and workflow automation in a single visual environment.

*Footage provided by ActionVFX*

## Why we built Canvas

Professionals need the freedom to combine different tools, compare creative directions, make precise adjustments, and revisit decisions throughout the production process. In reality, that often means stitching together multiple AI models alongside traditional compositing techniques, creating workflows that can quickly become fragmented and difficult to manage.

We built Canvas to bring those workflows together in a single visual environment, where AI generation and compositing work side by side rather than in separate applications.

![Canvas node graph for the "One Plate, Infinite Skies" workflow](https://desktop.beeble.ai/public/landing/cms/images/blog/launching-canvas-ai-compositing/one-plate-infinite-skies-interface.png)

## Bringing the best AI models together

Canvas provides native access to Beeble's core technologies, including [SwitchX](/research/switchx-2-0) for video-to-video transformation, [SwitchLight](/research/switchlight-3-0-is-here) for generating PBR passes such as normal maps, and AI-powered rotoscoping tools such as MatAnyone and [Corridor Key](/blog/corridorkey-beeble-green-screen-keying).

At the same time, we've designed Canvas as an open platform. Artists can also work with leading external models such as Seedance 2.0, Kling 3.0, Nano Banana, Depth Anything, and Qwen, making it easy to combine the best tools for each task without constantly switching between applications.

## Built for iteration

Creative work rarely follows a straight line. With Canvas, you can build reusable node graphs, generate and compare multiple variations, batch process workflows, and refine results without rebuilding pipelines for every shot.

Every step remains visible and editable, making experimentation faster while preserving full creative control.

![Canvas node graph for the Golden Hour Drive scene](https://desktop.beeble.ai/public/landing/cms/images/blog/launching-canvas-ai-compositing/golden-hour-drive-interface.png)

## More than an AI tool

Canvas is a production environment designed to connect AI generation, compositing, and workflow automation into a single system.

By bringing these elements together, you get a faster and more flexible way to create production-ready content while maintaining the precision and control that professional work demands.

![Detail view of a Canvas graph with multiple compositing nodes](https://desktop.beeble.ai/public/landing/cms/images/blog/launching-canvas-ai-compositing/golden-hour-detail-interface.png)

Canvas is available now, and we can't wait to see what you create with it. For further technical details, visit our [documentation page](https://docs.beeble.ai/beeble/canvas).


# AI relighting in Nuke: Foundry & Beeble tutorial
Source: https://beeble.ai/blog/ai-relighting-in-nuke-with-beeble-switchx-tutorial
Published: 2026-05-05

Learn how to relight footage in Nuke with Beeble SwitchX. This tutorial covers AI relighting, frequency separation, roto, cleanup, and regraining.

Artificial intelligence is opening up new creative possibilities inside compositing workflows, but integrating AI-generated results into high-end VFX pipelines still depends on precision, control, and finishing techniques.

In this tutorial from Foundry and Beeble, we walk through a complete workflow for relighting footage in Nuke using Beeble's AI-powered relighting tools, while maintaining the fidelity and flexibility professional compositors expect.

Learn how AI lighting can become part of a practical compositing workflow, from matching renders back into the original plate to restoring texture, fixing artifacts, and blending the final result seamlessly into a production-ready shot.

## What the tutorial covers

This walkthrough includes:

- Setting up a color-managed Nuke project for relighting work.
- Denoising source footage before AI processing.
- Using Beeble SwitchX for AI-generated relighting.
- Creating reference lighting using prompts.
- Importing relit renders with the Beeble Nuke plugin.
- Matching resolution and cleaning up artifacts.
- Using frequency separation to recover texture and sharpness.
- Refining the result with roto and color corrections.
- Repairing details with RotoPaint.
- Regraining the final composite for seamless integration.

## Bridging AI and traditional compositing

What makes this workflow compelling is not just the AI relighting itself, but how naturally it fits within established compositing practice. It combines the speed and creative flexibility of AI-generated lighting with the precision tools compositors already trust inside Nuke.

Explore our [Nuke plugin documentation](https://docs.beeble.ai/integration/nuke-plugin) for setup and configuration details.

Whether you're refining VFX shots, experimenting with AI-driven relighting, or looking for ways to integrate machine learning into a production pipeline, this tutorial shows how the two worlds can work together.


# NAB 2026 roundup: Putting a studio in a suitcase
Source: https://beeble.ai/blog/nab-2026-roundup-portable-virtual-production-studio-in-a-suitcase
Published: 2026-05-01

Beeble's NAB 2026 roundup — a portable virtual production studio at MGM Grand, spatial workflows with XGRIDS, Unreal Engine plugins, and AI relighting.

By Conrad Curtis | Production and Partnerships Lead at Beeble

NAB this year didn't feel like a glimpse of the future so much as a confirmation of something already underway. Across the show floor, in meeting rooms, and in improvised spaces off-site, the same idea kept surfacing in different forms: virtual production is no longer defined by stages, hardware, or fixed pipelines, but by how fluidly you can move between them.

What we saw, and what we built, pointed to a new baseline where production, post-production, and real-time tools are no longer separate phases but part of a continuous, iterative process. At the center of that shift is a simple but powerful concept: if your tools are flexible enough, your entire production can move with you.

## A portable virtual production studio in a hotel suite

![Portable virtual production studio set up in an MGM Grand hotel suite at NAB 2026](https://desktop.beeble.ai/public/landing/cms/images/blog/nab-2026-roundup-portable-virtual-production-studio-in-a-suitcase/portable-vp-hotel.png)

The clearest expression of that idea came not inside the Las Vegas Convention Center, but a short distance away, in a terrace suite at the MGM Grand, where we set out to answer a straightforward question: how much of a virtual production pipeline can you realistically compress into a portable setup without losing creative control?

Over the course of a few hours, the team from District Cinema transformed the suite into a fully functioning studio — not as a stunt, but as a working environment where capture, relighting, and inspiration could happen in real time and continue seamlessly into post.

What made this possible was not a single piece of technology, but the way various technologies worked together, particularly through SwitchX, which allowed us to treat lighting and comp as something that could evolve after the moment of capture rather than being locked in on set. Instead of being limited by our environment, we were able to focus on performance and composition, knowing that the look could be refined later with a level of control that traditionally required far more infrastructure. In that sense, our "studio in a suitcase" idea is less about portability alone and more about decoupling creative decisions from physical constraints.

## Spatial workflows and the XGRIDS panel

That same philosophy carried into our session with **XGRIDS**, where the conversation shifted from portability to spatial understanding. The introduction of **3D Gaussian Splatting** into the workflow is both a technical and conceptual upgrade, as it allows the scene itself to retain its structure, depth, and relationships across shots. In practical terms, this means that when you relight or recontextualize a scene using SwitchX, those decisions are no longer happening in isolation, but are informed by a coherent spatial model.

Spatial data is quickly becoming the connective tissue between traditionally separate domains such as capture, VFX, and real-time rendering. By integrating 3DGS workflows into a broader toolchain that includes tools like [Terra](https://districtcinema.com/terra) and [Captury](https://captury.com/), we are moving toward a production model where consistency is not something you fix in post, but is inherently preserved throughout the process.

![Conrad Curtis on the SwitchX panel at NAB 2026](https://desktop.beeble.ai/public/landing/cms/images/blog/nab-2026-roundup-portable-virtual-production-studio-in-a-suitcase/switchx-panel.jpg)

## Portable compute and CorridorKey in practice

If the hotel suite demonstrated what a portable studio looks like and the XGRIDS session explored how spatial workflows extend its capabilities, the presentation with **Puget Systems, District Cinema, and the Corridor Crew** grounded everything in the realities of day-to-day production. Running these workflows on compact, high-performance systems made it clear that portability goes beyond physical footprint, ensuring that the full creative pipeline remains accessible wherever you are working.

The introduction of our new **Beeble Plugin for Unreal Engine** — which we developed in collaboration with the **Corridor Crew and District Cinema** — and the integration of Corridor Crew's **CorridorKey** capped off our growing stack of Beeble tools. By addressing the long-standing challenges of green-screen extraction, particularly in fine detail, motion blur, and color accuracy, it complements the broader Beeble approach of preserving as much information as possible for downstream decisions. When combined with SwitchX and Unreal Engine, the result is a pipeline where capture, keying, relighting, and final pixel output are tightly interconnected rather than fragmented across tools and stages.

![NAB 2026 booth and demos with the Corridor Crew and District Cinema](https://desktop.beeble.ai/public/landing/cms/images/blog/nab-2026-roundup-portable-virtual-production-studio-in-a-suitcase/booth-1.jpg)

![Beeble plugin and CorridorKey demo at NAB 2026](https://desktop.beeble.ai/public/landing/cms/images/blog/nab-2026-roundup-portable-virtual-production-studio-in-a-suitcase/booth-2.jpg)

## Where it goes from here

What NAB ultimately reinforced is that this shift toward hybrid, generative production is not a distant possibility but an active transition already reshaping how work gets done. The ability to move between locations without losing capability, to defer creative decisions without sacrificing quality, and to maintain consistency through spatially aware workflows is changing how films are made and how teams think about the process.

If you saw the demos, visited the suite, or caught the sessions, we would love to hear how these ideas resonate with your own pipeline and where you see the opportunities or friction points. And if you missed it, this is the perfect moment to reconnect, because everything we explored at NAB is continuing to evolve, and the next phase will be shaped as much by collaboration as by technology.

Join the next phase of production.


# Beeble wins Webby Award for AI & machine learning
Source: https://beeble.ai/blog/beeble-webby-award-winners
Published: 2026-04-22

Beeble has won a Webby Award, receiving top honors in Apps, Software & Immersive – Best Use of AI & Machine Learning.

## A Webby Award for Beeble

Beeble has won a Webby Award, receiving top honors in the **Apps, Software & Immersive – Best Use of AI & Machine Learning** category.

The Webby win is a reflection of the work so far — from the team building the technology to the artists using it in ways we didn't initially expect. It's also a reminder that this space is moving quickly, and that the real opportunity is still ahead.

## What comes next

If this moment signals anything, it's that the boundaries between production and post are continuing to blur.

- Studios are becoming more portable.
- Workflows more iterative.
- Teams smaller, but more capable.

And the role of AI is becoming clearer: not replacing creativity, but expanding where and how it can happen.

## Want to be part of the new era of video to VFX?


# CorridorKey arrives in Beeble Background Remover
Source: https://beeble.ai/blog/corridorkey-beeble-green-screen-keying
Published: 2026-04-21

CorridorKey is now in Beeble's Background Remover—delivering physically accurate green screen keying with clean alpha and true foreground reconstruction.

Now available inside Beeble's Background Remover toolkit, CorridorKey introduces a fundamentally different approach to green screen extraction—one that doesn't just separate foreground from background, but reconstructs what the foreground should have been in the first place.

## From keying to unmixing

Traditional keying methods treat green screen removal as a masking problem. You isolate the subject, refine edges, and hope the result holds up under compositing. CorridorKey approaches it differently.

Instead of producing a binary mask, it solves what's known as the unmixing problem: predicting the true, un-multiplied color of the subject as if the green screen had never influenced it.

Hair retains its softness. Motion blur behaves naturally. Semi-transparent materials hold their integrity. And color contamination from the green screen is actively removed, not just hidden.

## A pipeline built for production

Inside Beeble, CorridorKey is paired with a second model to form a robust, production-ready pipeline:

- [MatAnyone](/models/matanyone) + [CorridorKey](/models/corridor-key) combine to deliver precise alpha generation and high-quality foreground reconstruction.
- Fine details like wispy hair and fast motion are preserved without manual intervention.
- Outputs are immediately usable in professional compositing environments.

## Formats that fit your pipeline

Beeble supports export formats built for VFX and post-production:

- 10-bit RGBA MOV for high-quality compositing.
- RGBA PNG sequences for frame-by-frame workflows.
- Alpha-only outputs for flexible pipeline integration.

Available now on the Beeble web app.


# Beeble named Webby Award nominee for AI-powered video-to-VFX
Source: https://beeble.ai/blog/beeble-webby-award-nominee
Published: 2026-04-08

Beeble has been named as a Webby Award Nominee, recognised among the top 11% of over 13,000 entries worldwide.

We're proud to share that Beeble has been named as a Webby Award Nominee — recognised among the top 11% of over 13,000 entries worldwide.

Beeble has been nominated in the Apps, Software & Immersive – Best Use of AI & Machine Learning category at the Webby Awards. Judged by the International Academy of Digital Arts and Sciences, the nomination recognises work pushing the boundaries of how digital tools shape creative industries.

## Rethinking how video is made

Most production workflows still follow a fixed sequence: decisions are made on set, and post-production is about refinement.

At Beeble, we've been building toward a different model — one where lighting, mood, and even environments don't have to be locked in at the moment of capture.

By combining AI relighting with video-to-video transformation, our technology allows creators to:

- Adjust lighting after filming
- Rework environments without reshoots
- Iterate creatively across production and post as one continuous process

This shifts filmmaking from something rigid into something far more flexible — and ultimately, more creative.

## Why this recognition matters

Being nominated for a Webby is a signal of where the industry is heading. AI in video has often been framed in terms of automation or imitation. But the real opportunity lies elsewhere: giving creators more control, not less.

"We're seeing production become more flexible and iterative, where creative intent can evolve throughout the process rather than being fixed at the point of capture." — Hoon Kim, CEO of Beeble

### From idea to execution, without the usual constraints

In practice, this approach opens up new possibilities:

- Smaller teams can achieve higher-end visuals
- Creative decisions can evolve later in the process
- Physical production constraints become less limiting

Instead of committing everything upfront, creators can explore, refine, and adapt as they go.

## What's next

We're continuing to work with partners across film, advertising, and content production to push these workflows further — moving toward a future where production and post-production are no longer separate stages, but part of a single creative loop.

If you're interested in exploring what this looks like in practice, we'd love to hear from you.


# #1 Background Remover for pros
Source: https://beeble.ai/blog/background-remover
Published: 2026-03-18

The best background remover for VFX pros — clean alpha mattes, video support, MOV with alpha, and PNG sequence export built for compositing pipelines.

Clean alpha mattes. Video support. MOV with alpha channel. PNG sequence export. Beeble Background Remover is the only tool that combines AI automation with the output quality and formats that professional compositing workflows demand.

## The background removal gap

If you work in post-production, you already know the landscape.

On one side, consumer tools like remove.bg and Canva give you one-click background removal for still images. They're fast, they're cheap, and they work well enough for social media graphics. But try running footage through them and you'll get soft edges, haloing around hair, and output formats that don't fit into a real pipeline. There's no alpha channel you'd trust in a comp.

On the other side, professional tools deliver broadcast-quality results. Rotobrush in After Effects, Nuke's keyers, DaVinci Resolve's Magic Mask — these produce mattes you can actually use. But they require significant manual work per shot, expensive licenses, and often frame-by-frame cleanup for difficult edges.

For a long time, Unscreen occupied the narrow space between these two worlds: dedicated video background removal that was easy to use and produced reasonable results. When Unscreen shut down on December 1, 2025, after its acquisition by Canva, it left a gap that neither consumer tools nor manual pro workflows could fill.

No tool currently offers AI automation, pro-grade alpha quality, and pro export formats in one package. Beeble Background Remover does — and that's why it's the best option for professionals.

## Why pros choose Beeble

Every feature in Beeble Background Remover is built around the things that actually matter in professional workflows: alpha quality, edge handling, and export formats that integrate with compositing software.

### Clean alpha mattes

The core of any background removal tool is its matte quality. Beeble Background Remover generates clean alpha channels that handle the hard cases — hair strands, fur, semi-transparent materials, and fine edges that trip up most automated tools.

This isn't the binary mask you get from consumer tools where every pixel is either fully opaque or fully transparent. The output preserves partial transparency at edges, which means you get natural-looking composites without the telltale fringe that screams "keyed footage."

### Video background removal

Most background removal tools are built for still images. The ones that claim video support typically process each frame independently, which produces temporal flickering and inconsistent edges across the sequence.

Beeble Background Remover processes video as video. The result is temporally coherent alpha mattes — edges that stay consistent from frame to frame, without the shimmer and crawl that make per-frame approaches unusable in production.

### Pro export formats — MOV with alpha and PNG sequences

This is where the tool starts to matter for professional workflows.

The output isn't a compressed video with a green screen behind it. Beeble gives you two pipeline-ready export options:

**RGBA video (MOV with alpha channel)** — A single MOV file with an embedded alpha channel, ready to drop into your timeline. This is the fastest path from extraction to comp: one file, full transparency, no assembly required. It works natively in After Effects, DaVinci Resolve, Final Cut Pro, and any editor or compositor that supports RGBA video.

**PNG sequences** — Individual PNG 8-bit frames with embedded alpha channels, the standard interchange format for VFX work. Lossless, frame-accurate, and compatible with every professional tool — After Effects, Nuke, DaVinci Resolve, and any compositing application that reads image sequences.

Both formats preserve the full alpha matte quality. Choose MOV when you want a single-file workflow; choose PNG sequences when you need per-frame control or maximum interoperability. No re-encoding, no format conversion, no quality loss between extraction and comp.

### Pro-level controllability

One-click tools make a single decision about what's foreground and what's background. That works when the answer is obvious. It doesn't work when you need to keep a specific object, isolate hands but not arms, or handle a scene where the "subject" isn't clearly defined.

Beeble Background Remover gives you control over the extraction process. You can guide the tool toward the result you need rather than hoping the automatic detection gets it right. This is the difference between a tool you use for quick social posts and a tool you build a pipeline around.

## Where it fits in real workflows

### Film and television VFX

Plate preparation is one of the most time-consuming parts of VFX work. Pulling clean mattes from live-action footage — especially footage that wasn't shot on a controlled stage — eats hours of artist time per shot. Beeble Background Remover handles the initial extraction, giving compositors a clean starting point that they can refine rather than build from scratch.

### YouTube and online video production

Content creators who produce at volume need background removal that scales. Shooting every video in a studio with controlled lighting isn't practical when you're publishing daily. Beeble lets you shoot anywhere and handle the background in post, with output quality that holds up on screen.

### Advertising and commercial work

Agency timelines are tight. When a client needs footage composited over a new background for a campaign, the traditional approach — manual rotoscoping or reshooting on green screen — takes days. Beeble compresses that to minutes, with quality that meets broadcast standards.

### Motion design and compositing

Motion designers regularly need isolated subjects for composites, title sequences, and multi-layer animations. MOV with alpha channel and PNG sequence output mean the extracted footage integrates directly into existing project files without intermediate rendering steps.

## How Beeble compares

- Capability
- Consumer tools (remove.bg, Canva)
- Pro tools (AE Rotobrush, Nuke)
- Beeble Background Remover
---
- Image background removal
- Yes
- Yes
- Yes
---
- Video background removal
- Limited
- Yes
- Yes
---
- Alpha matte quality
- Consumer-grade
- Broadcast-grade
- Pro-grade
---
- Hair and fine edge handling
- Soft, haloed
- Clean (manual work)
- Clean (automated)
---
- RGBA video (MOV with alpha)
- No
- Yes
- Yes
---
- PNG sequence export
- No
- Yes
- Yes
---
- Manual effort required
- None
- High
- Low
---
- Controllability
- None
- Full
- Adjustable
---
- Batch processing
- No
- No
- Yes
---
- Cloud processing
- Yes
- No
- Yes
---
- Pipeline integration
- Limited
- Native
- Native
---
- Cost
- Low
- High (license + labor)
- Credit-based

No other automated tool gives you broadcast-quality mattes in pipeline-ready formats. Traditional pro tools get you there, but at the cost of hours of manual work per shot. Beeble is the only option that delivers both.

## How it works

The workflow is straightforward:

1. **Upload** your image or video footage
1. **Configure** the extraction — adjust settings to guide the matte toward the result you need
1. **Process** — Beeble generates the alpha matte
1. **Download** your result as PNG (image), MOV with alpha channel, or PNG sequence (video) with embedded alpha

Processing runs on cloud GPUs, so it doesn't matter what hardware you're working on locally. The heavy computation happens server-side.

Beeble Background Remover is available on credit-based pricing as part of the Beeble platform.

## Part of a larger VFX platform

Background removal doesn't exist in isolation. In most workflows, it's one step in a larger pipeline — you extract the subject, then you need to relight it, place it in a new environment, or match it to a different scene's lighting.

Beeble is a full VFX platform that includes relighting, PBR extraction, and generative video transformation alongside background removal. A subject you extract with Background Remover can move directly into Beeble's other tools without leaving the platform or converting between formats.

This isn't a standalone utility bolted onto a website. It's a pipeline tool built by a team that understands how VFX work actually gets done.


# How AI learns to understand light
Source: https://beeble.ai/blog/ai-understand-light-creative-bloq
Published: 2026-03-17

In a new article for Creative Bloq, Beeble's founder explores why AI must move beyond imitation to truly understand light.

In a recent article for **Creative Bloq**, Hoon Kim, Founder and CEO of Beeble, explores a key limitation in today's AI lighting tools: they imitate what looks right, but don't understand how light actually works.

This distinction becomes critical in relighting. While many AI tools perform well under standard conditions, they often break when pushed, producing flat skin, unrealistic highlights, and inconsistent detail, particularly in complex areas like hair and materials.

The article argues that convincing relighting requires a shift from visual guesswork to a physics-based approach. By modelling how light interacts with surfaces — how it reflects, diffuses, and scatters — AI can produce results that remain stable and believable under creative control.

This change has broader implications for filmmaking and content creation. If lighting behaves predictably, it can be adjusted after capture rather than fixed on set, enabling more flexible and iterative workflows across video-to-VFX and virtual production.

Ultimately, the piece positions AI not as a creative replacement, but as a tool that becomes more useful as it becomes more reliable, supporting artists in shaping mood, atmosphere, and story.

That physics-based approach is what powers [3D Relight](https://beeble.ai/features/3d-relight), where every light you add behaves the way light does on set.

[Read the full article here](https://www.creativebloq.com/ai/we-taught-ai-to-understand-light-and-transformed-our-creative-process)


# Recreating One Battle After Another car chase on iPhone
Source: https://beeble.ai/blog/one-battle-after-another-iphone
Published: 2026-03-16

How far can accessible gear and AI tools push filmmaking? VP Land and VPX Lab recreated a multi-million dollar car chase using an iPhone to find out.

What happens when you take a multi-million dollar Hollywood car chase and try to recreate it with an iPhone?

That's exactly what VP Land, in collaboration with VPX Lab, set out to test using Beeble's AI tools within a hybrid filmmaking workflow. Not in theory, but in a real-world production: shooting plates, scanning environments, compositing, and pushing footage until it breaks.

The goal was simple: rebuild a sequence inspired by *One Battle After Another* using accessible tools — iPhone footage, AI relighting, and a hybrid production pipeline. The bigger question was harder: how far can this stack actually go, and where does it fall apart?

The process combined real-world capture and virtual production techniques. Exterior driving plates were shot on an iPhone, supported by 360-degree reference footage. Interior shots pushed further, testing both a traditional "poor man's process" (real car + blue screen) and a fully virtual approach, scanning a car into 3D and shooting the actor in an empty space.

Across the workflow, from shooting plates to compositing and relighting, the team ran into the core tension shaping AI filmmaking today: speed versus control. Traditional pipelines offered precision but required time and expertise. Newer AI approaches moved faster, but often broke consistency.

Using [Beeble SwitchLight](/research/switchlight-3-0-is-here) and later [Beeble SwitchX](https://app.beeble.ai/home), the team began exploring what happens when that gap starts to close: where relighting, restyling, and iteration can happen faster without completely losing control of the image.

This post is a first look at the experiment, a preview of what worked, what didn't, and what it reveals about where AI filmmaking stands right now.

Read the full breakdown in [VP Land](https://www.vp-land.com/p/shooting-the-one-battle-after-another-car-chase-on-iphone-what-worked-what-didn-t-and-what-s-next).


# Mastering SwitchX: 2 essential rules
Source: https://beeble.ai/blog/mastering-switchx
Published: 2026-03-02

Two essential principles for clean, production-ready results with SwitchX: master your mask and perfect your reference image.

Switch anything. Keep what matters.

When we built SwitchX, we designed it for one thing: control.

If you want clean, production-ready results, everything comes down to two essential principles:

- Master your mask.
- Perfect your reference image.

That's it.

Let's break it down.

## RULE #1: MASTER YOUR ALPHA MASK

Masking is the foundation of your result.

SwitchX needs to know what you want to preserve. Even if you use the same footage, prompt, and reference image, a different mask can completely change the outcome.

![Different masks applied to the same footage producing different SwitchX results](https://desktop.beeble.ai/public/landing/cms/images/blog/mastering-switchx/mask-comparison.avif)

### There are four ways to work with masks inside SwitchX.

### Auto mode

Auto mode is the fastest way to begin. SwitchX automatically detects and isolates the main subject.

![SwitchX Auto mode automatically detecting and isolating the main subject](https://desktop.beeble.ai/public/landing/cms/images/blog/mastering-switchx/auto-mode.avif)

From there, you can:

- Replace the background.
- Change the lighting.
- Drop your subject into a virtual production environment.

It's ideal when your subject is clearly separated, and you want fast, reliable results without manual setup.

### Select mode

Instead of letting the system decide, you tell it exactly what to keep. You can isolate a face, hands, or any object in the frame, preserving identity while transforming everything else.

Imagine changing a casual outfit into a doctor's coat while placing the subject inside a hospital. By protecting the face and hands, you keep recognizability intact while the world around it transforms.

![SwitchX Select mode isolating face and hands while transforming outfit and background](https://desktop.beeble.ai/public/landing/cms/images/blog/mastering-switchx/select-mode-face-hands.avif)

This is also where more advanced workflows open up. You can combine masks, invert them, or isolate multiple elements to support inpainting-style edits and more complex composites.

![Advanced masking with combined and inverted masks for inpainting workflows](https://desktop.beeble.ai/public/landing/cms/images/blog/mastering-switchx/advanced-masking.avif)

Advanced masking options include:

- Add multiple objects
- Invert masks
- Combine regions
- Use for inpainting workflows

This is where professionals unlock full creative control.

### Fill mode

Fill mode takes a different approach.

Instead of selecting part of the frame, you select all of it — and tell SwitchX to keep the structure while shifting the lighting or style.

![SwitchX Fill mode preserving scene structure while transforming lighting and mood](https://desktop.beeble.ai/public/landing/cms/images/blog/mastering-switchx/fill-mode-relighting.avif)

This is perfect when you don't want to replace objects or environments, but you do want to transform the mood. Turn midday into dusk. Make flat lighting cinematic. Restyle an entire scene without altering its physical layout.

Everything stays. The atmosphere changes.

### Upload mode

For full production control, Upload mode lets you bring your own black-and-white alpha map video, exported from tools like Nuke or Adobe After Effects.

This gives you frame-by-frame, pixel-level authority over what stays and what changes.

If you're working in a professional VFX pipeline, this is where SwitchX becomes deeply integrated rather than experimental.

![SwitchX Upload mode using a custom black-and-white alpha map for pixel-level control](https://desktop.beeble.ai/public/landing/cms/images/blog/mastering-switchx/upload-mode-alpha-map.avif)

## RULE #2: PERFECT YOUR REFERENCE IMAGE

Once your mask defines structure, your reference image defines intent.

Think of it as a visual contract.

Your reference should clearly describe the world you want: lighting direction, mood, background, color grade, costumes, and atmosphere. The more complete it is, the more predictable your result becomes.

Here's how SwitchX interprets it:

- In masked regions, it generates new elements inspired by the reference.
- In unmasked regions, it transfers lighting and style while preserving original pixel identity.

This is why faces remain consistent. Even if your reference image contains a slightly different version of the subject, perhaps generated elsewhere, SwitchX understands the original pixel data and applies lighting without overwriting identity.

![Reference image transferring lighting and style while preserving subject identity](https://desktop.beeble.ai/public/landing/cms/images/blog/mastering-switchx/reference-image-identity.avif)

The structure remains grounded. The performance stays intact.

### Pro tip: Refine before you generate

If you want expert-level results, don't just upload a reference image — edit it.

Open your reference in Adobe Photoshop. Adjust the color grade. Add props. Refine shadows. Shape the mood precisely.

Then bring that refined image into SwitchX.

---

And that's it!

Master your alpha mask.

Perfect your reference image.

And enjoy SwitchX.


# Shooting for AI relighting: Best practices
Source: https://beeble.ai/blog/shooting-for-ai-relighting
Published: 2026-02-27

This guide outlines the practical on-set considerations that help ensure footage works optimally with AI relighting workflows.

## Good versus bad footage for AI relighting

AI relighting opens up powerful possibilities in post-production. With tools like Beeble, cinematographers and VFX teams can adjust lighting, generate VFX passes such as depth and normals, and integrate footage into new environments without reshooting.

However, the quality of AI relighting is highly dependent on the quality and structure of the original footage. Even though our AI models generate VFX passes from your footage, we don't invent missing visual information. Our model interprets and reconstructs lighting based on what is present in the frame.

This guide outlines the practical on-set considerations that help ensure footage works optimally with AI relighting workflows.

### Why input footage matters

Our AI model generates VFX passes such as depth maps, normal maps, and lighting layers directly from your footage. These passes rely on visible visual cues like texture, edges, lighting gradients, and color information.

If those signals are missing or degraded, the generated passes will be less accurate, limiting how convincingly lighting can be changed later.

For this reason, capturing clean, well-lit, well-exposed footage with clear subject separation is essential.

## Good vs bad footage for AI relighting

### Good footage

Footage that works well with AI relighting typically shares several characteristics:

- **Clear subjects.** The subject should be easily identifiable — people, objects, or characters clearly visible in the frame.
- **Strong subject separation.** A clean distinction between the subject and the background helps the system understand scene structure and depth.
- **High image quality.** Sharp focus, sufficient bitrate, and detailed visuals allow the AI to generate accurate VFX passes.
- **Soft and even lighting.** Balanced lighting avoids harsh shadows or blown highlights, preserving detail across the image.
- **Correct exposure and white balance.** Maintaining natural color values and avoiding clipped highlights ensures lighting adjustments remain flexible in post.

### Bad footage

Certain types of footage make relighting significantly harder. Avoid:

- **Footage without a clear subject.** If the subject is too small or unclear, the model cannot reliably reconstruct lighting or geometry.
- **Low-quality footage.** Heavy compression, blur, noise, or out-of-focus images reduce the detail needed for VFX pass generation.
- **Incorrect exposure.** Overexposed or underexposed footage destroys lighting information that the model needs to reconstruct illumination.
- **Non-standard color spaces.** Footage captured in Log or non-standard color spaces can degrade results if not converted properly.
- **Excessive camera movement.** Shaky or unstable footage makes it harder to maintain consistent spatial understanding across frames.

## On-set tips for AI-ready footage

### 1. Preserve lighting detail

AI relighting relies on subtle lighting information. Avoid extreme contrast or heavy shadow clipping that removes surface detail.

When possible, use soft lighting setups that maintain gradation across the subject.

### 2. Maintain subject isolation

Relighting works best when the system can clearly identify the subject.

Good practices include:

- Green screen or controlled backgrounds.
- Avoiding overlapping foreground elements.
- Ensuring the subject occupies a meaningful portion of the frame.

### 3. Capture stable footage

Excessive handheld movement can introduce inconsistencies across frames.

For best results:

- Use tripods or stabilized rigs.
- Avoid sudden camera motion.
- Maintain consistent framing when possible.

## Preparing footage for AI relighting

Before processing footage:

- Confirm correct color space (Rec.709 / sRGB).
- Check exposure and white balance.
- Ensure the subject is clear and in focus.
- Stabilize footage if necessary.
- Verify file size, resolution, and frame limits.

Taking these steps ensures the AI has the visual information it needs to generate accurate relighting and VFX passes.

Once your footage is ready, generate VFX passes and relight it in [3D Relight](https://beeble.ai/features/3d-relight) — Beeble's browser-based relighting editor.


# What professional VFX artists actually need from AI
Source: https://beeble.ai/blog/vfx-artists-ai-broadcast-tech
Published: 2026-02-27

In a new article for Broadcast Tech, Beeble's CEO outlines the four things AI must deliver to be useful in real production environments.

In a recent article for Broadcast Tech, Hoon Kim, CEO of Beeble, explores a growing shift in how the VFX industry evaluates AI tools, moving beyond novelty to real production value.

While AI has captured attention for its creative potential, professional environments demand something different: reliability under pressure. In high-end production and broadcast workflows, unpredictable tools are not just inconvenient; they are unusable.

The article outlines four key requirements that artists consistently demand from AI systems:

- **Predictability.** Tools must respond consistently to creative input, without unexpected changes.
- **Repeatability.** Results must be reproducible across iterations and revisions.
- **Physical grounding.** Outputs must follow the rules of light, perspective, and real-world behaviour.
- **Pipeline integration.** AI must fit into existing tools and workflows, not disrupt them.

Together, these define the difference between AI as a novelty and AI as infrastructure.

The piece argues that the future of AI in VFX lies not in replacing artists, but in supporting them — handling repetitive tasks while preserving creative intent and fitting seamlessly into established production pipelines.

[Read the full article here](https://www.broadcastnow.co.uk/industry-opinions/what-professional-vfx-artists-actually-need-from-ai/5214371.article)


# Meet Edit Alpha, rebuilt for the reality of rotoscoping
Source: https://beeble.ai/blog/new-edit-alpha-rotoscoping
Published: 2026-02-16

Rotoscoping has always been part of visual effects. If you want to replace a background, combine shots, or integrate 3D elements, you need a clean matte.

AI can generate a strong starting point, but most shots still benefit from a bit of refinement. Edges and shapes will often need adjusting, and those small corrections make the difference between something that almost works and something that actually holds up in a final shot.

That's why we rebuilt Edit Alpha from the ground up.

## A simpler way to refine your matte

The new Edit Alpha in Beeble Cloud provides a reorganized interface that makes mask editing faster and more direct.

The layout is cleaner and easier to navigate. You can refine masks directly over your footage, so you always see exactly how your changes affect the shot.

Instead of digging through panels or hunting for controls, you stay focused on the matte.

## Full control over your masks

Rotoscoping lives in the details, and those details demand flexibility.

With the updated Edit Alpha, you can invert a mask instantly, switching foreground and background without rebuilding the shape. You can also adjust mask opacity so edges are easier to judge while refining.

If your workflow starts elsewhere, you can now import a custom alpha and refine it directly in Beeble Cloud.

The goal is simple: clear control over your masks without slowing down the work.

## Bezier curves support

The rebuilt pen tool now supports full Bezier curves, making it easier to draw clean shapes and smooth edges.

- Hold Shift to create perfectly straight lines.
- Tap Alt to switch between brush and erase modes while you work.

Keyboard shortcuts enable you to switch tools and refine masks without breaking your rhythm.

The result is a smoother editing experience where adjustments feel quick and precise.

## Built for Beeble Cloud

Everything runs in your browser, so there's nothing to install and no hardware setup to manage. Upload your footage, generate your alpha, and refine it in the same place. New to the workflow? See how [AI rotoscoping](/features/ai-rotoscoping) replaces spline-by-spline manual work.

Rotoscoping isn't going away, but the tools around it are improving. With a cleaner interface and faster controls, refining a matte becomes a standard step in the workflow rather than a bottleneck.


# The AI tools for virtual production you need to know about
Source: https://beeble.ai/blog/ai-powered-vfx-software-for-major-studios
Published: 2025-12-18

AI-native tools are revolutionizing virtual production and saving critical time and budget.

**AI-native tools are revolutionizing virtual production by compressing the entire workflow from planning to final relighting, saving critical time and budget.**

Virtual production is powerful; however, it can also be unforgiving. The moment the volume powers up or the stage is locked, the clock is ticking, and every minute costs money. 

Environments need to be ready, assets need to behave, cameras need to hit their marks and performance, lighting and playback all have to line up. If anything slips, you're burning budget just to stand still.

The reality for most teams is that there's never quite enough time to:

- Previs complex scenes properly before you're on stage.
- Turn quick concept art into navigable, testable shots.
- Capture performance data without a full mocap footprint.
- Build, scan and update environments at the speed of the schedule.
- Fix lighting and continuity issues once the set has been struck.

That's where a new generation of AI-native tools comes in. They compress the slowest and most fragile parts of the workflow: planning, asset prep, performance capture, worldbuilding, and final relighting. They help you get to "good enough to shoot" faster, and keep your options open when the schedule inevitably changes.

## **From idea to stage: smarter planning and virtual cameras**

Most problems in virtual production start long before anyone stands on a volume. If you can plan better, scout smarter and test coverage early, everything downstream gets easier.

1. ### [**Lightcraft Jetset**](https://lightcraft.pro/)

![Lightcraft Jetset real-time virtual production camera tool on a mobile device](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/lightcraft-jetset.avif)

**Lightcraft Jetset** is a good example of that shift. Running on a phone or tablet, it turns your device into a real-time VP tool for stage planning and virtual camera operation. You can walk through a space, frame shots, and see your live camera feed composited into 3D environments as you move. That means:

- Directors and DPs can explore blocking and lens choices during tech scouts, not just once they're on an LED stage.
- VP supervisors can test whether a shot is even practical before committing time, crew and budget.

It's essentially a lightweight rehearsal version of your stage, in your pocket.

2. ### [**Moonvalley**](https://www.moonvalley.com/)

![Moonvalley turning a single image into a navigable 3D camera exploration](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/moonvalley-3d-camera.avif)

**Moonvalley** takes the same idea of fast camera exploration and applies it to 2D source material. Its Marey model turns single images or simple scenes into navigable 3D spaces. Instead of staring at a storyboard and guessing how a dolly or crane move might feel, you can:

- Fly a virtual camera through a still
- Design precise paths and timing
- Share those moves with the VP team as a clear reference

That closes the gap between concept and volume. The creative team can work in the language they already use, frames, mood boards, concept art, and still deliver something spatially meaningful to the VP crew.

3. ### [**Intangible**](https://www.intangible.ai/about-us)

![Intangible browser-based spatial scene blocking and collaboration tool](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/intangible-spatial-blocking.avif)

To keep all of this aligned, **Intangible** adds a browser-based spatial layer. Think of it as a shared sandbox where you can block out scenes, drop in assets, sketch camera paths and iterate visually with clients and collaborators. Generative tools sit on top of that spatial understanding, helping you experiment with lighting and layout, without needing to be a full-time 3D artist. For virtual production, that means everyone is looking at the same scene, not just imagining it from different decks.

## **Performance and characters: from mocap to digital humans**

Once the world and cameras are blocked, you need believable people moving through it. That’s another area where AI is compressing timelines.

4. ### [**Movin3D**](https://www.movin3d.com/)

![Movin3D capturing full-body motion with a single LiDAR-enabled device](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/movin3d-motion-capture.webp)

**Movin3D** focuses on motion and spatial understanding using a single LiDAR-enabled device. No suits, no big mocap volume: you can capture full-body performance in minutes. For VP teams, that’s ideal for:

- Blocking stunts and action beats in rehearsal
- Generating quick previs for complicated scenes
- Testing whether a performance reads with the planned camera before scheduling full-scale capture

It turns motion capture into something you can do on more stages, more often.

5. ### [**Meshcapade**](https://meshcapade.com/)

![Meshcapade generating realistic rigged digital human characters from photos](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/meshcapade-digital-humans.avif)

On the character side, **Meshcapade** gives you realistic digital humans on tap. Feed it photos, scans or other inputs, and it generates rigged 3D characters that behave consistently across shots and projects. Once those avatars exist, they can become:

- Digital doubles for tricky setups or safety-critical stunts
- Background actors for crowd extension
- Consistent stand-ins for lighting, reflections and techvis

6. ### [**Wonder Dynamics**](https://help.wonderdynamics.com/introduction/)

![Wonder Dynamics automating CG character integration into live-action footage](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/wonder-dynamics-cg-integration.avif)

Layered on top of that, **Wonder Dynamics** (now part of Autodesk as Flow Studio) automates much of the grunt work in putting CG characters into live-action plates. It tracks actors, infers performance data and helps align animation and lighting to the plate. In practice, that lets you:

- Turn rehearsal or test footage into working character shots
- Validate creature and character beats early in the process
- Save full hero animation passes for the shots that really need them

Together, these tools turn performance and character work into something you can iterate on constantly, instead of a one-shot pipeline that's too expensive to change.

## **Worlds, scans and compute: building the backbone**

Virtual production environments need to be accurate, flexible and performant, especially when they’re based on real locations.

7. #### [**World Labs**](https://www.worldlabs.ai/)

![World Labs Marble model generating explorable 3D worlds from text and images](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/world-labs-3d-worlds.avif)

**World Labs** sits on the generative side of that backbone. Marble, its world model, turns text prompts, reference images, video clips or rough 3D layouts into spatially consistent 3D worlds you can move through and edit. In practice, lookbooks and location photos can become explorable sets rather than just reference decks, with outputs like Gaussian splats and meshes that drop into real-time engines and offline renders. For VP teams, it shortens the distance between concept art and shoot-ready environments, especially for smaller stages and hybrid LED/green-screen workflows.

8. #### [**Xgrids**](https://www.xgrids.com/intl?page=geomatics)

![Xgrids Lixel handheld LiDAR scanner for fast reality capture and digital twins](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/xgrids-lidar-scanner.avif)

**Xgrids** the anchors those worlds in reality. Its Lixel handheld LiDAR scanners and pipelines are designed for fast, precise reality capture, turning physical spaces into high-quality digital twins in a short time. Those scans feed straight into 3D workflows, giving VP teams:

- Clean, consistent world geometry for previs, techvis and final environments
- A single source of truth for sets, streets and interiors
- A way to keep location-based VP aligned across multiple shoots and seasons

On the compute side, Xgrids is also leaning into formats and infrastructure for modern 3D representation, like 3D Gaussian Splatting. The upshot is that heavy, volumetric scenes can be processed and served at scale without you having to reinvent the infrastructure.

Paired with Intangible, Moonvalley and Jetset, you end up with a pipeline where ideas become navigable worlds quickly, and those worlds stay accurate all the way to final.

## **Beeble: Treating real footage like CG**

All of that gets you to the same place: a shot with real performances, virtual or physical environments, and a director who still wants options.

This is where Beeble comes in.

![Beeble generating PBR passes from live-action footage for CG-like relighting](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/beeble-pbr-passes.avif)

Beeble takes live-action footage and generates full physically based rendering (PBR) passes, normals, depth, albedo, roughness, metallic and more. In other words, it reconstructs enough of the scene's geometry and material response that you can [relight it like CG](https://beeble.ai/features/3d-relight).

In a virtual production context, that changes what "locked" really means:

- If volume content and plate don't quite match, you can rebalance the lighting in post instead of booking another day on the stage.
- If a director wants to try a mood shift later in the day, more contrast, different ambience, you can explore that look interactively.
- If you're blending location photography with LED environments, you can use those PBR passes to create more physically consistent composites.

![Beeble Studio running locally on GPU with 4K relighting and on-prem security](https://desktop.beeble.ai/public/landing/cms/images/blog/ai-powered-vfx-software-for-major-studios/beeble-studio-4k-relighting.avif)

With [**Beeble Studio**](https://beeble.ai/beeble-studio), all of this happens locally on your own GPUs, in true 4K, with unlimited processing and on-prem security. That matters when you're dealing with unreleased shows, high-profile talent or confidential sets: the footage never leaves your facility, but you still get the speed and flexibility of AI-driven relighting.

In short, Beeble is the bridge between virtual production and traditional post, letting you treat captured footage with the same freedom you expect from CG renders.

## **Putting it together: an AI-native VP stack**

None of these tools replaces a director, a DP, a VFX supervisor or a stage crew. What they do, collectively, is compress the distance between intention and image.

A modern AI-assisted VP stack might look like this:

- Plan and explore with Intangible, Moonvalley and Lightcraft Jetset.
- Capture and perform with Movin3D, Meshcapade and Wonder Dynamics.
- Build and ground worlds with World Labs' generative environments and Xgrids scans and scalable 3D infrastructure.
- Finish and extend with Beeble, relighting and refining footage as if it were CG.

The teams with an edge in virtual production won't be the ones ignoring these tools; they'll be the ones squeezing real value out of them, using AI to take the friction out of planning, prep and fixes when the clock is ticking.

If you're already experimenting with virtual production and want more control over how you light and relight your worlds, Beeble is built to slot straight into that toolkit. Treat real footage like CG, keep your data where it belongs, and use AI to move faster when it matters most.


# Introducing Beeble Camera
Source: https://beeble.ai/blog/introducing-beeble-camera-iphone-vfx
Published: 2025-08-20

The simplest way to capture and process footage that's camera-tracked, rotoscoped, and ready for relighting.

**Turn your iPhone into a mini Hollywood studio. This is the simplest way to capture and process footage that's camera-tracked, rotoscoped, and ready for relighting.**

We're excited to introduce **Beeble Camera —** iPhone App that brings camera tracking to Beeble. This is the simplest way to capture and process footage that's **camera-tracked, rotoscoped, and ready for relighting**.

With Beeble Camera, your iPhone becomes a **mini Hollywood studio**:

- **Shoot** with your iPhone and get real-time camera tracking.
- **Process** instantly in Beeble Web with AI-powered roto & VFX Passes.
- **Create** virtual productions in minutes using Beeble Editor and Blender.

With Beeble Camera and Beeble Web, virtual production has never been this simple. Download Beeble Camera today, and start creating.

[**More details here**](https://beeble.ai/)


# Beeble at SIGGRAPH 2025
Source: https://beeble.ai/blog/beeble-siggraph-2025-vancouver-showcase
Published: 2025-08-06

We're excited to share that Beeble will be at SIGGRAPH 2025 in Vancouver from August 12–14.

**We're excited to share that Beeble will be at SIGGRAPH 2025 in Vancouver from August 12–14**

## **Booth Location**

\
We're at **Booth 712** in **West Exhibit Halls B & C**, **Vancouver Convention Centre**.

We'll be showcasing our latest updates in AI-powered Virtual Production — be sure to stop by if you're attending.

## **Birds of a Feather: Joint Session with Lightcraft**

![Beeble and Lightcraft Birds of a Feather session at SIGGRAPH 2025](https://desktop.beeble.ai/public/landing/cms/images/blog/beeble-siggraph-2025-vancouver-showcase/birds-of-a-feather-session.avif)

Join **Hoon Kim (Beeble)** and **Eliot Mack (Lightcraft)** for a casual session where we'll demonstrate innovative workflows for AI-powered Virtual Production:

- Camera tracking
- Rotoscoping
- Relighting

All without breaking your budget.

📅 **Aug 13 (Wed), 12:30–2:00 PM**

📍 **Room 13, East Building**

[**More details here**](https://c.vialoops.com/CL0/https:%2F%2Fs2025.conference-schedule.org%2Fpresentation%2F%3Fid=bof_190%26sess=sess616/1/0100019880d3e8d9-1502f801-bb3d-46e2-83e0-92b1299f3076-000000/91lBHAnhXg9Od783yzqdUn2p1bpeEJX0FB_glViYmBA=417)

We're building this future together, and we'd love to hear from you.

Come by, share your work, exchange ideas, and let's talk.


# Now with UE 5.6 support & project-based installation
Source: https://beeble.ai/blog/beeble-unreal-engine-5-6-plugin-update
Published: 2025-06-20

The Beeble plugin for Unreal Engine just got a major upgrade: UE 5.6 compatibility & Project-Based Installation

**The Beeble plugin for Unreal Engine just got a major upgrade: UE 5.6 compatibility & Project-Based Installation**

- **UE 5.6 Support:** UE Plugin is now compatible with 5.6
- **No more executing Python script every time**—just download the plugin, unzip it, and drop it into your project folder. It will stay activated as long as you are using the same project.
- **Camera Control to match real world:** The Beeble Camera and Beeble Plate are now anchored together - just like in the Blender add-on. We added focal length and distance control, allowing you to match the camera parameters to your real-world camera setup.  The media plane will **`auto-scale based on focal length and distance from the camera`**.


# SwitchFrame: Expand the frame. Preserve the shot
Source: https://beeble.ai/research/switchframe
Published: 2026-09-16

Meet SwitchFrame, Beeble’s video outpainting model. Expand your footage with visual reference guidance while preserving the shot’s motion, timing, and color.

### **Introducing SwitchFrame**

SwitchFrame is Beeble’s own video outpainting model, developed and trained specifically to extend existing footage beyond its original boundaries.

It generates the expanded scene while preserving the original performance, motion, and frame count. A reference image guides the full composition, with control over aspect ratio, source position, scale, and crop.

SwitchFrame powers Reframe on Beeble, supporting shots up to 20 seconds with 4K output and 10-bit color.

## **SwitchFrame at a glance**

- **Any aspect ratio.** Choose a preset or set your own custom width-to-height ratio to fit the screen or composition you have in mind.
- **20 seconds. 4K. 10-bit color.** Sub-second clips are supported, too.
- **Preserve shot.** Your source stays completely intact, with the original performance, motion, and frame count preserved as the surrounding scene expands.
- **Expand freely.** Choose where to position the source, how to scale it, and what to crop. Use a reference image to visually direct the expanded scene.

## **Any aspect ratio**

Turn horizontal footage into vertical, expand a vertical shot into ultrawide, or create a composition for a custom screen. Reframe supports 9:21, 9:16, 3:4, 1:1, 4:3, 16:9, and 21:9, alongside custom aspect ratios. The scene expands beyond the original boundaries to fill the format you choose.

## **Preserve the shot. Expand freely.**

SwitchFrame preserves the original performance and camera movement as the surrounding scene expands. With your chosen position, scale, and crop applied, the source region stays aligned frame by frame, and the original frame count is maintained. Position your footage to leave room on any side, scale it down to reveal more of the scene, or crop it to refine the composition. A reference image guides the environment, lighting, and layout of the expanded scene. Create one with Auto Fill or upload your own.

## **Up to 20 seconds. 4K. 10-bit color.**

SwitchFrame supports everything from sub-second inserts to shots up to 20 seconds. Output is available up to 4K with 10-bit color, providing the resolution and tonal precision needed for further editing and color work. Short clips keep their intended length, and longer takes retain the timing of the original performance.

## **See the expansion**

Drag the divider to compare each source with its reframed result. The source is shown at its position and scale inside the output canvas; black space marks the area beyond the original frame.

## **Put SwitchFrame to work**

SwitchFrame is available in the Beeble cloud app: as a Reframe node in the canvas, or on the standalone Reframe page.


# Meet SwitchX 2.0 — 4K, 20s, 10-bit, Camera Tracking & Fast Motion
Source: https://beeble.ai/research/switchx-2-0
Published: 2026-09-09

SwitchX 2.0 is the video-to-video model for high-end production. Now in 4K, 10-bit, 20 seconds, camera tracking, and fast motion.

### **Introducing SwitchX 2.0**

SwitchX is a video-to-video model where you can switch the background, lighting, and literally anything in your video with production-level quality.

Today we're announcing SwitchX 2.0 — 4K, 10-bit, 20 seconds, camera tracking, and improved source tracking for fast motion.

## **Core technology: Source-Locked Rendering**

Every generative video model can change a scene. Very few can do it without changing the subject. And when identity matters, even a subtle shift is a deal-breaker. Source-Locked Rendering is our answer: complete pixel alignment with your source, so what you protect stays untouched.

And because the output is pixel-aligned, you can treat it like a compositing layer: overlay the result on your source and nothing drifts. It works on three inputs.

#### 01 · Source video

Your footage: the performance, framing, and timing you already shot.

#### 02 · Mask

White means keep these pixels exactly, only relight them to match the new environment. Black means generate this region fresh, following the reference.

#### 03 · Reference image

A single frame showing how you want the shot to look. Outside the white region of the mask, the model brings in the reference's environment; inside it, only the reference's lighting. The reference is the single most important input to get right, and the full guide to designing one is in the [SwitchX docs](https://docs.beeble.ai/beeble/switchx#designing-the-reference-image).

## **What's new in 2.0**

1.0 introduced the workflow: source, mask, reference. 2.0 keeps it untouched and brings five core updates:

- **Longer shots**, raising the maximum duration to 20 seconds at 24fps.
- **Native 4K output**, increased from 1080p.
- **10-bit color depth**, up from 8-bit.
- **Improved camera tracking**, so generated results follow the camera movement of the source footage far more closely.
- **Enhanced motion fidelity**, with better handling of fast movement and lip synchronization.

**20 seconds.** The maximum clip length goes from 10 to 20 seconds: up to 480 frames, always matching the source frame count.

**4K.** Output resolution goes from 1080p to 3840×2160. The model reads the full-resolution source while rendering, so the detail comes from your footage, not from an upscaler.

**10-bit color.** Output bit depth goes from 8 to 10 bits per channel: 1,024 levels instead of 256. The extra levels keep gradients from banding when you push exposure or contrast in the grade.

**Camera tracking.** One of the two places 1.0 struggled hardest. 2.0 follows the source camera through handheld, pans, and parallax, so the generated environment moves exactly with your shot.

**Fast motion.** The other: fast movement (lip sync, dance, gesture), where video models smear or drift off the performance. 2.0 holds the detail.

**Iterate fast, finish in 4K.** A 4K pass is expensive, so 2.0 builds the workflow around it: iterate at 720p or 1080p, and finalize in 4K only when you are happy with the result. The final pass goes back to your original source and recovers its real detail, the same result as generating natively in 4K, without spending 4K credits on drafts.

## **Everything from 1.0 carries over**

Clips shorter than 3 seconds, usually a dead zone for video models, work. Frame count matches exactly. Aspect ratio never changes. The details that make a model usable in real VFX work.

## **What you can do with it**

The core workflow is background + relighting. But the mask is a general tool: most of what a VFX breakdown calls a shot is within reach of the same three inputs. A run through the cases, every pair straight from 2.0:

And this is the short list. The full reel, masks and all, lives in the [SwitchX showcase](https://app.beeble.ai/switchx).

## **Put SwitchX 2.0 to work**

SwitchX 2.0 is available in the Beeble cloud app: as a node in the canvas, or on the standalone [SwitchX page](https://app.beeble.ai/switchx).


# SwitchHDR - 16bit HDR from any video
Source: https://beeble.ai/research/switchhdr
Published: 2026-07-03

SwitchHDR is Beeble's in-house model for true HDR recovery, rebuilding 16-bit dynamic range from ordinary 8-bit SDR footage. Controllable by region and prompt.

### **Introducing SwitchHDR**

True HDR recovery from ordinary footage

SwitchHDR is our own frontier model for true HDR recovery. It rebuilds 16-bit scene radiance from ordinary 8-bit SDR footage. Trained on real HDR, temporally consistent, and controllable by region and prompt.

## **Recovering true 16-bit from 8-bit SDR**

Conventional SDR footage captures only a portion of the scene's dynamic range. Highlights clip to pure white. Shadows collapse into sensor noise. Once that detail is lost, it's no longer represented in the SDR signal.

A blown-out region could have been many different things: the sun, a window, a bright sky. There is no single correct answer hiding in the pixels. Conventional tools can only stretch what remains, and 8-bit precision leaves no room for it: stretching exposes banding, flattens highlights, and amplifies noise.

SwitchHDR reconstructs it. Instead of stretching the existing signal, the model infers the high-dynamic-range scene the footage came from. It rebuilds clipped highlight detail, recovers clean structure from noisy shadows, and outputs a true 16-bit result in a scene-linear, professional container.

It's a recovery model, not a generator. The performance, the framing, and everything the camera actually captured stay intact. SwitchHDR only restores the range the SDR signal never carried.

## **Recovery, grounded in real light**

SwitchHDR was trained on real high-dynamic-range footage, not synthetic exposures or tone-mapped approximations. It learns how light actually behaves across the full dynamic range: how a highlight rolls off, how much energy a light source really carries, how shadow detail sits above the noise floor. Its recovery is physically plausible, not a stylized guess.

#### 01 · Analyze the 8-bit SDR signal

The model identifies where the footage has clipped in the highlights and where shadow detail has collapsed into noise. It does this consistently across the whole sequence, so recovery regions stay stable from frame to frame.

#### 02 · Reconstruct the scene radiance

Highlight detail is rebuilt and shadow noise is suppressed, guided by what the model learned from real HDR footage. Because SwitchHDR processes the sequence as a whole, moving highlights stay coherent across frames. No shimmer, no flicker, no ghosting.

#### 03 · Respect your direction

Luminance-threshold masks and text prompts steer the recovery. Outside the mask, the original is preserved pixel for pixel. Inside, the model reconstructs with more latitude toward your stated intent. Highlights and shadows can be directed independently, because a blown sky and a crushed interior rarely want the same treatment.

#### 04 · Output in a professional container

Results are written as a 16-bit EXR sequence in scene-linear ACES AP0 (ACES2065-1). Because the output is true scene radiance, you can push exposure several stops in either direction in the grade and the image holds. Highlights reveal structure instead of going gray. Shadows open up instead of dissolving into noise.

## **The only HDR recovery you can direct**

Recovering clipped regions is fundamentally a one-to-many problem. The same white pixels could resolve into a dozen plausible scenes. Every other approach picks one for you: a single automatic output you're asked to accept. SwitchHDR takes direction at the model level, by region and by intent.

#### Region · luminance masks

Define shadow and highlight regions with a luminance threshold. Outside the mask, SwitchHDR stays faithful to the source, pixel for pixel. Inside, it reconstructs lost detail with more latitude, so recovery happens only where you want it.

#### Intent · per-region prompts

Tell the model what should fill each recovered region, with separate prompts for highlights and shadows. *Highlights: sunset sky breaking through clouds. Shadows: weathered wooden texture.* The reconstruction follows your shot's intent instead of the model's default guess, and the guidance lands only where it belongs.

## **See the difference**

The clearest evidence is visual: the same SDR source, recovered. Drag each slider to compare the original against the HDR result. Since browsers can't display true HDR, every result below is the 16-bit HDR EXR graded to SDR in DaVinci Resolve for web playback.

## **Put SwitchHDR to work**

Recover true HDR from your own footage: controlled, frame-accurate, temporally stable, and ready for your pipeline. See SwitchHDR at work inside the SDR to HDR tool. Try it for yourself!


# SwitchX - Change background and lighting in a single click
Source: https://beeble.ai/research/switchx
Published: 2026-02-06

Meet SwitchX, our first video to video model. Modify lighting, background, props, and more, while keeping your original subject intact.

### **Introducing SwitchX: Switch anything, keep what matters.**

Meet **SwitchX**, our first video to video model. It lets you modify lighting, background, props, and more, while keeping your original subject intact.

For filmmakers, creators, and VFX artists, changing the environment or lighting of a shot after it's been filmed has always been a complex, time-consuming challenge. Generative AI promised a solution, but traditional models often hallucinate, morphing the subject and losing the original performance.

SwitchX changes that. Guided directly by your source footage at the pixel level, SwitchX lets you modify lighting, backgrounds, props, and more, all while keeping your original subject completely intact. It ensures consistent identity, preserved details, and highly controllable results that stay grounded in your original performance.

SwitchX is now available on the [cloud app](https://app.beeble.ai/home). Try it for yourself!


# SwitchLight 3.0 is here
Source: https://beeble.ai/research/switchlight-3-0-is-here
Published: 2025-11-05

SwitchLight 3.0 is the best Video-to-PBR model in the world, delivering unmatched quality in generation.

### **Introducing the best Video-to-PBR model in the world**

SwitchLight 3.0 is the best Video-to-PBR model in the world. It delivers unmatched quality in generating physically based rendering (PBR) passes from any video, setting a new standard for VFX professionals and filmmakers who demand true production-level relighting.

Light defines every scene in movies, commercials, music videos, and every piece of visual media. The quality of lighting determines the quality of storytelling. Being able to change lighting digitally in post-production has always been every creator's dream.

SwitchLight 3.0 makes this possible. We're bringing SwitchLight 3.0 to both the Beeble cloud app and our new desktop app, Beeble Studio.

- In the [**Beeble cloud app**](https://app.beeble.ai/), all footage is now processed using SwitchLight 3.0, upgraded seamlessly from version 2.0.
- The new [**Beeble Studio desktop app**](https://beeble.ai/beeble-studio) runs SwitchLight 3.0 entirely on your local GPU, designed for professionals who need secure, local workflows. It unlocks 4K processing, 16-bit EXR outputs, and supports videos up to 1 hour in length.

SwitchLight 3.0 is available today. If you're using Beeble Cloud, your footage will be automatically processed with the new model, no price changes. If you're using Beeble Studio, the app will automatically download and install the new model.

## **What's new in SwitchLight 3.0**

SwitchLight 3.0 has 2 main changes:

- **True Video Model:** For the first time, SwitchLight is a **true video model** that processes multiple frames simultaneously. Earlier versions relied on single-frame image processing followed by a separate deflicker step. In version 3.0, temporal consistency is built directly into the model, delivering smoother, more stable results while preserving sharp detail.

- **10× Larger Training Set:** Trained on a dataset ten times bigger, SwitchLight 3.0 captures a broader range of lighting conditions, materials, and environments for more realistic results. This expanded data foundation significantly enhances realism and generalization across different scenes.

![](https://desktop.beeble.ai/public/landing/cms/images/research/switchlight-3-0-is-here/data-size-comparison.avif)

### **CAPTURE EVERY DETAIL UP TO 4K**

SwitchLight 3.0 achieves a new level of detail and visual clarity. Compared to 2.0, it captures **finer facial definition, intricate fabric wrinkles, and sophisticated surface patterns** with far greater accuracy. Materials look more natural and consistent, and the overall clarity of PBR outputs has significantly improved.

### **TRUE VIDEO MODEL**

SwitchLight 3.0 is a **true end-to-end video model** that understands motion natively. Unlike SwitchLight 2.0, which relied on an image model and post-smoothing, it eliminates **flicker and ghosting** even in extreme motion, shaking cameras, or vibrating subjects. Every frame maintains **coherent materials** across time.

### **Improved Stability and Refinement**

SwitchLight 3.0 resolves previous artifacts such as the strong green tint and incorrect normal orientation on the top of the head, restoring accurate surface color and lighting. Hair are now rendered with greater precision and maintain smooth temporal consistency across frames. Even in animated sequences, SwitchLight 3.0 preserves fine structural details.

SwitchLight 3.0 is now available on both the [cloud app](https://app.beeble.ai/home) and the [desktop app](https://beeble.ai/beeble-studio). Try it for yourself!


# SwitchLight 2.0 is here
Source: https://beeble.ai/research/switchlight-2-0-is-here
Published: 2025-07-01

Our Next Generation Model for Video-to-PBR and Relighting.

### **Our Next Generation Model for Video-to-PBR and Relighting**

## What is SwitchLight 2.0

**SwitchLight 2.0** is our most advanced **Video-to-PBR** model, designed to generate accurate physically based rendering (PBR) maps—including **normal, basecolor, metallic, roughness, and specular maps**—from any type of footage.

Compared to SwitchLight 1.0, it delivers **significantly enhanced accuracy and understands the whole scene**, including both background elements and common objects. It is built to empower creators of all levels with full relighting and compositing capabilities.

## SwitchLight 1.0 vs 2.0

SwitchLight 2.0 features a completely new architecture, purpose-built for high-fidelity relighting of any footage. It's trained on 13× more data, spanning a much wider range of objects and environments. Combined with a 10x larger model size, SwitchLight 2.0 delivers sharper details—now including more extreme scenarios like complex environments.

### **Ultra-detailed PBR maps**

SwitchLight 2.0 can capture the details of skin texture, fabric, and facial and scene detail with stunning accuracy. Even in challenging real-world scenarios such as outdoors with strong shadow and highlight, making a significant improvement from SwitchLight 1.0, 2.0 model faithfully recovers PBR maps.

### **Full-Scene Understanding Beyond Humans**

SwitchLight 1.0 required an alpha mask to specify the foreground subject, as it could only generate PBR passes for isolated humans. In contrast, SwitchLight 2.0 requires no alpha input and generates accurate PBR maps for the entire scene. Trained on a diverse dataset of objects beyond just people, SwitchLight 2.0 enables high-quality relighting and seamless compositing in any setting.

## Image-Based Lighting on Real Video

Image-Based Lighting (IBL) is a gold standard in 3D rendering—using HDRIs to create realistic, environment-driven lighting with accurate reflections, shadows, and ambient tones. But until now, IBL was limited to CGI, because real footage lacked the necessary PBR data.

SwitchLight 2.0 unlocks this capability by generating high-quality PBR maps directly from video. You can apply HDRIs to live-action scenes and get physically accurate relighting, just like in Blender or Unreal. No 3D reconstruction—just instant, natural lighting in our real-time web editor or any 3D pipeline.

## Real-World Examples

## More Examples

## Export to Blender, Unreal, Nuke

You can export the PBR maps generated in Beeble Studio and import them into your favorite tools—such as Blender, Unreal Engine, After Effects, or Nuke. For Blender and Unreal specifically, we offer dedicated add-ons and plugins that let you import your footage as a fully relightable asset, making the integration seamless and powerful.

## Try SwitchLight 2.0 at Beeble Editor

The easiest way to experience SwitchLight 2.0 is at Beeble Editor. Simply upload your green screen footage, and SwitchLight 2.0—running in the cloud—will automatically extract all the PBR passes. From there, you can experiment with real-time lighting and compositing to bring your shots to life.

Try SwitchLight 2.0 for free on web! (No download required)


# Hearts2Hearts: Combining AI, 3D, and compositing for a K-pop music video
Source: https://beeble.ai/showcase/ean-hearts2hearts-lemon-tang-music-video
Published: 2026-10-16

VFX artist EAN used SwitchX and Canvas alongside 3D and traditional compositing to extend practical sets while preserving the artists’ performances.

EAN is a VFX artist working across music videos and brand films. With a background in Cinema 4D-based 3D and compositing, EAN has recently explored how generative AI can fit naturally into existing VFX pipelines alongside live-action footage and established production techniques.

For Hearts2Hearts’ “Lemon Tang” music video, EAN handled AI VFX, 3D production and compositing. The project combined practical blue-screen photography with generated environments and purpose-built 3D elements, choosing different techniques based on each shot’s needs.

SwitchX was used for shots where the performers needed to remain intact while the environment around them changed. Canvas helped organize iterations, masks, and generated elements before the shots moved into final compositing.

## What was your role on the Hearts2Hearts project?

The project was Hearts2Hearts’ “Lemon Tang,” and I was responsible for AI VFX, 3D production, and compositing.

Starting with the original footage, I created the spaces and objects we needed. For some scenes, I built the image in 3D. Where AI was effective, I actively used Beeble AI’s SwitchX to extend the original blue-screen footage into new environments. At the end, I composited the generated AI material, 3D results, and original footage back together, then refined the color and details. So rather than a video completed with AI alone, this project followed a VFX workflow that combined 3D, AI, and compositing differently depending on the scene.

## What was the visual concept for the video?

As the project name “Lemon Tang” suggests, the video needed to feel fresh and bright throughout. We needed to create a natural resort atmosphere through the environments, color, object design, and materials. Overall, the aim was to extend a fantasy world from realistic live-action photography while maintaining the bright, polished visual style associated with K-pop music videos.

## How did the workflow develop from production through post?

Before the shoot, we created references for the overall concept and the spaces we needed, and established the visual direction.

Then, for each shot, we separated what needed to remain from the original photography from the areas we could transform or recreate with AI. Once we received the footage, I first analyzed the performers’ movement and camera movement, then created reference frames close to the final image I wanted.

From there, I used AI to change the background or the environment's shape. Where necessary, I recombined several results or supplemented them with 3D and conventional VFX techniques. For scenes where we needed to keep the filmed performers while significantly changing the surrounding environment, I used SwitchX.

The final stage followed a conventional VFX process, with masking, compositing, color matching, and detail work to make the shots connect naturally across the sequence. I approached AI less as a replacement for the whole pipeline and more as another production method added within the existing VFX process.

Personally, I think it is important to choose the production method and tech that best fits each shot.

## Where did Beeble fit into the workflow, and why did you choose SwitchX and Canvas?

With general video generation models, even when I could create the visual I wanted, the original performer’s face, movement, or silhouette often changed slightly. In a music video, maintaining the artist’s performance and identity is very important, so I needed a tool that gave me control over that.

What was particularly useful about SwitchX was being able to approach the parts of the original video that needed to remain and the parts that needed to change separately. It suited this project because I could preserve the performer’s movement and form as much as possible while changing the surrounding space, lighting, and atmosphere.

I used Canvas because it enabled me to connect these processes in a single node-based workflow. AI work rarely ends with a single generation. You keep adjusting masks, trying different references, and compositing the results again. Being able to organize that process visually was useful.

## What was important in making the performers and new environments feel like part of the same shot?

The most important thing in those shots was not simply making the background more elaborate. The performer and the new space needed to look as though they had been photographed together from the beginning.

I paid close attention to how light fell on the performers and surrounding objects, and how the color connected with the new environment. I wanted the boundary between the live-action photography and the generated space to be as difficult to perceive as possible.

## Can you take us through one shot, from the original footage to the finished result?

First, I defined the area containing the performer that absolutely needed to be preserved. Then I created a reference image for the space I ultimately wanted to make.

Before generating video, I prefer to establish the direction as specifically as possible using a single frame first. I repeatedly adjust the background structure, color, and lighting direction in the still image. Once those things are reasonably resolved, I can move into video with much more stable control over the result.

Using that reference, I changed the background and overall environment in SwitchX while retaining the performer’s movement. I then brought the result into Canvas, readjusted masks where needed, and combined it with other generated results to refine the details.

Finally, I returned to my existing compositing tools to refine the color, edges, and smaller details before completing the shot.

The key to this project was treating the AI result as a VFX pass rather than as the finished shot.

## What were the most difficult creative or technical problems?

The hardest part wasn't making a major change to an individual shot. It was keeping multiple shots in a scene consistent in tone, mood, and environment.

With AI, you can quickly create an interesting result for one shot, but making it feel like the same space when the camera angle or the performer’s position changes is another problem. If the background structure, object positions, materials, color, or lighting atmosphere change slightly from shot to shot, you notice the difference when they are edited together as one scene.

So I first defined the scene's overall visual reference and spatial structure as specifically as possible. Then I repeatedly compared each reference and generated result to make sure individual shots didn't stray too far from that baseline. When needed, I selected specific areas for compositing or added 3D elements to maintain continuity in the space.

At the time I was working on the project, there were also limitations around generating text within images or video in exactly the form I wanted. Typography and signage needed precise design and legibility, so I couldn't rely solely on AI for those elements. I created them separately in 3D and combined them during final compositing.

I used AI to quickly extend the atmosphere and visual character of the space, while 3D and compositing handled elements like forms and text that required precise control.

## Did you develop any techniques you would recommend to other artists using SwitchX and Canvas?

The most important thing I found on this project was to establish the visual baseline for the whole scene rather than concentrating only on making a single shot work well.

When using SwitchX, I also found it more effective to create a good reference and clearly define the elements that needed to remain than to repeatedly adjust the prompt.

In particular, rather than starting with video, I recommend spending time on a single frame and getting it as close as possible to the final image. If you clearly establish the space's structure, lighting, and color at that stage, I found I could get much more consistent results later in SwitchX.

## After this project, where would you like to take this kind of work next?

In the future, I’d like to blur the line between shooting and AI more actively. For example, it could be interesting to shoot only a minimal set and then extend it into a completely different world, or take the same performance and place it into different spaces and lighting environments.

In my personal work, I’d also like to create more unrealistic and surreal spaces. I’m interested in putting real people into environments with structures or physical qualities that would be difficult to shoot in reality.

Ultimately, I want to explore less “what can I make with AI?” and more how existing photography, 3D, VFX, and AI can connect into one natural production language. I’d like to continue experimenting with Beeble as a tool within that kind of hybrid production process.


# Keeping performance real while rebuilding the world: The hybrid workflow behind Crooks
Source: https://beeble.ai/showcase/bykai-crooks-hybrid-workflow
Published: 2026-10-06

How B-01 Films and BYKAI kept real performances at the centre of Crooks, using Beeble to relight actors into digitally built worlds.

For *Crooks*, the ambition was never to make an AI-generated drama. It was to find a way to make a live-action production whose world could extend far beyond what its budget and schedule would traditionally allow.

Produced by [B-01 Films](https://www.b01films.com/), with [BYKAI](https://www.bykai.ai/) developing and integrating the AI-hybrid workflow, *Crooks* brings together traditional filmmaking expertise and new production techniques. B-01 Films provides the production foundation; BYKAI contributes the planning, R&D and workflow integration that connect live-action capture with digitally constructed environments.

Developed as a nine-part premium series, *Crooks* moves through historically specific locations and environments, while keeping real actors, performances and cinematography at the centre of the production. For the first episode alone, the team needed to capture 22 different locations across an eight-day shoot.

To make that possible, showrunner **Boi Kwong of B-01 Films, who is also CEO of BYKAI**, and **Lim Kim Hui, Chief AI Strategist at BYKAI**, developed a hybrid production model that combined live-action filmmaking with 3D environments, AI, and VFX techniques. The team separated the process into three layers: capture the performance first, build the world around it, then bring everything together in the finish.

BYKAI's contribution was to connect these layers into a production infrastructure, making decisions about what to capture practically, what could be built digitally, and how to preserve continuity through the finished sequence. That expertise is central to its proposition for producers and studios considering AI-hybrid production.

The team integrated Beeble into the workflow, giving the filmmakers greater flexibility to work with the original photographed performance while adapting the lighting and environments around it.

*Crooks* also provides a case study for BYKAI's international ambitions: applying production expertise developed in Singapore to projects with overseas producers, directors and studios.

We spoke with Boi and Kim about the creative ambition behind *Crooks*, why performance became the non-negotiable part of their workflow, and how changing what could happen after the shoot influenced what was possible on set.

## Could you introduce yourselves, your roles on Crooks, and how you became involved in the project?

**Boi:** I'm Boi Kwong, founder of B-01 Films, the production company behind *Crooks*, and CEO of BYKAI. I come from a traditional filmmaking background, and I'm the showrunner on *Crooks*. I was actually very sceptical about AI at the beginning. I'm fiercely protective of actors and authentic performance, and I didn't want technology taking over the creative part of filmmaking.

But *Crooks* was an ambitious project, and we reached a point where our budget wasn't enough to realise the vision in the traditional way. I hit a wall and started looking for different ways to move the project forward. That's when I met Kim. He started showing me what some of these tools could actually do. What changed my perspective was realising that we weren't talking about replacing actors or replacing filmmaking. We could preserve the performance and use technology to make more of the world around it possible.

**Kim:** I'm Lim Kim Hui, Chief AI Strategist at BYKAI, and my background is much more on the technology side. Boi and I came into this from very different perspectives, so we spent a lot of time in pre-production doing R&D and testing what could actually work reliably enough for production.

## You describe Crooks as a "character-led premium series where the world can grow as quickly as the story demands." What was the creative ambition behind it?

**Boi:** *Crooks* is rooted in a real story and takes us into an earlier period of Singapore, so the world needed to feel historically believable.

We had to create environments including a prison, a hospital, a two-storey restaurant, and historically accurate buildings and locations that either no longer exist or would be very difficult to find and control for a shoot today.

The scale was huge for the schedule we had. Episode one alone involved 22 locations across eight shooting days.

But I didn't want those constraints to make the world of the story smaller. The question became: how do we keep the performances and the filmmaking real while giving ourselves much more freedom over the environment around them?

## You've said that "technology should make the world larger without making the performance feel smaller." How did that philosophy shape the production?

**Boi:** The performance was always the most important thing. Actors needed to perform together and have something physical to interact with. Green-screen acting isn't something every actor here is used to, so wherever interaction mattered, we used practical props, markers and physical elements to give them a sense of the space around them.

I also didn't want to restrict the camera. We shot handheld. We used dynamic camera movement. For me, the technology had to adapt to the filmmaking. I didn't want the filmmaking adapting to the limitations of the technology.

## What were the main production or VFX challenges you needed to solve to make that possible?

**Kim:** One of the biggest challenges was consistency. It's possible to generate a very convincing individual image, but filmmaking is about sequences. The camera moves, you change perspective, you cut to another angle, and the world and performance still need to feel like they belong to the same scene.

With generative tools, if you give the model too much freedom, it can start changing things you don't want changed. It might change the environment, but it can also start changing the composition, camera perspective or the character.

So we couldn't treat a shot as one image that we handed over to AI. We needed to separate the elements and define what could be altered and what needed to remain untouched.

## Where did Beeble fit into the workflow?

**Kim:** Beeble became an important part of the process of bringing the photographed performance and the world together.

We worked closely with the Beeble team during R&D and production, particularly on how we could use the technology to relight and integrate the live-action material into different environments.

For us, the key was starting from the original performance. The actor's movement and expressions were already there, and camera choices were already there. We didn't want those things to be regenerated.

Beeble gave us another level of control over what we could change around that source material.

**Boi:** Beeble saved our lives. I didn't want to sacrifice the actors' performances or change the way I wanted to shoot just to make the technology work.

## What did Beeble allow you to do differently?

**Boi:** One of the biggest areas was lighting.

We could capture the actors under more controlled, relatively flat lighting rather than needing to commit to every aspect of the final look while we were shooting.

That gave us much more flexibility afterwards. We weren't always dependent on the natural light being perfect or waiting for something like golden hour. We could focus on getting the actors and the performance right, knowing we had more control over the lighting later.

That becomes very important when you're working on a schedule like ours.

## Could you describe a sequence where Beeble played an important role?

**Boi:** We had scenes shot in the studio that needed to become very dark, atmospheric night environments.

If we'd shot the actors under the final lighting we wanted for those scenes, we would have committed to that look very early and potentially limited what we could do with the footage afterwards.

Instead, we could capture the performance with the information we needed and then use Beeble to relight the actors for the finished environment.

What mattered was that the performance didn't change. We still had the actor's face, expressions and movement, and we still had our original camera work. We were changing the conditions around that performance.

## Did flexibility in post affect production decisions?

**Boi:** Definitely. When you know you have that flexibility later, you start making different decisions on set. Lighting is one example, but it also affects locations and how much of an environment physically needs to exist when you're shooting. That meant we could focus our time on what had to happen in front of the camera.

At the same time, we didn't want that freedom to become an excuse to fix everything later. If something was important to the actor's performance or physical interaction, we still wanted it there on set.

That's the hybrid part of the workflow. You're constantly deciding what needs to be real and what can become flexible.

## How did you choose between AI practical filmmaking and traditional VFX?

**Boi:** We never wanted to use AI just because we could.

Performance remained real. Physical interaction remained important. Then we looked at the environment around that and asked where technology could give us more freedom.

**Kim:** It's not really AI versus VFX. Different parts of the image need different tools.

We worked alongside established VFX processes throughout the production. The question was always: what gives us the control we need for this particular shot?

AI becomes another part of that pipeline, rather than replacing it.

## Has Crooks changed how you think about future productions?

**Boi:** I'm opening the drawer again. I've put away projects because I knew what they would cost to make. Some ideas were too big to realise with the resources available. Now I'm looking at those projects again.

That's what has changed for me. It's not about removing actors, crews or filmmaking from the process. It's being able to look at a story that previously felt impossible and ask whether there is now another way to make it.

Technology can make the world larger without making the performance smaller.

## What could this mean for BYKAI and overseas productions?

**Boi:** *Crooks* gives us a practical case study to discuss with other filmmakers. Through BYKAI, we want to work with producers and studios internationally to explore whether this approach can help realise their stories. We would start with their script and creative goals, identify scenes that suit a hybrid approach, and test a defined sequence before discussing a larger production.

**Kim:** Every project needs its own assessment. We would look at what must be captured on set, which environments can be developed digitally, and how the work will fit the existing production and post-production teams. The aim is to make our experience useful to another production while keeping creative decisions with the filmmakers.


# Going bigger: UW Credit Union's game-day campaign
Source: https://beeble.ai/showcase/mcm-creative-uw-credit-union-game-day-campaign
Published: 2026-10-05

How MCM Creative and Re(Ai)lism used Beeble to put a packed 63,000-seat stadium behind green-screen actors for a UW Credit Union commercial.

For a new UW Credit Union Rewards Checking commercial, creative agency [The Weaponry](https://www.theweaponry.com/) had an idea built around game day: a referee on the field, fans in the stands and the unmistakable energy of a packed Camp Randall Stadium. Instead of calling a foul on the game, the referee would call one on an unnecessary bank fee. There was just one problem – there was no game.

The production already had six other commercials to shoot in Milwaukee, while Camp Randall Stadium was in Madison, around an hour and 45 minutes away. Shooting there would have meant moving the unit, managing hundreds of extras and still likely extending the crowd in post. The traditional alternative was to compress the action with a long lens, keep the angles tight and never let the audience see just how empty the stadium really was.

Director Adam Jones wanted the opposite. He wanted to go wide. Working with The Weaponry and a partnership between [MCM Creative](http://mcmcreativestudios.com/) and [Re(Ai)lism](https://www.re-ai-lism.com/), Jones shot the performances on green screen. The team then built the game-day world around the actors, combining generative environments, traditional post-production and Beeble to create the scale of a packed stadium while preserving the performances at the heart of the commercial.

The technology didn't just help them execute the original idea. It gave them room to make it bigger. "We can have 63,000 fans in the background and celebrate that with the camera angles," says Jones. "We can create the energy of a full stadium." That freedom continued after the cameras stopped rolling. Lighting could be developed in post. Environments could be refined. An improvised joke between the actors eventually grew into an entirely new ending involving thousands of fans, a stadium chant and a "Kevin Cam."

We spoke with **Jones**; **Adam Albrecht**, head of creative agency The Weaponry; **Mike Canzoniero**, owner of MCM Creative; and **Edward McGurn**, founder of Re(Ai)lism and AI Creative Director and AI/VFX Supervisor on the campaign, about how they built the workflow, what they learned from putting it into production, and where they plan to take it next.

## Where did the original idea for the campaign come from?

**Adam Albrecht:** UW Credit Union was launching Rewards Checking, which doesn't have scheduled fees. Our concept was essentially: *call foul on unnecessary fees.*

We developed a scenario where a referee calls a foul, but instead of calling something that happened on the field, he calls a foul on a bank fee. Then we break the fourth wall between the referee and the people in the stands.

We could have shot that very small. We could have gone to a local game with 20 people and cropped everything in. But Adam saw an opportunity to make it much grander using this technology.

The team brought the concept to life exactly as we imagined it, without compromise.

## What would creating that stadium environment have involved traditionally?

**Adam Jones:** We had six other commercials to shoot in Milwaukee, and the stadium is in Madison, about an hour and 45 minutes away.

Traditionally, we'd probably use a 200mm lens and keep the field of view very narrow. Maybe we'd have 200 extras, spread them out, wardrobe them, feed them, wrangle them and then augment that crowd with VFX.

But we'd also have to move the entire unit to Madison, potentially put the crew up overnight and deal with all the logistics around that.

Beeble enabled us to do something completely different. We can have 63,000 fans in the background and actually celebrate that with the camera angles. We can create the energy of a full stadium. That scale lets our imagination, as creative people, be limitless.

**Edward McGurn:** We weren't just creating a generic stadium either. Adam picked the exact seats where the characters would be sitting. We got a LiDAR scan of the stadium and sent a scout to photograph the views from those seats, so when you're looking left, right or straight ahead, those perspectives correspond to where the characters actually are.

## How did MCM approach the project?

**Edward McGurn:** We're building what we see as an integrated AI production studio centred on performance and hybrid production. MCM already has the infrastructure: green-screen stages, editing rooms, sound mix, colour and post; so with tools like Beeble we're able to bring the whole process together under one roof.

**Mike Canzoniero:** I'm coming at this as a film director who loves indie film. When AI came along, I saw it as a leveller with Hollywood. Indie film could never keep up with Hollywood in terms of spending and effects, and suddenly there's an opportunity to level that playing field.

At the same time, I love actors. I got into filmmaking largely because I love the experience of directing actors. What Beeble offers me is the opportunity to have my cake and eat it. I can dream up sets we might not have been able to afford before, but I still get the luxury of working with great actors and getting great performances.

For me, that's the hybrid model. It's not right for every scenario, but keeping actors in the mix is a really important part of the magic.

## How did you know the approach was going to work?

**Edward McGurn:** We tested it. We had a lot of questions to answer before putting the company's reputation behind the workflow. How would the backgrounds hold up? How would the perspective work? What would happen with green spill? Could we integrate the foreground and background convincingly?

We took the early script, shot a couple of test scenes and put them through SwitchX with a basic generated background. The first results blew us away.

We were trying to establish whether we could credibly offer this as a service and stake our reputation on being able to deliver it. Once we saw those tests, we knew there was something there.

**Mike Canzoniero:** It was shockingly intuitive how it handled our test, and it gave us the confidence that we could deliver. Perspective was something I was particularly worried about. We've been doing green screen forever, and I know what a nightmare it can be getting a proper plate to work. Now you're adding 60,000 extras and trying to make them feel real and make the perspective match the lens.

**Edward McGurn:** The important distinction is that this commercial wasn't impossible to make before. It wasn't possible at this budget range before. We saved at least 30% compared to a traditional post workflow.

**Adam Albrecht:** Not even close. This isn't a national bank buying a Super Bowl commercial. It's a regional credit union in Wisconsin.

That's what was exciting – suddenly we could think at that scale.

## Where did Beeble make the biggest difference?

**Mike Canzoniero:** For me, the lightbulb moment was the Beeble relight. Until then, it still felt like a great green-screen composite with this dynamic background. But once we got the relighting right across the plate and the actors, it tightened everything up and became one world.

It didn't just happen. Our team played with it a lot to find the right relight and the right reference. But when we got it right, it clicked. That's when I thought: this is a tool we can use in movies, we can use in commercials.

**Adam Jones:** We were able to shoot a lot faster because of the flat lighting. We weren't constantly moving things around. We could concentrate on performance and then make some of those cinematography decisions in post – colour, light, time of day.

I didn't feel like I had less creative control. I felt creatively liberated.

## What did the workflow look like from set through to the finished spot?

**Edward McGurn:** We shot on green screen and had reference plates available on set, but those weren't the final backgrounds. We deliberately lit everything fairly flat and neutral, knowing that we could develop the lighting later with Beeble.

**Mike Canzoniero:** We actually had our Beeble tech on set, and within about 20 minutes of shooting we were able to show the client something much closer to what the final result was going to look like.

The person we trained on Beeble came to the table with skills in DaVinci, After Effects and Blender. He's a colourist and an editor, so he was already well versed in post and effects work. But he was essentially iterating a workflow for himself where there wasn't an existing model to follow.

We were learning rules as we went and becoming more confident about what we'd do differently next time. I see the post workflow as an assembly line of skilled people: an editor, sound designer, generative artist working on plates, compositing and then the Beeble and colour work.

There were a lot of people working on this spot and a lot of labour involved. It's about elevating the quality, not one kid sitting in a basement prompting three things and suddenly having an ad. That's not the workflow.

**Edward McGurn:** This still went through colour, sound mix, sound design and editing. It was a full post-production process, just in a slightly different configuration.

## Did working this way change how you directed the actors?

**Adam Jones:** Not really. What I would have had the actors do would have been exactly the same if we were at Camp Randall Stadium. Professional actors are already used to imagining what's around them.

I love being on location, and I don't see this replacing physical production, but there are a lot of situations where it makes sense to create the world around the actors.

**Mike Canzoniero:** I would challenge you to get that kind of nuanced comedy performance from a generative actor right now. We're working with generative performance ourselves, and it's not there yet – especially for comedy or anything subtle. Right now, real actors are the way to go.

If you can put those real actors into AI environments where the audience can't tell the difference, that's where this hybrid approach becomes really interesting.

## And it sounds like the flexibility continued after the shoot.

**Adam Albrecht:** That's where we really saw the best of both worlds. Because we had real actors, we could try alternate lines on set. One of those gave us a completely different idea for the ending.

The character asks, "How did the ref know your name was Kevin?" and the response is basically, "Everyone knows I'm Kevin."

Once we saw that in the edit, we realised there was an opportunity to create a completely different ending.

**Edward McGurn:** That actually came from James, our generative artist. I asked if we could do anything bigger, and he suggested having the crowd chant "Kevin." Then it became, "Let's put him on the jumbotron," and from there we got to the "Kevin Cam."

**Mike Canzoniero:** To me, that sounds like human creativity, not AI coming up with ideas. It's people being creative with tools and getting the chance to iterate in a way that human creativity does. It comes back to the quality of the work you can get when you use AI as a tool and don't expect it to do everything for you. There's no magic button.

**Adam Albrecht:** It was collaboration between the director, agency and production team. We weren't protecting our individual ideas. We were asking: how do we make the best spot?

## Adam, from an agency perspective, did the experience change how you think about what's possible for clients?

**Adam Albrecht:** We went into it with high hopes around access, timing and scale, and the technology lived up to that.

Before we started in June 2026, so much of this was theoretical. By the end of the summer, we'd seen that it was real, implementable and capable of producing what we wanted.

For people who come up with ideas, that's incredibly exciting. **We can think big and believe we can bring those ideas to life. It's not just theory anymore.**

## What are you looking to push further on the next production?

**Mike Canzoniero:** Movement. That's the next thing for us. These were locked-off cameras. On the next spot, we're moving the camera and seeing how Beeble handles parallax.

We're also using an XR production workflow. We're pre-building an Unreal environment that we can work with on set, and then we'll develop that further in post with Beeble.

## What does having this capability mean for you as a director?

**Adam Jones:** As a commercial director, I'm competing against two or three other directors on every job I pitch. Right on the heels of this UW project, I was up for a gig where my competitors' approaches included a constructed set, a volume wall, and a live comp on green screen. I was the only one saying, "What about this?"

Ed helped me with the treatment, so we could practically visualise the spot and show what this approach could offer, and we landed it.

Right now there's a window where understanding these tools can keep me on the cusp of the wave.

For me, it's a superpower. I'm embracing it.

## Credits

- **Adam Jones** – Director of the UWCU spot
- **Adam Albrecht** – Head of the creative agency on the project (The Weaponry)
- **Edward McGurn** – Executive Producer | Head of AI Production, MCM Creative Studios | Founder, Re(Ai)lism
- **Mike Canzoniero** – Owner | MCM Creative Studios
- **Jeremiah Black** – VFX Artist and Beeble Artist
- **James Hussey** – Generative Artist
- **Gab Grieco** – Marketing Manager | MCM Creative Studios
- **Sophie Gould** – Production Coordinator | MCM Creative Studios


# FilmEvolve: Iterative filmmaking with Beeble
Source: https://beeble.ai/showcase/filmevolve-iterative-filmmaking-with-beeble
Published: 2026-09-17

FilmEvolve founders Mark Harris and Avi Mack on relighting captured footage with Beeble while keeping creative decisions with the filmmaker.

For filmmakers, some of the biggest creative decisions have traditionally become fixed the moment the camera stops rolling. The weather you had, the location you could afford, and the on-set lighting all become part of the image you take into post-production. [FilmEvolve](https://filmevolve.ai/) sees AI changing that relationship, giving filmmakers more room to refine what they have already created.

Founded by **filmmakers Mark Harris and Avi Mack**, FilmEvolve is a production and consulting company focused on integrating AI ethically into professional filmmaking workflows. Its approach is firmly rooted in practical production: working with actors, crews and captured footage, then using tools such as Beeble to extend what can be achieved with those ingredients.

FilmEvolve is now applying that philosophy to a slate of projects, including upcoming feature films for which Beeble is being deployed from pre-production through production and post.

We spoke with Mark and Avi about why they see AI as a filmmaking tool, how relighting is changing decisions on set, and why creative control remains at the center of their approach.

## Can you introduce FilmEvolve and what you’re working on?

**Mark Harris:** I’m a film writer, TV producer and executive by trade, and a partner with Avi in FilmEvolve. We’re a consulting and production services company helping integrate AI ethically into professional filmmaking and media workflows.

We also have our own slate. That includes *The Platform*, a short film currently on the festival circuit, as well as premium series and multiple feature films in development.

A big part of what we do is helping filmmakers and producers understand how these technologies can fit into professional production without replacing the people who make films.

## What does integrating AI “ethically” mean to you?

**Avi Mack:** From the beginning, our question was: how do we engage with this technology without using it to replace people?

We could see that AI was developing incredibly quickly and would change filmmaking substantially. We wanted to understand how it could help us make films differently, make more ambitious work, and give more people access to filmmaking tools, without replacing actors or crew members.

There are filmmakers around the world who don’t necessarily have access to large crews, infrastructure, or budgets. If these tools allow them to explore their creativity and tell stories they otherwise couldn’t make, that’s incredibly exciting.

For us, it’s about amplification rather than replacement.

**Mark Harris:** When people hear AI, they often think about generative images or video. But I think its strongest application is as a production tool.

That’s why Beeble was so interesting to us. It solves recognizable filmmaking problems. It isn’t about asking AI to make the film for you; it’s about giving filmmakers another set of tools with which to make creative decisions.

## What opportunities does that open up in production?

**Mark:** We think about it as a hybrid workflow. We’re shooting practically, with actors, and we’re not replacing performances. But once that photography exists, AI gives us another creative pass.

Previously, you were largely committed to the lighting you captured or the background that was there. Now those elements can become iterative.

Production starts moving from **“we hope we got it” to “let’s keep refining it.”**

There’s also an opportunity to spend more of the production day on the things that matter most. If you can reduce the time spent capturing certain inserts or dealing with logistical limitations, you can spend more time working with performers and concentrating on the emotional center of the scene.

## How does Beeble specifically change that relationship with what you capture?

**Avi:** Weather is a really simple example. Every filmmaker has been in a situation where a scene is supposed to take place on a beautiful sunny day, but the day you have to shoot is overcast.

Traditionally, you might simply have to accept that because the cast, crew, and location are booked and you can’t afford to move the shoot.

Knowing you can revisit the lighting afterward removes some of that pressure. You still want to get the shot as close as possible in camera. That doesn’t change, but you’re no longer completely stuck with circumstances you couldn’t control.

We have an example on a shoot in Greece. We had been told the sun would set over the ocean at the beach we were using, but when we arrived, we discovered it actually set on the other side of the island.

The footage itself was beautiful, but it wasn’t what we had originally imagined. As a test, I used Beeble to relight it as sunset footage. We created both a grounded version, with the subject more silhouetted, and a more cinematic version. The original performance and photography remained intact, but suddenly we had several creative directions available to us.

That’s the difference: you don’t just have the opportunity to fix something. You have options.

## Does knowing that affect how you plan a shoot in the first place?

**Mark:** Completely.

I think Beeble has the ability to make lighting and location feel iterative rather than permanent. It becomes less about capturing the absolute final image when you shoot and more about capturing the **ingredients for the final image**.

We’re developing a slate of three feature films in Oklahoma and thinking about Beeble across pre-production, production, and post.

It changes some very fundamental production questions. If we have a scene inside the hull of a spacecraft, for example, do we need an expensive practical build? Do we shoot it on an LED volume? Or could we shoot in a much simpler environment and reimagine that space later?

Knowing these options exist changes the conversation before you ever get to set.

## Could these tools become part of the on-set workflow too?

**Avi:** That’s something we’re very interested in.

Could we have someone on set whose job is to work with these tools while we’re shooting? We could capture a shot, quickly explore what it might look like with a different lighting treatment or environment, and use that information to influence the next decision.

The important part is that it would still be iterative afterward. We’d be capturing high-quality footage, but we’d also have a much clearer indication of what the eventual shot could become.

Large productions already use extensive previs and on-set VFX infrastructure to help filmmakers understand what a final image might look like. What’s exciting is the possibility of bringing some of that capability to productions that don't have a $200 million budget.

## Has that made you consider ideas you might previously have dismissed as too ambitious?

**Mark:** All the time. One of our upcoming projects is *Affliction*, a psychological thriller that we’re currently planning and likely including as part of the Oklahoma slate. From the outset, we’ve been exploring how virtual production and Beeble could help us reimagine environments and build scale into the practical shoot.

That fundamentally affects what you believe is possible for an independent film.

## You also tested Beeble on *The Platform*. What did you learn from that?

**Mark:** *The Platform* includes a sequence set in Grand Central Station, which we shot using an LED volume. We tested Beeble on some of that footage and immediately saw that the relighting could add richness, atmosphere, and a surprising level of realism. It was extremely encouraging.

At the same time, Grand Central is an unusually difficult environment because it’s iconic and instantly recognizable. Very small architectural inconsistencies can break the illusion. That made it a useful test because it showed both where these tools are already incredibly strong and where specific use cases still present challenges.

## Have there been smaller, everyday production problems where Beeble has proved useful?

**Avi:** Absolutely. I interviewed a composer in a very small studio that wasn’t particularly well lit. There was neither the time nor the space to create the kind of lighting setup I would ideally have wanted.

I brought the footage into Beeble and turned something quite flat into an image that felt much more considered and professional.

That wasn’t about creating something fantastical. It solved a very ordinary production problem that filmmakers encounter all the time.

## Creative control comes up repeatedly when you talk about AI. Why is that so important?

**Mark:** Creative control is everything.

AI should execute creative intent; it shouldn’t make the creative decisions.

That’s what I find compelling about Beeble. Understanding what the technology can do informs how you shoot, how you plan production, and how you approach post, but the filmmaker is still making the decisions.

**Avi:** Filmmaking is ultimately about having a point of view and having taste.

A lot of what these technologies enable was technically possible before, whether that was relighting, replacing backgrounds, or extensive visual effects. But it could require a huge amount of infrastructure, money, and specialist work.

Making those capabilities more accessible gives more filmmakers creative control.

And Beeble feels specifically like a filmmaker’s tool. It isn’t simply asking someone to type a prompt and let a model make all the decisions. It helps bridge the gap between what you practically photographed and what you had in your mind’s eye.

It allows you to keep the cinematography, performance, and creative choices you made on set while continuing to shape the image.

## Where would you like to see this approach go next?

**Mark:** I think resistance around AI is beginning to soften as people understand the breadth of what the technology actually offers. Guardrails around creative rights and training data remain incredibly important, but I think there’s also a growing understanding that these technologies can become legitimate filmmaking tools.

What excites us is giving filmmakers who might not have had access to certain production capabilities the opportunity to make much more ambitious work.

**Avi:** I’d also love to see more collaboration between AI companies and the existing crafts in filmmaking.

Imagine a lighting artist or gaffer being able to develop their own distinctive approach within these systems, or camera manufacturers integrating AI tools directly into the way filmmakers work on set.

The technical barriers may continue to fall, but the creative part doesn’t disappear. If anything, taste, storytelling, and having something original to say become even more important.

That’s where I think this becomes really interesting.

## Finally, what makes Beeble particularly interesting to FilmEvolve?

**Mark:** When I first met Hoon and learned what Beeble was building, I immediately saw it as an incredibly powerful producer’s tool because it was solving real-world production problems.

We’re now building an AI-native production workflow across our slate, and we see Beeble becoming a foundational technology within that.

There are countless generative tools, and they keep leapfrogging one another. Beeble feels different because it’s addressing very specific filmmaking needs.

It began as something we understood primarily through relighting, but it is becoming much broader: a compositing tool, a VFX tool, and ultimately a production tool.

For filmmakers, that’s incredibly exciting.


# How Second Image used Beeble to build four game worlds for Google Play
Source: https://beeble.ai/showcase/second-image-google-play-indonesia-game-worlds
Published: 2026-09-04

Second Image Studio combined practical sets, Beeble, AI tools, and traditional VFX to build four game worlds for a Google Play Indonesia campaign.

When Bangkok-based Second Image Studio took on a Google Play Indonesia commercial, the brief called for its protagonist to move through four distinct game-inspired worlds, each with its own environments, lighting, costumes, and visual language.

The creative ambition was substantial. The schedule was even more challenging. Second Image had around two weeks of pre-production and three weeks of post-production, with more than 40 shots moving through review cycles lasting five to seven days.

Building every environment using a conventional CG pipeline would have required considerably more artists, rendering time, and effort, and would have made major creative changes increasingly difficult as the deadline approached.

Instead, VFX Supervisor **Ronnachai Changsri (Nat)** developed a hybrid workflow that combines practical set pieces, AI tools, Beeble, and traditional VFX. Working closely with director **Chaw Khana**, the Second Image team developed and refined the visual design of the environments throughout the project.

The approach allowed Second Image to iterate on environments quickly while keeping artists in control of the look, compositing, and final finish.

We spoke with Nat about designing the workflow, what he learned from using Beeble on a production for the first time, and why he sees AI becoming another part of the VFX pipeline rather than a replacement for it.

## Nat, can you tell us a little about yourself and Second Image Studio?

I’m a CG Supervisor, and I run Second Image Studio in Bangkok. I started the company around four years ago. Before that, I worked at studios in South Korea, Malaysia, Singapore, and Thailand, eventually returning to Bangkok to start my own project.

We’re actually a very small team. There are only two permanent employees, and then we build out the rest of the team with freelancers and contract artists depending on the project.

## What was the brief for the Google Play Indonesia commercial?

Google Play was launching a campaign to promote its Play Points rewards, offering users additional points when they top up or make purchases on the platform.

The promo involved several gaming partners, including MLBB, Free Fire, and Roblox, and the director wanted to bring those different game worlds into the film. To achieve that, we divided the commercial into four distinct worlds. The main character remains the same, but his outfit, environment, lighting, and what happens around him change from world to world.

One of our biggest challenges was the schedule. We had around two weeks for pre-production and approximately three weeks for post. We knew from the beginning that the environments would continue developing through close collaboration between the director, client, agency, and our VFX team, so we needed a workflow flexible enough to support that process.

If we had built all those environments using a traditional CG workflow, it would have required many more artists and much more time. More importantly, whenever something changed, we might have had to go back and rebuild or rework a large part of it.

That was where Beeble came in.

## How did you approach the shoot knowing you would be using this workflow?

The director wanted to shoot everything on green screen, but we also used practical props and set pieces.

I had actually spoken with someone from Beeble on Discord beforehand about working with camera movement. They suggested including practical objects in the scene to provide information about the space and camera movement.

So I encouraged the production to include as many useful, practical elements on set as possible, including stones, ruins, glass, leaves, vehicles, and other objects that belonged in the scenes. Then we could build out the rest afterward.

That turned out to be really important.

## Was this your first time using Beeble on a production?

Yes. I had heard about Beeble before and had tested it because I thought the results looked amazing, but those were really just experiments.

This was the first time I used it on a real production.

## What did the workflow look like once the footage came into post?

It wasn't a case of shooting an actor on green screen, putting the footage into Beeble and delivering the result. There were artists involved both before and after Beeble.

We would start by developing the environment and creating a rough slap comp using the original footage and the background. At that point, you might still have the green screen, hard edges, and other things that obviously weren't integrated.

We used other AI image tools to help develop a single reference frame, blending edges and working out things like lighting and contact shadows. Once we had a reference we were happy with, we could import it into Beeble and use it to handle the footage, including camera movement and environment replacement.

After that, we would return to the traditional VFX process and add everything needed to complete the shot: smoke, leaves, crystals, lens flares, atmosphere, and other CG or compositing elements.

So Beeble became one part of a much larger pipeline.

## How important was iteration on this project?

It was probably one of the biggest advantages. We had more than 40 shots, and in some review cycles we only had five to seven days. There were also significant creative changes.

For one environment, for example, we already had a background that was working well when the director decided he wanted to introduce additional ruined structures. We didn't want to regenerate everything from scratch because we needed to preserve what was already working.

We could retouch the reference, create a new slap comp with the additional structures, and then use Beeble to bring that revised environment back into the moving shot.

That flexibility meant we could leave some of the technical problems behind and spend more of our available time making creative decisions. When you get results faster, you have more time to ask, *“How can we make this better?”*

That was something new for us. We had never worked this fast before.

## Did anything particularly surprise you about what Beeble could handle?

The gun transformation really surprised me. Normally, we used Beeble mainly for backgrounds and relatively static elements, and then we would composite moving elements. In this sequence, however, the character's gun needed to transform from a small gun into a much larger weapon.

Originally, I planned to do that using traditional CG. We had already created the model and textures. But I decided to test whether we could achieve it with the AI workflow instead. The result from Beeble really surprised me. There were other things in the shot that still needed work, but the gun itself worked incredibly well. I didn't expect that.

## How did you maintain consistency when the same world appeared from different camera angles?

That was another important part of the workflow. We might see the same world from the left, then from the right, and then from another angle. Each shot needed to look believable, but the audience also had to immediately understand that all the shots were taking place in the same world. It's not enough for every shot to look good on its own. They have to belong together.

The speed of iteration helped us develop that consistency across the sequence rather than simply generating a completely different environment every time.

## Which sequence was the most challenging?

The final castle sequence was probably the toughest. The director was closely involved in developing the environment with us, and we went through several major iterations of the castle design before arriving at the final look. If we had been working entirely in traditional CG, those kinds of changes could have meant going back into modeling and repeatedly reworking the geometry.

Using this workflow allowed us to make those creative decisions much faster. Beeble gave us a strong starting point for the environment, but because this was the final sequence of the film, where everyone comes together for a celebration, the director wanted it to feel much bigger and more alive.

We therefore added extensive CG and compositing on top of the Beeble result to create the festival atmosphere and complete the sequence.

## What difference did the workflow make to the overall production?

For the full film, it's difficult to make an exact comparison, but a conventional approach would likely have required a significantly longer schedule and a larger team, given the number of environments, revisions, and level of detail involved. There are cost benefits, too, particularly when you consider the number of artists and the render time required to build all these environments traditionally.

But I think it's important to be clear that this doesn't remove the artist. We still needed artists to create the slap comps and develop the references before Beeble, and later on we still needed artists for smoke, lens flares, atmosphere, CG elements, and all the other details that make the final image work.

## What did the project teach you about shooting for an AI-assisted VFX workflow?

Practical elements matter a lot, particularly when the camera is moving. We had real objects on set, including things like the car in one of the sequences, and they gave the shot structure. That helped enormously when we built the environment around the footage.

Since then, I've had clients ask whether we can simply shoot an actor alone against a green screen and create everything else afterward with AI. For a static camera, you might be able to do that. But with a moving camera, I wouldn't approach it that way. Based on what we learned from this project, I would definitely plan practical props and set elements into the shoot again.

## Having used AI throughout this production, do you think it changes the role of the VFX artist?

Clients ask me this a lot: *Is AI going to replace CG artists?*

After using so many AI tools on this project, I feel they cannot think for us. We need to think for them and then let them do their work. There are still creative decisions being made by artists before and after the AI. For me, I see AI becoming almost like a new department within the pipeline rather than simply replacing the existing ones.

There are certainly tasks it can reduce. On this project, for example, we didn't do traditional rotoscoping. We only needed some simple masking around the face. That alone saves a lot of time. But the creative decisions still come from the artists.

## Ready to explore Beeble?

---

## Project credits

- **Client:** Google Play
- **Production House:** Player 2 Bangkok
- **Film Director:** Chaw Khana
- **Executive Producer:** Woei Woei
- **Film Producer:** Melvin Wong, Ing Mahayotsanun
- **Assistant Director:** Boum Wongwatpong
- **Post Producer:** Panitta Trang
- **VFX:** Second Image Studio
- **VFX Supervisor:** Ronnachai Changsri (Nat)
- **VFX Producer:** Kanyanat Chomchoei
- **CG Supervisor:** Nattapong Surakittidate
- **Assets Artist:** Worachot Ruangsri
- **Rigging:** Sorawit Limkertpol
- **Animator:** Chayanee Denmuanvong, Aniwut Yodkaew
- **Lighting Artist:** First Nattapong, Teerasan Suppasri
- **Compositor:** Tanaklit Ketkaew, Phanin Paichoe, Ronnachai Changsri


# Ocean Jet: Creating a Dreamlike World for the Entwined music video with SwitchX
Source: https://beeble.ai/showcase/ocean-jet-entwined-music-video-switchx
Published: 2026-07-22

Musician and filmmaker Max Scotch used SwitchX to build the dreamlike world of Ocean Jet's Entwined music video — creating 95% of it entirely on his own.

When musician and filmmaker Max Scotch set out to create the music video for Ocean Jet's latest single Entwined, he envisioned an otherworldly journey through a dreamlike landscape suspended between heaven and hell.

Originally planned as a traditional green-screen production with environments built in Unreal Engine, the project quickly grew beyond what was practical for an independent creator. Discovering Beeble's SwitchX transformed the production, allowing Max to complete nearly the entire visual world himself.

We spoke with Max about the creative vision behind Entwined, his experience using SwitchX, and how AI-powered filmmaking is opening new possibilities for independent artists.

## Can you introduce yourself and Ocean Jet, and tell us a little about the song behind this music video?

Hi, I'm Max Scotch. I'm 39 years old, and I've been passionate about music and visual arts since I was a kid.

In 2013, my friend Jon Myred and I formed the band Ocean Jet. We write and perform music in English, blending Synth Pop and Alternative Rock. Our song Entwined was co-written with my wife, Lesia Scotch, in 2022, the same year she joined the band.

The song is about two people who love each other and choose to stay together, no matter what happens.

## What kind of visual world were you hoping to create?

I wanted to create an unreal, dreamlike world where the audience feels that everything is happening somewhere beyond our reality.

Only at the end of the video do we discover that this place is a kind of limbo, a space between heaven and hell, where the two main characters are searching for each other so they can be reborn and meet again in their next life.

I also wanted to pay tribute to one of my favorite films, What Dreams May Come by Vincent Ward.

![Dreamlike environment from the Entwined music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-02.webp)

![Still from the Entwined music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-03.webp)

## You took on almost every creative role yourself. What was the biggest challenge in bringing that vision to life?

I wrote the script, directed the video, edited it, created the visual effects, and handled the color grading.

This was my first experience working on a project with visual effects at this scale. Before tools like SwitchX, creating something like this would have required far more time, resources, and technical experience. It completely changed what was possible for an independent creator working alone.

## The project originally started with plans to build the environments in Unreal Engine. What changed, and how did SwitchX become part of your workflow?

I first discovered SwitchX through an Instagram Reel introducing the technology. My immediate thought was that it could be a game-changer and finally allow me to complete this music video by myself.

We originally shot the footage on a green screen in May 2025, intending to build the entire world in Unreal Engine. Unfortunately, I couldn't make that approach work, even with help from 3D artists.

That's when I decided to rebuild everything using SwitchX, and the final result exceeded my expectations.

![Environment built with SwitchX for the Entwined music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-04.webp)

![Performer composited into a SwitchX environment](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-05.webp)

## How much of the finished video relied on SwitchX, and what did it make possible that would otherwise have been difficult or out of reach?

Around 95% of the finished music video was created using SwitchX. The remaining 5% consists of shots that weren't filmed during production, so I created those separately using other tools.

SwitchX was exactly what I needed because it could replace backgrounds while preserving the performers naturally. No other solution offered that combination at the time.

Building the environments in Unreal Engine would have taken significantly longer, required much more work, and been far more expensive. Some of the visual ideas simply wouldn't have been practical to achieve with the same level of realism.

![SwitchX background replacement in the Entwined music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-06.webp)

![Still from the Entwined music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-07.webp)

## As someone creating large-scale visual effects for the first time, what surprised you most about working with SwitchX?

This was my first serious visual effects project, so I had to learn everything from scratch.

SwitchX made that process surprisingly approachable. Instead of spending months learning complex VFX pipelines, I could focus on the creative side of building the world I imagined.

I was especially impressed by how naturally the characters blended into each environment. Details like reflections on wet surfaces, water interacting with the actors, and the automatic adaptation of lighting and color made everything feel incredibly realistic.

Those kinds of effects would normally require a huge amount of manual work, so having them happen automatically made the entire workflow much faster.

![Reflections and lighting adapted automatically by SwitchX](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-08.webp)

![Character blended into a SwitchX environment](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-09.webp)

## Is there one moment in Entwined that best represents what you were able to achieve with SwitchX?

One of my favorite moments is actually quite a simple composition, the sequence where the girl walks across the water.

Because I had originally planned to create the entire environment in 3D, that scene felt like the biggest technical challenge. With SwitchX, however, everything came together exactly as I had imagined.

![The walking-on-water sequence from Entwined](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-10.webp)

![Still from the walking-on-water sequence](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-11.webp)

## What do tools like SwitchX mean for independent musicians and filmmakers?

It opens the door to ideas that would previously have been impossible because of budget or technical limitations.

Now I can film almost anywhere and build an entirely new world around the footage from my laptop at home, without needing a large team, huge amounts of time, or an expensive production.

![Environment created for the Entwined music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-12.webp)

![Still from the Entwined music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-13.webp)

## What's next for Ocean Jet, and how do you see AI fitting into your future projects?

The members of Ocean Jet all live in different cities (although my wife and I live together), and today's music industry demands a constant stream of video content.

Using AI tools like SwitchX makes that much easier. We can each film our parts at home, and I can bring everyone together in the same environment as though we were all on set.

I'm already working on another short music video. Including Entwined, I've now completed three projects using SwitchX, and I expect many more to come.

![Ocean Jet band members composited into one environment](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-14.webp)

![Final still from the Entwined music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/ocean-jet-entwined-music-video-switchx/entwined-15.webp)

**Credits**\
Director, Screenwriter, Editor, VFX & Colorist: Max Scotch\
Assistant Director: Anna Ozhigova\
Director of Photography: Pavel Matveev\
Gaffer: David Tarasov

---


# How Peregrine Films combined practical filmmaking and AI to create an epic sci-fi music video
Source: https://beeble.ai/showcase/peregrine-films-switchx-sci-fi-music-video
Published: 2026-07-22

Berlin filmmaker Lucas Fiederling turned a €4,000 green screen shoot into a Kung Fury-inspired sci-fi music video with SwitchX, processing around 140 shots.

Berlin-based filmmaker [Lucas Fiederling](http://instagram.com/lucasfiederling) has spent most of his career moving between directing, cinematography, VFX, documentaries, music videos and commercial work. As founder of [Peregrine Films](https://peregrine-films.com/), he has always combined practical filmmaking with visual effects, making him uniquely positioned to explore the emerging world of AI-assisted production.

Recently, Lucas used SwitchX to create a stylised sci-fi music video inspired by Kung Fury, transforming a green screen studio shoot into a world of dinosaurs, sports cars and retro action sequences. At the same time, he's evaluating the technology for an ambitious independent sci-fi feature film currently in development.

We spoke with Lucas about hybrid filmmaking, anamorphic lenses, and why he believes AI works best when combined with traditional production techniques.

## Tell us about the music video project. What was the concept, and what role did SwitchX play?

The project was for the German metal band Endseeker. It was their final music video before breaking up, so they wanted to do something special.

The concept was inspired by Kung Fury. We wanted dinosaurs, a white Lamborghini, car chases and all the crazy sci-fi action that comes with that aesthetic.

The problem was that the budget was only around €4,000.

A majority of the shots were filmed in a green screen studio and then transformed using SwitchX. In total, I think I processed around 140 shots. Additional I took photos of the band and shot them sitting and performing in a car, and used Seadance and Kling to get some additional wide shots and so on.

This kind of project would not have been possible on this budget a few years ago. It would have needed a team of VFX artists working for weeks.

![Green screen studio shoot for the Endseeker music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-02.webp)

![SwitchX-transformed shot from the Endseeker music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-03.webp)

![Sci-fi environment created with SwitchX](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-04.webp)

## What creative or production challenges were you trying to solve?

For me, it was really about testing how far the workflow could go. I didn't want to shoot everything locked off on a tripod and play it safe. I wanted to film it like a normal performance music video.

I was running around with a 21 mm lens, shooting handheld between the performers, doing whip pans, motion blur, and all the things that traditionally make VFX work more difficult.

Also my only other crew member was Joran Ost, my gaffer, and we couldn't always build every shots lighting situation out perfectly to what it would be later in the generated world. So I had to trust SwitchX also on the relighting and recoloring for some shots.

Sometimes I was even shooting into the wrong side of the green screen, and parts of the studio were visible.

Basically every 'mistake' you could make. What surprised me was how well it still worked.

![Handheld performance shot transformed with SwitchX](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-05.webp)

![Band performance in a SwitchX-generated world](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-06.webp)

![Retro sci-fi action sequence from the music video](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-07.webp)

## Were there any shots that particularly demonstrated what SwitchX could do?

The quick pans were probably the biggest surprise.

There are shots where I start on one band member, whip pan across the frame, and land on another person. During the pan, there's almost nothing visible except motion blur.

When I compared the original footage with the SwitchX result, it was tracking things incredibly accurately.

Sometimes I'd think something looked strange in the transformed shot, then I'd switch back to the original footage and realise that's exactly how the motion blur looked in-camera.

It's pretty amazing that it can understand what's happening when there are barely any visible details in the frame.

![Whip pan shot preserved through the SwitchX transformation](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-08.webp)

![Motion blur handled accurately by SwitchX](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-09.webp)

## You're also testing SwitchX for a feature film. Can you tell us about that project?

It's an independent sci-fi feature I'm developing together with Invideo.io.

We've spent the last few months writing the screenplay and creating an 80-page treatment. Almost every scene has already been visualised.

The goal is to make an ambitious science-fiction film using a hybrid workflow. I still want miniatures. I still want practical effects. I still want real sets wherever possible. But now we can combine those things with AI tools and stretch the budget much further.

If we build an asteroid miniature, for example, it's actually more valuable than before because we can generate additional angles, backgrounds, or extensions around it later.

![Previs test for the independent sci-fi feature](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-10.webp)

![Proof-of-concept environment transformation](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-11.webp)

## How are you evaluating SwitchX for the feature?

We've already used it for previs and proof-of-concept work. We shot test scenes in my house and transformed them into completely different environments to see what was possible.

A lot of the testing has been around lenses and cameras because I wanted to understand where the limitations were before we start production.

I haven't really found an example where it doesn't work yet.

## You mentioned testing Alexa Mini footage with anamorphic lenses. What surprised you?

I honestly expected anamorphic footage to be much harder. Those lenses have all kinds of distortions, chromatic aberration, and focus characteristics that usually make VFX work more complicated. But it worked incredibly well.

In some ways, the more character and imperfection the footage has, the more convincing the final result becomes.

The AI does understand what the lens is doing. It helps make everything feel more real.

## Has AI changed the way you think about production design, locations or scheduling?

Definitely. For one, we can now talk to our production designers first and figure out how the budget limits them. Then we can figure out if it potentially helps to extend the set from there using AI; may it be to add backgrounds or anything else we would usually consider to add in VFX but then rarely be able to afford it, at least on the smaller to medium scale projects that I'm used to working on.

Another big change is flexibility. You don't have to go back to a location because you missed a coverage shot. You can test ideas much earlier. You can visualise scenes before production.

A project like this simply wouldn't have been achievable at this budget level before.

![Set extension created with SwitchX](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-12.webp)

![Environment transformation test shot](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-13.webp)

![Scene visualised before production with AI tools](https://desktop.beeble.ai/public/landing/cms/images/showcase/peregrine-films-switchx-sci-fi-music-video/peregrine-14.webp)

## What does your current workflow look like?

I edit in Premiere Pro. I grade in DaVinci Resolve. I use Photoshop heavily for image work and references. For AI generation, I use tools like GPT Image 2 and Nano Banana 2, and then SwitchX sits inside that workflow for compositing, relighting and environment transformation.

I usually like to keep control over things like grain, texture and final image treatment myself, so I often generate something relatively clean and then shape the final look afterwards.

## Looking ahead, how do you think tools like SwitchX will change filmmaking?

I don't think audiences are going to immediately accept fully AI-generated films with no real performances behind them. People still connect with real actors and real emotion.

What excites me is combining those performances with tools that let us create films we couldn't afford before.

If you can work in this hybrid world, I think that is the future right now.

---


# Vortex Films: Production-grade AI for commercial storytelling
Source: https://beeble.ai/showcase/vortex-films
Published: 2026-07-01

How Bangalore studio Vortex Films uses Beeble relighting in a production-grade AI and VFX pipeline for AMD, Nothing India, and RCB campaigns.

For Naveen Kumar, filmmaking has always lived at the intersection of creativity and technology.

As the founder of [Vortex Films](https://vortexfilms.in/), a Bangalore-based production studio working across commercials, music videos, and branded content, Naveen has spent more than a decade exploring how emerging tools can expand what small creative teams can achieve.

Today, Vortex delivers campaigns for brands including AMD, Nothing India, and major sports franchises, combining traditional filmmaking, VFX and AI-assisted workflows to meet the demanding timelines of commercial production.

We spoke with Naveen about building a technology-first production studio, solving impossible post-production challenges, and why he believes AI's greatest value is giving creatives more time to be creative.

## Can you tell us a little about Vortex and how you got started?

I started Vortex when I was still in school. I originally wanted to make music, but one of my teachers suggested I try video instead. I learned Premiere Pro and After Effects through YouTube tutorials and started making videos, short films and explainer animations. Over time, that grew into freelance work, then commercial projects, and eventually Vortex.

One thing that has always been important to me is that we don't separate filmmaking from technology. My background is in engineering, so I've always been excited by new tools and new workflows. I don't consider myself a filmmaking purist. If there's a better way to solve a problem, I want to explore it.

Today we're a small team based in Bangalore, and despite our size, we've worked on projects that have reached hundreds of millions of viewers.

![A Vortex Films production still](https://desktop.beeble.ai/public/landing/cms/images/showcase/vortex-films/vortex-films-2.webp)

## When did Beeble first become part of your workflow?

I've been following the team for years. I remember seeing early demonstrations of SwitchLight and signing up as soon as I could. At the time, we were downloading passes and manually importing them into Blender.

Over the years, it became one of those tools that moved from interesting to essential. My VFX lead started using it on almost every project. Eventually, we realised there wasn't really another tool that could replace what it was doing for relighting. At this point, it's become a core part of our workflow.

The biggest thing is that it works. That's why I keep calling Beeble production-grade. A lot of AI tools look impressive in demos, but when you're working on real client projects under hard deadlines, reliability matters more than anything else.

## You recently shared an AMD campaign that reached more than 80 million views. What was the challenge on that project?

There was one particular shot that became extremely difficult. The concept was that the company's executive appears multiple times inside a server room, representing the idea that he's doing the work of several people at once.

The problem was that the footage wasn't originally shot with that final creative direction in mind. During post-production, the director wanted changes that weren't really supported by the footage we had.

We tried traditional compositing approaches. We tried standard relighting methods. Nothing was landing emotionally.

We tried to avoid bringing the shot into a Blender pipeline because we wanted a faster solution. Eventually, we realised we should have started with Beeble from the beginning. That gave us enough information to relight the character and place him convincingly inside the environment we needed.

What should have been impossible suddenly became achievable. The shot worked exactly as intended, but only because we had the flexibility to rebuild the lighting after production.

## Another project you mentioned was a campaign for Nothing India and RCB. What made that project unusual?

That project was incredibly challenging because we couldn't shoot the cricketers.

The campaign needed to showcase the new Nothing-branded RCB jerseys, but there were restrictions around how the players could be represented. We couldn't generate faces. We couldn't create synthetic versions of the athletes.

So we had to work from existing footage. Every shot came from a different source. Different lighting. Different environments. Different cameras.

We extracted the players, rebuilt jerseys and branding, created new environments, and then used Beeble to relight the athletes so they matched those environments. The relighting step was critical because without it, everything would have looked composited and artificial.

That project really demonstrated how AI can be used within a professional VFX workflow rather than as a shortcut around one.

## You've said AI isn't replacing creativity. How do you see its role?

For me, AI compresses the distance between the idea and the final output. Before, I was spending 14 to 16 hours a day executing work. Today, many repetitive tasks are streamlined, which gives me more time to focus on writing, directing, and developing ideas. Creativity doesn't come from sitting in front of a computer all day. It comes from experiences. It comes from writing. It comes from observing the world and bringing those ideas back into your work. If a tool can reduce a three-day task to thirty minutes, that's not taking away creativity. It's giving me more time to be creative.

What I've noticed personally is that I'm spending less time fighting software and more time thinking about ideas. That's a huge shift.


# How Makaco used SwitchX and Nuke to deliver photoreal VFX at record speed
Source: https://beeble.ai/showcase/makaco-switchx-nuke-photoreal-vfx
Published: 2026-06-19

How Peru's Makaco studio combined SwitchX and Nuke to recreate Paris, London, and New York for a beer campaign — 30+ green screen shots in 2.5 weeks.

[Makaco](https://www.makaco.pe/) is a Peru-based VFX and post-production studio led by Diego Velásquez. Known for its photorealistic visual effects work across film and advertising, the studio specializes in environment replacement, set extensions, character animation, and complex compositing workflows.

With over 15 years of experience working in Nuke, Diego and his team have increasingly explored how AI tools can integrate into high-end production pipelines — not as a replacement for traditional VFX, but as a way to expand creative possibilities, accelerate delivery, and rethink how modern compositing teams operate.

## Tell us about the recent TV commercial you worked on — over 30 green screen shots is no small task. What was the brief?

We had a very important project: the relaunch of Peru's most important beer for the international market. We needed a world-class look and had to recreate locations from Paris, London, and New York, as well as scenes from ancient Peru. The director wanted complete freedom to move the camera with a very handheld, dynamic style.

We were also working against a very tight deadline of two and a half weeks. Recreating all the environments and keying the characters traditionally would have been extremely difficult within that timeline.

We started experimenting with SwitchX. I filmed myself at home without a green screen and placed myself into locations like Sydney and Brooklyn to demonstrate the relighting, tracking, and keying capabilities. Once the director saw the results, he was sold on the workflow.

Using SwitchX even allowed us to make quick tests on set, which helped us choose the correct camera angles and movements during production.

## How did you approach replacing environments and relighting those shots using SwitchX?

First, we created concepts for the backgrounds using AI tools before filming. That way, we already knew how we wanted each shot to look, and we could light the set and position the camera accordingly. But even with careful planning, the lighting on set rarely matched the concepts perfectly. That's where the relighting capabilities of SwitchX became extremely useful.

One of the biggest breakthroughs during the project came when we discovered that we could feed the original concept art directly into SwitchX — even if the character in the image wasn't the same person — and it would recreate the lighting and overall mood almost perfectly.

Once we figured that out, the workflow became much faster and more flexible. We were able to move quickly through the remaining shots while still maintaining a very high-quality result. We then combined the SwitchX output back with the original plates in Nuke to recover the full color range, especially for skin tones, while giving the DP complete control over the final grading process.

The client especially praised the camera movement, background tracking, and the quality of the keying — particularly how much hair detail we were able to preserve. Since then, some of our other work has also been featured on Beeble's Instagram page, including a personal project I had wanted to make for almost five years. Thanks to SwitchX, I'm finally close to finishing it in my free time.

## You said you completed it in "record time" — how does that compare to a more traditional Nuke pipeline?

That first commercial was completed in about two weeks. Using only traditional tools, the budget would probably have been three or four times larger because we would have needed to recreate, track, and project all the backgrounds in 3D.

In terms of timing, it likely would have taken around two months. Even with that schedule, I'm not sure we could have achieved the same results.

Since then, we've continued using SwitchX more and more. Just last week, we finished another spot shot entirely on green screen in only five days after filming, with results that people genuinely believed were shot on location.

## Any other projects you'd like to mention?

Yes — we recently completed another project for the largest bank in Peru in an extremely short timeframe.

We shot everything on green screen and delivered photorealistic results in only five days. We filmed on Wednesday, had the offline ready by Thursday, and delivered the final commercial on Tuesday.

Interestingly, all the client feedback focused on acting and continuity between shots — not on the environments or compositing — almost as if they forgot it had been shot on green screen.

## When did you first come across Beeble, and what made you start using it in production?

Actually, one of the artists in the office first discovered it while looking for alternatives for rotoscoping. As we started using it, we quickly realized how powerful it was for generating normal passes as well.

It soon became our everyday tool of choice for roto and relighting. Once SwitchX came out, we kept finding more and more ways to integrate it into our pipeline.

## You mentioned you're now using Beeble SwitchX every day — why has it become part of your core workflow?

Mainly the speed and ease of use. Like with most new tools, we first experimented with it on personal artistic projects. That helped us understand both its strengths and its limitations. We realized that many of those limitations could be solved by combining the results back with the original plates inside Nuke.

Once we finished our internal tests, we started using it on professional work. The results have been amazing both in terms of output quality and turnaround speed.

## You've been working with Nuke and AI for a while — what drew you into combining those two worlds?

Nuke has been the foundation of our pipeline for around 15 years. Over the last two or three years, we started experimenting more seriously with AI — initially as a way to create short films faster and at a lower cost.

As the technology evolved, we realized it could also replace or enhance parts of our traditional workflows, like face animation or set extensions. Since then, we've been combining the two approaches more and more.

## Has this experience changed how you think about your pipeline or the role of your team?

Completely. Now we approach every project with the question: "Can we do this with SwitchX or AI?" That mindset is influencing how we plan pipeline updates and even how we structure the studio moving forward.

Beeble is allowing our artists to focus much more on the creative side of the work instead of spending so much time on technically complicated green screen workflows.

## How do you balance AI automation with creative control and precision in your work?

AI is affecting our work in almost every department now. Our art direction team uses it daily, not only for concept art, but also for asset creation and even final shots. Our compositors can focus more on creative iterations rather than repetitive roto and keying tasks.

At the same time, our AI department continues to grow each month. Just last week, for example, we completed another commercial mixing fully AI-generated shots with SwitchX footage, using video recorded remotely by a celebrity in Spain while we directed the shoot over Zoom.

## Ready to explore the workflow yourself?

Try SwitchX and explore how studios like Makaco are combining AI and traditional compositing workflows to accelerate production without sacrificing creative control.


# The Month Between April and May: A human-made animation pipeline
Source: https://beeble.ai/showcase/2000blue-month-between-april-and-may
Published: 2026-05-30

Director Kai Tattersall on building a hybrid live-action and rotoscope animation pipeline for 2000blue's immersive film, using Beeble for PBR extraction.

### How 2000blue's Kai Tattersall built a hybrid live-action and rotoscope pipeline for *The Month Between April and May*, using Beeble for PBR extraction while keeping the artwork 100% human-made.

For *The Month Between April and May*, director and visual artist Kai Tattersall is building a film that moves between live action, painted animation and immersive installation.

Created with composer **Myles Ortiz-Green** for **Qualcomm Institute's Recombient Lab**, the project combines wraparound projection and 32-channel spatial sound, but its experimentation extends beyond the exhibition format. Behind the film is a hybrid production process Kai has been refining for years, one that combines performance, rotoscope animation, and technical tools to deliver a fully human-made result.

We spoke with Kai about building that pipeline, why Beeble became part of it, and where he draws the line between assistance and authorship.

## How did *The Month Between April and May* begin?

Myles and I had been talking for a long time about making something shaped by the way we listen to music — especially ambient music, where slowness and duration become part of the emotional experience. We wanted to make a film that moved like that, where things could unfold without urgency.

A lot of that also came out of conversations about where we come from. I'm from Okinawa, and his lineage is from Puerto Rico, and we had very different relationships to island identity, memory and return. There was a tension there that felt worth building a film around. That became the emotional core.

The immersive installation really grew out of the film rather than the other way around. Adapting it for Qualcomm's Recombient Lab let us expand that atmosphere through 32-channel sound and wraparound projection, but I still think of the project as fundamentally filmic. The sound leads a lot of it, which I think is unusual, but that's because Myles has been creatively involved from the very beginning.

[Watch the director sample reel](https://2000blue.com/project/media/kaitattersall/tattersallKai_dirSample_v01.mp4) (MP4, ~159 MB)

## Your work sits somewhere between filmmaking and animation. What draws you to that space?

I came out of film, but I was constantly drifting toward animation, and over time I stopped thinking of them as separate disciplines.

What interests me is that this process lets me tell stories that would be impossible for me to make through traditional means. If this project were fully live action, the budget would be enormous. If it were fully animated, it would be years of labour. This method sits in a kind of middle space where scale becomes possible.

I sometimes describe it as rotoscoping with a lot of cheating, and I mean that affectionately. It's about creative shortcuts — finding ways to move faster without giving up authorship. I can control what I need to preserve the language of the film, like the performance, the look, the production design, while letting animation expand what the world can become.

## You describe the project as 100% human-made. Why is that distinction important?

Because I think where you draw the line matters, and you have to be able to defend where you draw it.

I've been experimenting with AI tools for years, so I'm not speaking from a place of outright rejection of technology. But I didn't want those systems making creative decisions for me. If AI is generating line work, inventing colour, designing environments, for me, that moves away from a human-made product.

What we're doing here feels very different. We're using AI for extraction, not authorship. We're asking for material information — normal maps, albedo maps, matte extraction — things that help us understand the footage, not create the final image.

That distinction is really important to me. The drawing, the painting, the lighting decisions, the visual style, those are still ours. The tool helps us get to the work, but it isn't making the work.

## Where does Beeble fit into your pipeline?

It enters almost immediately after production. We shoot on a soundstage, often with multiple cameras, almost more like theatre than traditional filmmaking, and then that footage comes into the animation pipeline.

Beeble is especially important at the PBR extraction stage. The albedo maps have been hugely useful for us, particularly because they remove green spill from the source footage in a way traditional tools never really solved well enough. That alone is a big part of why it became part of our process.

The normal maps are equally important, but for a different reason. They give us lighting reference, and in our workflow, that becomes incredibly valuable because it helps guide line work and painted lighting decisions later. It's not an exaggeration to say I don't think we could do this process the same way without those passes.

## Can you talk more about how you use the normal maps?

I tend to think of them as guide lighting. That's really what they are for me.

If I were painting shadows and highlights frame by frame entirely from imagination, the labour involved would be enormous. That's what whole careers in traditional animation are built around. I simply don't have the resources for that.

But with the normal maps, I have something truthful underneath to react to. I can see how light moves across a face or body frame to frame, and then I paint over that. I'm still making the decisions, but I'm doing so much faster and with much more confidence.

That's what makes it so useful. It doesn't replace artistic judgement — it accelerates it.

## Did it affect how you approached production on set?

Completely. It changed the shooting methodology.

Because we knew lighting decisions could be developed later, we shot with flat lighting on set, which gave us enormous flexibility. It meant faster scene changes, multi-camera setups, and fewer resets.

We were shooting eleven pages of script a day, which is kind of absurd. But a lot of that was possible because we weren't treating lighting as something that had to be fully resolved before the camera was set.

That really changed what scale of production felt possible for a small team.

## Was there a moment when it made a noticeable creative difference?

There's a scene with Thomas sitting by a window that comes to mind. It was a deceptively difficult shot because he's moving constantly, and the lighting is shifting in subtle ways.

What Beeble made easier was experimentation. I could audition lighting ideas quickly, move a source, change an edge light, alter colour relationships, and understand those decisions visually before committing.

That mattered less as a time saver and more as a creative tool. It made exploration easier.

And honestly, that's maybe the part I find most exciting.

## Do you see this becoming your standard workflow?

Yes, completely.

I think this is the pipeline now, and what excites me about that isn't just efficiency. It suggests another way filmmaking might work: smaller teams, people wearing more hats, more creative ownership.

I always compare it to music production. There was a time when making a record required enormous infrastructure, and now extraordinary work can be made much more independently.

I'd love to see filmmaking continue moving in that direction.

## Do you see this as part of a broader shift in animation or filmmaking?

Maybe. I try not to make huge predictions.

But I can imagine more people becoming interested in treating relighting in post as a serious creative stage rather than just corrective work. And I do think for rotoscope-based work, especially, we're only beginning to understand what tools like this can support.

[You can support the film here.](https://2000blue.com/DUST)

## Credits

Kai Tattersall — Director / Writer / Animator

Myles Ortiz-Green — Composer / Producer

Swati Shetty — Executive Producer

Mike Patterson — Creative Advisor


# Afterglow: Rebuilding light and atmosphere with AI
Source: https://beeble.ai/showcase/ai-relighting-music-video-rgby-beeble
Published: 2026-04-28

Peter Dalwoo Kim used Beeble for motivated relighting, background replacement, and atmospheric lighting on the cinematic Afterglow music video.

### How RGBY Entertainment's Peter Dalwoo Kim built *Afterglow* with motivated relighting, background replacement, and a lightweight production workflow.

For *Afterglow*, a music video for Red Chair, Peter Dalwoo Kim took on the entire production himself — from concept and shooting to editing and AI-driven post.

The result is a cinematic piece that blends nostalgia, soft emotional storytelling, and carefully controlled lighting, much of it shaped after the shoot.

Working through [RGBY Entertainment](https://rgby.kr/), both a creative studio and production company, Kim represents a new kind of creator: one who combines directing, music production, and technical workflow design into a single, streamlined process.

## Behind the scenes

### Can you explain your role on *Afterglow*?

The artist came to me with just the music and a wish to make a video. From there, I handled story development, concept planning, text-based storyboarding, shooting, editing, AI post-production, color grading, and location scouting.

### What was the core idea behind the video?

The group wanted their album to feel like the 90s. I came across some school uniform photos on their Instagram, and it clicked — two young women in familiar classroom spaces.

The theme was love, but filtered through innocence: a gentle jealousy and a close friendship. The bluebird represents the love they were chasing — it may fly away in the end — but what stays is the friendship between them.

### Where did your visual references come from?

My main reference was a previous music video I directed, [Milkyway](https://www.youtube.com/watch?v=t6gnGF4cuNY), inspired by the Japanese drama Hanako and Anne. I was drawn to the feeling of warm afternoon light and the quiet atmosphere of a rural town.

The school we used is actually the same filming location as that drama, which helped anchor the visual tone.

### How did you approach planning and previs?

I don't use traditional storyboards. I work entirely in text — I build everything in a spreadsheet. It works well for me as a solo creator because it keeps things structured without being too rigid.

## Beeble in production

### How did you approach shooting, knowing much of the look would be shaped in post?

Most shot decisions happen on location. Being there opens up ideas I can't plan in advance.

I used to pre-plan all AI replacements before shooting, but when things didn't go as expected, it was hard to adapt. Now I capture first, then develop ideas from the footage itself.

### The bird is a key element — how did you make it feel consistent?

AI-generated animals tend to change with every generation, so consistency is a challenge.

My solution was to give the bird a scarf. That detail makes viewers recognize it as the same character across scenes, and it adds personality.

### Where did Beeble fit into your workflow?

I use Beeble in three stages.

First, I establish the overall cinematic direction. Things like 16mm film, golden hour, or volumetric light — using a single test frame.

Second, I use Beeble for lighting adjustments without changing the background. This is especially useful for maintaining consistent, natural-looking light across different shots.

Third, for more complex shots, I use it for full background replacement, especially when there's heavy camera movement. That's where it's most powerful.

## Solving for light

### What were the biggest challenges?

Volumetric light was one of the hardest things to get right. Traditional tools often look unnatural.

I found that if I create a single reference frame with volumetric light already applied, Beeble can propagate that light across the entire clip in a very natural way. Getting that first frame right is key.

### How did you maintain consistency across different locations?

I created a custom LUT and applied it across the entire edit, then fine-tuned individual shots.

Beeble also helped by producing a naturally consistent tonal output, which made it easier to unify everything.

## Rethinking the workflow

### Did this project change how you think about production?

Completely.

Before, lighting alone required heavy equipment, power setups, and rentals. It was difficult to manage as a solo creator.

Now my shoots are very lightweight, and all the time I used to spend on logistics goes into the creative work instead.

### What did this workflow enable that wasn't possible before?

I can translate imagination directly into the frame.

Music videos require a lot of different environments, and preparing those used to take a lot of time and money. Now that overhead is essentially gone. As long as I know what I want, any environment is achievable.


# AI relighting in post: Rebuilding the look of a TV commercial
Source: https://beeble.ai/showcase/ai-relighting-in-post-for-a-tv-commercial-judas-studio
Published: 2026-04-20

See how Judas Studio used Beeble's AI relighting to refine lighting in a Napralief TV commercial, enabling faster turnaround and consistent results.

### How AI relighting helped Judas Studio finish a broadcast-ready Napralief commercial under tight timelines.

For this Napralief commercial, the production team at [Judas Studio](http://judasstudio.com/) faced a familiar challenge: footage that looked good in-camera but revealed lighting variations and color spill once in post. Traditional grading alone would have been limiting under tight timelines.

To bring the spot to the finish line, the team turned to AI relighting, using Beeble to dynamically adjust lighting on the talent and correct mismatches across multiple environments.

## The challenge

During production, the team encountered:

- Noticeable color contamination, including blue cast on skin tones.
- Inconsistent lighting across changing environments.
- Green screen setups that called for careful keying and compositing.

The result: footage that required additional work to integrate seamlessly across scenes. Traditional color grading in tools like DaVinci Resolve alone would have been insufficient to fully resolve these issues, particularly across multiple environment transitions.

## The solution: AI relighting

Using Beeble, the team applied AI-driven relighting directly within their Nuke compositing workflow. This allowed them to:

- Dynamically re-light the subject to match each environment.
- Neutralize color spill and restore natural skin tones.
- Maintain consistency across multiple scene transitions.
- Iterate quickly without manually rebuilding lighting setups.

The relighting pass worked alongside existing keys and mattes — enhancing, not replacing, the traditional pipeline.

**Judas Studio Creative Director CBevil** reflects on the impact of the tool:

> "Beeble was used to dynamically re-light the subject as the environments changed… it was a lifesaver in post."

These frames demonstrate how AI relighting helped correct color balance and lighting direction while preserving natural skin-tone detail.

## A shift in commercial production

This ad highlights a key shift in commercial production: AI is extending what's possible when production conditions are less than ideal.

For brands and agencies, that means:

- Reduced production risk.
- Greater flexibility after the shoot.
- Faster delivery timelines.

If you're working on commercial, film, or branded content and want greater control over lighting in post, now's the time to bring that control into your own workflow.


# Day-for-night filmmaking with Beeble SwitchX
Source: https://beeble.ai/showcase/alden-peters-vampire-night
Published: 2026-04-15

See how filmmaker Alden Peters used Beeble SwitchX to relight a daylight vampire scene into a convincing night sequence using real reference footage.

### Filmmaker Alden Peters set himself a simple but ambitious goal: shoot a dramatic vampire scene in broad daylight, then convincingly turn it into night using Beeble SwitchX.

Instead of relying on fully AI-generated reference images, Alden took a different approach. He went back to the same location after dark and captured real reference footage — same angles, same location, real lighting. That single decision dramatically improved shot consistency.

## The setup: day shoot, night reference

The production was intentionally simple:

- Film all principal photography in the shade during the day.
- Keep exposure wide open using ND filters to mimic low-light behaviour.
- Avoid direct sunlight on actors and the background.
- Return later at night to capture reference footage using real lighting setups.

Because the daytime footage was shot clean and controlled, the relighting step could focus on realism instead of correction.

## A smarter way to relight with SwitchX

Instead of generating a look with prompts, Alden used his own nighttime reference shots to drive the relight. This anchored the relighting process in real-world lighting direction and intensity rather than randomised AI outputs.

He explains: "If you're generating reference images, you're rolling the dice every time. Using real footage gives you consistent lighting logic across the whole scene."

This method meant:

- Consistent light direction across multiple angles.
- More natural skin response and specular highlights.
- No random lighting shifts between cuts.
- Dramatically reduced trial-and-error.

The result: consistent, believable night lighting across an entire sequence.

## Where it shines

The relight held up across wide shots, close-ups, movement and rack focus — with only minor challenges on one extreme close-up, which Alden solved by generating a cleaner unlit pass through Beeble Studio before running it back through SwitchX.

He also noted:

- 4K inputs produce visibly cleaner results.
- Minor flicker can be cleaned up easily in post.
- Real reference footage consistently outperforms generated stills.

## Why this matters for real productions

Night shoots are expensive, slow, and physically demanding for cast and crew. By shooting during the day and relighting in post:

- Productions can avoid costly night shoots.
- Schedules become more flexible.
- More takes can be captured in daylight.
- Directors retain creative lighting control in post.

Alden sums it up: "If you're shooting a longer project with lots of night exteriors, this approach can save time, energy, and money — without sacrificing visual quality."


# Relight explosive scenes directly inside Nuke with Beeble
Source: https://beeble.ai/showcase/nuke-compositing-academy
Published: 2026-01-16

Alex Hanneman from Compositing Academy shows how Beeble's Nuke plugin makes real footage relightable.

### Alex Hanneman from Compositing Academy shows how Beeble's Nuke plugin makes real footage relightable.

A flat greenscreen plate is usually a one-way door: you can key it, grade it, add CG, but you can't re-light the performance as if it were a CG asset. However, Alex does exactly that in his tutorial.

## The setup

Alex's workflow is deliberately practical: capture a simple greenscreen plate, build the CG scene in Blender, then use Beeble Studio to generate physically based passes from the live-action footage, so Nuke can treat the plate like a relightable element.

That means you're no longer locked into whatever lighting you managed on the day, which is especially useful when:

- You didn't have DMX/interactive lighting on set.
- You're still deciding where the CG event (explosion, muzzle flash, lightning) will land.
- You need to push a performance into a more dramatic, stylised environment.

## The process step-by-step

### 1) Generate PBR + utility passes from the plate

Using Beeble Studio with local processing, Alex generates a full set of passes, including normals, which are key to relighting, plus depth and material maps.

In the breakdown, he calls out passes like:

- Albedo
- Depth
- Normals
- Roughness, specular, metallic
- A rough roto/matte

He also notes an interesting side effect: the albedo result can even out uneven greenscreen lighting, making downstream keying and cleanup easier.

### 2) Relight directly inside Nuke with the Beeble Nuke plugin

Alex then brings those passes into Nuke via the [Beeble Nuke Plugin](https://app.beeble.ai/plugins#nuke), which packages the passes into a workflow that plugs into relighting nodes—environment/HDRI, directional, point lights—with controls to rebalance diffuse vs spec.

This is the core win: you can now do the classic CG relighting trick, but on real footage.

### 3) The explosion reflection trick, and why it looks so convincing

Instead of faking the explosion light with a simple grade, Alex drives a reflection-style projection using a free Nuke gizmo called [ReflectionBuddy](https://www.nukepedia.com/tools/gizmos/3d/reflectionbuddy/), feeding it the normals Beeble generated.

That lets him project the explosion footage itself so it:

- Hits at the exact right time.
- Shifts correctly across the surface as the normals change.
- Targets different areas based on the material.

It's also where good comp judgement matters. Alex flags the common plastic look pitfall and fixes it the right way: breaking up highlights, targeting materials separately, and doing proper, localised corrections rather than a single global glow.


# Red Mercury's I Still Remain: Zero-budget sci-fi, powered by Beeble
Source: https://beeble.ai/showcase/red-mercury-sci-fi-short
Published: 2026-01-15

Red Mercury shows how Beeble unlocks cinematic sci-fi visuals for small teams with AI rotoscoping and relighting.

### In I Still Remain, Red Mercury shows how Beeble can unlock big, cinematic sci-fi visuals for small teams, taking care of the heavy lifting on rotoscoping, relighting, and getting live-action performances into Unreal Engine.

When London-based electronic duo [Red Mercury](https://redmercurymusic.com/) started work on their I Still Remain music video, they didn't have a studio, a VFX team, or a budget – just a DSLR, a £20 blue sheet, and a one-bedroom flat.

Nine months later, they'd crafted a cinematic sci-fi short about cloning, identity, and survival in a hostile city, filled with Unreal Engine environments, gunfire lighting, swarms of creatures, and a grounded central performance. The key was using AI not to generate content, but to handle the behind-the-scenes VFX work that usually eats up time and resources.

The key was a streamlined, flexible pipeline. Performances were shot against a blue sheet, processed through Beeble for rotoscoping and relighting, and composited into Unreal Engine environments. What began as a test to see if one person could pull off cinematic VFX from home became a nine-month production – and a proof of concept for a faster iteration. That shift turned Beeble into the backbone of the production, handling rotoscoping and video relighting so a solo creator could work at a level normally reserved for full post houses.

Producer Steve Gregory and vocalist and songwriter Isobel Crawford shared with us how the project came to fruition.

## Behind the scenes

**Isobel Crawford:** We're an electronic band. We specialise in techno and drum and bass. Steve and I met in 2020, which is when we started Red Mercury together, and now we're husband and wife. I have a musical background and bring vocals and songwriting, combined with Steve's production.

During lockdown, when we couldn't film outside, we started exploring virtual production options, and over the last three or four years, our videos have been more focused on VFX and virtual production.

## The project

**Isobel:** When we start a project, we discuss what the theme might be, the message. For this one, we were looking at ideas like identity, cloning, not losing yourself in a crowd – the kind of concept you can explore through sci-fi. We wanted to make something that felt cinematic, not just a performance piece. That evolved into a storyline about a clone navigating through a surreal, dystopian cityscape, facing resistance at every turn. We shot it at that time. The video took nine months to make, working on it every night and on weekends.

The first three months were spent trying lots of different tools to figure out what was even possible. I must have tested 30 different pieces of software, including building my own, to build a pipeline that reliably generated shots.

We did three or four separate shoots, and it was literally us in a one-bedroom flat with wide lenses because I couldn't stand back far enough. And I even fit a 12-foot crane into the flat.

**Steve:** Beeble actually enabled us to do this project. I don't think we would have been able to produce it otherwise. I didn't want to commit unless I knew it was technically possible; I was really concerned it would look stuck-on and cheap, because that's often what happens when you try to integrate a real person into CGI: the lighting's wrong, the edges around it don't look right.

Rotoscoping is amazingly complex and has been largely manual for years. And relighting is also almost impossible to do well when you shoot under a single lighting source and have to match it to another; it's really hard.

For me, Beeble became a huge time-saver and an essential tool for combining real performances with CG environments. We shot everything in our flat, sometimes with an iPhone mounted above the camera for tracking reference. The footage then went straight into Beeble's web editor, which handled the rotoscoping and keying, and let us relight Izi to sit naturally inside the Unreal environments – from overall mood and direction to interactive details like gunfire flashes.

We brought those Beeble passes into Unreal Engine to build and animate the wider world, occasionally sending renders back through Beeble for final lighting tweaks.

As we grew more confident in Beeble's ability to remove backgrounds and relight footage, we even began dropping the blue screen for some setups, focusing on performance and camera movement, knowing we could refine the lighting in post.

There were three kinds of shots: simple shots with no movement, more complex shots in which the camera moves around the subject, and faked moving shots. The moving shots are more difficult because if the background doesn't track, you get sliding, and it doesn't look real.

Beeble didn't support camera tracking at the time, so finding a camera-tracking solution that would integrate was a challenge. I tried Lightcraft, and in the end, I used Wonder Dynamics and simplified shots.

At a high level, the pipeline was the following: shoot on bluescreen, take into Beeble and rotoscope to a transparent background; then either relight and export into Unreal, or what I ended up doing was rendering stuff in Unreal and combining it with the Beeble footage, because I found the Beeble web editor produced much better results than taking it into Unreal.

**Steve:** Shots on the bridge: gunfire going off in Unreal, lights flashing, and the footage being relit with those gun flashes. There's no way I could have done that with anything else.

Also, the flying bees: originally, I was going to animate them in 3D, but time-wise, it was too much. I used Kling to generate some animation, then used Beeble to remove the background and relight, and I flew the video through it. It turned what would have been a complex 3D shot into something I could produce very quickly, and it looked perfect.

**Steve:** Get started. I Still Remain was made with no budget, a £20 blue sheet, a DSLR camera, and a lot of evenings and weekends. The barrier to entry has never been lower – the main cost is your commitment.

The hardest part isn't gear, it's finding software you can trust. For us, Beeble became one of those tools: a huge time-saver and a crucial component in transforming our footage into something that looks far bigger than the room it was shot in.


# Relighting Sounds of a Sphere with Jonathan Hislop
Source: https://beeble.ai/showcase/sounds-of-sphere
Published: 2025-11-30

How VFX artist Jonathan Hislop uses Beeble to reshape online noise into calm, immersive worlds.

### How a VFX artist reshapes online noise into calm, immersive worlds using Beeble.

[Sounds of a Sphere](https://www.youtube.com/@soundsofasphere) invites viewers to relax, using a blend of native music and stunning backdrops. But creating such videos requires extensive planning, shooting, and post-production work. That's where Beeble steps in, taking care of the heavy lifting when it comes to video relighting.

VFX artist Jonathan Hislop discovered his passion for CG visuals when working on 'drum and bass' flyers twenty years ago. Since then, he has mastered Photoshop, matte painting, and 3D, and worked with notable clients such as Atlantic Records UK, Warner Music Australia, and Flux Pavilion. Jonathan's latest project aims to defuse the fast-paced nature of the internet, with video relighting powered by Beeble.

## How did the idea of Sounds of a Sphere come about?

Sounds of a Sphere was born from a desire to counterbalance the chaotic nature of YouTube. Instead of adding to the noise, I wanted to create something calming and meaningful – a space where music and visuals come together to tell cultural stories. Every episode centers around a different artist's homeland, designed to immerse viewers in their world. The Bali episode is a great example of this. I collaborated with [Balinese musician Gus Teja](https://www.youtube.com/watch?v=-KKAy66EJEI) and ensured the backdrop matched the country's artistic history. I turned the Legong dancer into a ghost, reflecting Balinese spiritual beliefs, and added a deer, an animal that commonly represents a connection to the spirit world.

## What is the workflow, from planning to post?

I start by listening to cultural music from around the world and seeing what would fit Sounds of a Sphere best – relaxing and heartfelt melodies are best. Then I'll reach out to the musician and organize travel. Once I'm there, I'll rent some lights, a green screen backdrop, and a place to record. Sometimes, that involves traveling cross-country to find the most budget-friendly options. Then, I record the artist's performance in front of the green screen.

Once I'm back home, I master the music, edit the video, and start building the artist's environment – extracting the green screen in After Effects, creating the world in Unreal Engine, and sculpting in Zbrush. Sometimes, I use photogrammetry to scan statues or other assets into the background.

The next stage is lighting. It's always dynamic, mystical, and cements the artist in realism. Each new piece demands hours of intricate light adjustments to get that otherworldly look I need. That's when I discovered Beeble, a tool that lets me relight video efficiently without losing creative control.

## What drew you to explore Beeble for video relighting?

A friend recommended it to me after I shared some underwhelming experiences with AI tools in the past. When I downloaded Beeble, I instantly had more control over the output than other tools on the market.

I only needed a couple of minutes to get used to Beeble. The point light system worked instantly – it's an out-of-the-box solution. I could also add controls to the normal maps to make the lighting look more or less pronounced. The video relighting methods I used previously took hours, sometimes even days. This was much, much faster.

## Are you using Beeble in future projects?

Yes, it's cemented into the workflow now. In the past, relighting was a frustrating experience, but now it's so simple and I actually look forward to it! [SwitchLight 3.0](../research/switchlight-3-0-is-here) has also made a huge difference to the video relighting quality.

## What has the reception been like so far?

It's been great, and Sounds of a Sphere has since traveled outside of YouTube. I've helped create a 'chill out lounge' for two Diwali festivals here in New Zealand, where my videos are projected on a ten-meter-wide screen, and I'm planning to showcase my work at other festivals soon. The Maori living painting played as the opening for the Maoriland Film Festival and was shared with 60 embassies worldwide. It's great to have a project so close to home that's appreciated by so many.


# VFX artist Suryam Singh on creating realistic fire with Beeble
Source: https://beeble.ai/showcase/suryam-vfx-breakdown
Published: 2025-10-26

Suryam Singh combines traditional compositing with AI-driven CG lighting to create cinematic fire effects.

### How CG lighting to bring fire-filled drama to Aksomaniac's music video.

VFX artist Suryam Singh has built a career through self-teaching, starting out by editing memes and progressing to high-profile music videos and commercials for clients such as Hyundai, Microsoft, and Google. Today, his expertise spans motion graphics through to advanced post-production, where CG lighting plays a central role in his work.

For musician Aksomaniac's latest single, Paapam, Singh was tasked with delivering a dramatic, fire-filled finale that mirrored the track's themes of shame and release. With only nine days to final delivery, he combined traditional compositing with AI-driven CG lighting to create cinematic realism under intense deadlines.

## How It All Started

### How did you get involved with Paapam?

A friend of mine introduced me to the Director, Madhavan Krishneswaran. The brief was clear: the final scenes needed to show large-scale fires and flames, with enough drama and visual weight to support the song's emotional payoff. Creating realism required up to 50 layers of fire, ember, and smoke elements, carefully composited to achieve the sense of an environment consumed by flames.

### Can you break down the shot shown on your Instagram account?

The base compositing was done in After Effects. I pulled together fire, ember, and smoke assets, layering them to build depth and drama. The goal was to make the fire feel like part of the environment rather than an effect added on top. Once the composites were in place, I turned to Beeble to refine the CG lighting and integrate the flames with the rest of the scene.

## Bringing Fire to Life with Beeble

### How did Beeble help make the fire effects more realistic?

Matching the atmosphere to the fire additions was essential. Beeble's depth map output gave me the precision I needed to relight each shot. I extracted highlights and applied them back onto the footage so the flames interacted naturally with the characters and environment. I also added fog and haze to reflect the impact of smoke in the scene. These details are what make the fire feel convincing rather than stylised.

### What drew you to explore AI tools like Beeble in this workflow?

As VFX artists, it's our responsibility to stay at the cutting edge of technology. AI is becoming part of our workflows, but many tools still rely on broad text-based prompts that don't give you much control. Beeble stood out because it lets me adjust placement, intensity, and hue with precision. That kind of control is critical when you're working with something as dynamic as fire.

### Will you be using it in future projects?

Definitely. Blender is already central to my workflow, and experimenting with the Beeble add-on is high on my list. The creative freedom it gives me in lighting has already proven valuable, and I expect it will become a bigger part of how I approach upcoming projects. There's something else I want to try it for: photogrammetry. I should be able to take a 3D scan of an object, remove all light and shadow in Beeble, and use it in various projects without the multi-camera setup.

## The Future of VFX

### What are your predictions on emerging technology in VFX?

Real-time applications will become the standard across the industry, and virtual production will transition from a niche to a mainstream practice. AI is here to stay, too, but for it to succeed, the tools must remain customisable and accessible. At the end of the day, artists need to keep creative control; the technology should enable, not dictate.


# Bad Beetle Entertainment uses Beeble for AI in filmmaking
Source: https://beeble.ai/showcase/bad-beetle-entrenched-filmmaking
Published: 2025-09-24

Bad Beetle Entertainment uses Beeble to deliver cinema-grade VFX for their Star Wars fan film.

### Bad Beetle Entertainment uses Beeble for its fan film teaser trailer.

With just four people behind the scenes, [Bad Beetle Entertainment](https://badbeetleentertainment.com/)'s Star Wars: Entrenched is already earning praise from major studios and directors. The trailer alone demonstrates that high-end production values are no longer limited to big teams or big budgets.

Thanks to accessible, AI-powered tools like Beeble, independent creators can take full creative control—relighting scenes, customising looks, and delivering professional-quality visuals, all without traditional gear or complex software pipelines.

In this interview, studio head Navaz Dowling shares how he and his son Roman launched Bad Beetle Entertainment, the creative process behind Entrenched, and how Beeble enabled them to achieve a cinematic finish on their own terms.

## The Origin Story

### How did Bad Beetle Entertainment come about?

I've spent my whole career helping teams build and fix products — I'm a project and engineering manager by trade. But my passion has always been storytelling, design, and film. About a year and a half ago, my son Roman, who's 17 and has real talent for VFX, visual design, and cinematography, told me he wanted to get into the industry. That's when I suggested we start our own entertainment company. We started making things together, learning while teaching ourselves new techniques.

### Tell us more about Star Wars Entrenched.

It's a fan film that retells Empire Strikes Back from a different perspective. It's based on a comic book of the same name. As soon as I read it, I knew it would work amazingly in an audio-visual format, so we got to work deciding on the creative direction and our methodology.

This is our second project as Bad Beetle Entertainment, and although we've only released the teaser so far, we're receiving great feedback already. We've been invited to visit Pixar, and the creator of the comic has also given us kind words about our progress and vision.

## Production & Process

### How did you make a high-quality production on such a small scale?

We used what we had to the best of our ability. We were four people doing the job of 50, including concepting, shooting, editing, and post-production. Alongside myself and Roman, we had Lucas, who's training for VFX, and Zoe, who's creating the costumes for each character by hand. That's what makes this so special: we're all bringing our own unique skillsets to the project.

In terms of VFX, Roman taught himself a great deal about Blender and DaVinci Resolve, and we refined our work until it became as lifelike as possible. However, the real challenge was relighting, as enterprise-grade tools seemed to be our only option, until we discovered Beeble.

### How did you find out about Beeble?

I was watching YouTuber and VFX artist [Joshua M Kerr,](https://www.youtube.com/c/JoshuaMKerr) who was walking through Beeble's features and showcasing its quality. He demonstrated how easy it was to use from home, and that suited our needs perfectly, so we decided to give it a try.

We got the hang of the software in just a day or two. The help received from the Beeble team during that time was crucial to our success, and they were quick to answer any questions we had.

## Beeble's Impact

### How did Beeble enhance the filmmaking process in Star Wars Entrenched?

Relighting can be one of the most difficult and time-consuming aspects of post-production, but Beeble made it easier for us. Even in traditional, large-scale pipelines, it's challenging to change the lighting once the scene has been shot. But Beeble enables us to completely remove light and shadow and add our own, customisable effects. That's an edge it gives indie filmmakers that the rest of the industry doesn't currently have.

With reshoots now complete, we're entering the final stages of production, where Beeble will really shine. Thanks to our constant upskilling and Beeble's professional finish, we're moving past just showcasing our work. We're now in competition with major film studios in terms of quality.

### SwitchLight 2.0 launched after the teaser trailer. Have you tried it yet?

Yes, we love it! It's highly improved from the original model, and truly brings it to a cinema-grade tool. It's clear the Beeble team has taken customer suggestions and implemented them in this new model. We're excited to see how far it can take us for Star Wars Entrenched.

Whether you're an indie creator or a professional VFX artist, Beeble gives you full control over lighting, no green screens, no 3D, just cinematic results.

Ready to try it yourself? Explore Beeble and start relighting your own scenes—no green screens or 3D required.


# Relight your shots in Blender with Beeble AI
Source: https://beeble.ai/showcase/thies-relight-blender-workflow
Published: 2025-09-24

Thies Grunewald demonstrates a production-ready workflow for relighting footage using Beeble and Blender.

### Thies Grunewald's Frankenstein workflow

What if you could turn a flat greenscreen plate into a fully relightable shot, complete with point lights, HDRIs, cast shadows, and depth-aware displacement, without the hassle of rebuilding on-set lighting or fighting roto?

With Beeble SwitchLight 2.0 and Blender, that's exactly what VFX artist Thies Grunewald demonstrates. In this three-part tutorial, he walks through a streamlined, production-ready workflow that transforms simple plates into cinematic, relightable elements.

- Prepare a clean plate (with a greenscreen or neutral background).
- Process VFX passes in Beeble (alpha, base colour, normals, roughness, specular, AO, depth, flow).
- Import them with the official Blender add-on, ready for instant relighting—just like working with full CG assets.

## Frankenstein Blender. A new greenscreen approach

### Watch as Thies transforms flat footage into a stunning shot.

Virtual production fascinated me ever since the first Mandalorian episodes, and SwitchLight 2.0 seemed like the perfect way to show people that they could tell bigger stories without needing massive resources.

## Building Frankenstein: Shot Breakdown

- **Crisp 2.5D elements.** SwitchLight 2.0 generates relightable passes with improved normals, resulting in more realistic outcomes.
- **Seamless Blender integration.** Upload footage to Beeble, auto-generate passes, then import directly into Blender with everything wired up.
- **Advanced relighting.** Cast shadows and use depth features to add realism, rim lights, and dramatic lighting shifts in post-production.
- **Plan with intent.** Gather strong visual references, storyboard with tools like Storyboarder, and preview lighting in Beeble's app before shooting.
- **Keep it practical.** Props, costumes, and simple shoots come alive once relit in post.
- **Polish in post.** Compositing, grading, and extra effects (lightning, rain, electricity) transform flat renders into cinematic shots.


# AI relighting for indie filmmakers: Interview with Joshua M Kerr
Source: https://beeble.ai/showcase/interview-with-joshuakerr
Published: 2025-09-19

Joshua M Kerr discusses how Beeble's SwitchLight 2.0 transforms indie filmmaking with AI relighting.

### Joshua M Kerr Transforms Indie Films with Beeble's SwitchLight 2.0

Filmmaker and YouTuber [Joshua M Kerr](https://www.youtube.com/@JoshuaMKerr) began experimenting with film at just nine years old, making Wallace and Gromit-style clay animations in his living room back in the '90s. That early passion never left him.

What drives Kerr today is the same instinct that motivated him then: the desire to make things easily and freely from home, without needing big budgets or large crews. Through his YouTube channel, he now shares that approach with others, teaching creators how to produce impressive, cinematic results in their own spaces, without breaking the bank.

In this interview, Kerr discusses his journey into virtual production, his discovery of Beeble, and how a [24-hour challenge using SwitchLight 2.0](https://www.youtube.com/watch?v=y9CnSF8FgKo) demonstrated just how far AI relighting has come.

## Background

### What sparked your interest in virtual production and filmmaking?

I've been making things since I was nine, starting with stop-motion clay animations in my living room. Eventually, I went on to study TV and broadcast production at university. A few years ago, I discovered virtual production — the idea of using tools like Unreal Engine to create environments and integrate real-world performances. I started sharing that process on YouTube. The more I explored, the more my audience responded. It felt like the perfect way to show people that they could tell bigger stories without needing massive resources.

### Your YouTube content has attracted a wide audience. Who are you creating for?

It's a broad mix. There are complete beginners who have never touched Unreal or virtual production, as well as industry professionals who watch along. Even if you're on a big LED wall shoot, you don't necessarily have access to that tech for your own projects.

Many of us want to create things ourselves and tell our own stories. My channel is about learning as I go and inviting people to follow the process, including the mistakes.

## Choosing Tools

### How do you decide which tools to feature in your videos?

I'm very picky. A tool must solve a real problem in the storytelling process and be accessible. A lot of virtual production gear is expensive, so there's no point covering something my audience can't use.

### How did you approach finding tools that could actually work for your process?

I'd been searching for ways to do relighting for more than a year before Beeble even existed. I tested other tools, but they weren't good enough to use seriously.

Then someone mentioned Beeble on Discord. I gave it a try and immediately realised the quality was way ahead of anything else. At the time, it only produced two kinds of passes — one to even out the lighting and one to give a 3D normal map — but it worked in a way nothing else did.

## Evolving Workflows

### What improvements in relighting workflows have you noticed as the tools have developed?

It's evolved massively. Early on, integration into Unreal was completely manual — hours of work. Now it's almost one click.

The quality of the maps is also much higher, especially the normals. And instead of two passes, you've got six different types, which makes the lighting feel much more realistic. It has evolved from a proof-of-concept to a production-ready solution in just a couple of years. It's been a nice thing to have been on that journey with them as they progress and improve.

### You recently set yourself a 24-hour challenge — what did that experiment show you about working faster with these tools?

That came from a dare, really. Hoon Kim, Beeble's CEO, suggested I try making something in 24 hours. I don't think he expected me actually to do it — but I came back the next day with a short that worked.

What would have taken me days now takes hours, and the results still look cinematic. That's the kind of breakthrough that makes independent creators sit up and think: okay, this is possible.

## Consistency Matters

### From your perspective, what separates a useful AI tool from one that isn't production-ready?

Most generative tools are chaotic — you can't control them, you can't iterate, and they don't fit into production pipelines. Generative can't do what Beeble does, because it will change the way your character looks. Other AI systems I've tested can perform relighting, but often alter the appearance of a character. That's not acceptable. Consistency matters. Beeble manages it, and my money's on them for getting there first.

Beeble generates something familiar, useful, and completely production-ready. Any VFX artist can look at it and say: yes, this makes sense. It fills a gap that would otherwise be a very manual and difficult job.

### Looking ahead, how do you see relighting fitting into your future projects and experiments?

I like to run small individual tests — relighting, green screen, different workflows — and then pull everything together into a bigger project. Beeble is going to be a central part of that.

I've also seen some of what Hoon is working on next, and I can't wait for it to be released.

## Conclusion

For Joshua M Kerr, AI relighting isn't just about saving time; it's about unlocking creative freedom. By removing the need for complex dual-lighting setups and speeding up iteration, Beeble's SwitchLight makes it possible for creators to achieve cinematic results from anywhere, on any budget.

Whether you're experimenting at home or pushing the boundaries of professional VFX, tools like Beeble are proving that relighting can be faster, more consistent, and accessible to everyone. Try it yourself in [3D Relight](https://beeble.ai/features/3d-relight) — AI relighting in a real-time browser 3D scene.


# Superman & Lois: Pushing limits with AI relighting
Source: https://beeble.ai/showcase/superman-lois-relighting-vfx
Published: 2025-08-19

See how Boxel Studio used Beeble's AI relighting tools to transform Superman & Lois VFX shots.

### Boxel Studio used Beeble's SwitchLight to deliver faster, high-quality VFX.

Since its establishment in 2009, [Boxel Studio](https://boxelstudio.com/beeble-switchlight/) has worked on major productions from the likes of Sony, Lionsgate, and Disney. When Founder Andres Reyes Botello and Creative Director for AI & Innovation Freddy Chavez Olmos started work on HBO Max's Superman & Lois, they needed lighting software that would speed up delivery time while retaining high quality, as creating VFX for a television show requires meticulous planning and fast turnarounds.

## Behind the Scenes

### Can you explain a little bit about your work on Superman & Lois?

**Andres:** We were across a lot of different shots. Some involved full CG environments, creature animation, and digital double creation, and others had more subtle effects added, like destruction, scenery augmentation, and compositing. Luckily, we worked very closely with the on-set VFX supervisor, so we had good reference material and on-set data, including LiDAR scans for accurate geometry and HDRIs for lighting reference.

**Freddy:** When I first saw SwitchLight in action, I knew there was a use for it. A colleague, Bruna Berford, ran the first test on real Superman & Lois footage. When she showed me the result — normal maps, depth, and other passes extracted from a live-action plate — I was floored. Our comp supervisor realized Beeble was essentially performing footage virtualization, generating different AOV passes straight from the live-action plate. And she did all of this within Nuke without CG and digi-doubles, so we knew it would completely shake up our workflow.

We showed the results to the VFX Producer, Matt Gore, on Superman & Lois, and he was very impressed. The team is always looking for ways to boost production value while staying mindful of financial constraints, and Beeble could do just that.

### How did you discover Beeble, and what made you decide to try it?

**Freddy:** I'm always on the lookout for technologies that can reshape the way we work — things that aren't just incremental improvements, but real paradigm shifts. I first learned about Beeble through reading a research paper and was immediately intrigued by what it could mean for VFX professionals. I reached out to Hoon Kim, Beeble's CEO, met him at SIGGRAPH that year, and our working relationship began. Since then, we've been integrating Beeble into several of our projects. Initially, we used it for smaller parts like green screen comps, then we saw how fast and flexible relighting made a difference.

**Andres:** Once we understood its potential, we implemented it into our main project pipeline. It was one of the first machine learning-assisted tools we ever used in a real production setting. Those early tests on green screens worked great, and since then, it's become our go-to solution for a much broader range of compositing work.

## Beeble in Production

### What kind of work did you do with Beeble?

The software enabled us to create truly intricate, interactive lighting, no matter how small the detail. For example, we relit the edges of Superman's eyes for his laser vision, simulated subsurface glows, and made characters react to nearby explosions. In a traditional pipeline, FX like this would be more difficult to create and would involve the appropriate lighting on-set. But using Beeble, we didn't need any complex 3D setups.

The artists we worked with found Beeble to be a game-changer. Our compositing team had direct creative control over the lighting, something relatively unheard of before, especially without going back to rendering or requiring additional passes. It really helped to tighten the feedback loop and enabled the team to iterate faster and more freely.

### What challenges did Beeble solve during post-production?

We handled many shots in a short amount of time during Superman & Lois, so deadlines were a big factor in our work. We mean hundreds of shots to be delivered in under three weeks, so Beeble's ability to generate AOVs almost instantly was absolutely crucial. It completely removed the need for rotomation and geo-tracking, saving us a lot of time and labour while retaining the quality.

It also helped us in a creative sense, not just operationally. Because we had shorter render cycles, we could use lighting in unique ways to drive the narrative forward, which helped us convey the emotional tone and polish many of the shots we delivered.

### Superman & Lois was released before the SwitchLight 2.0 launch. Have you tried it yet?

Yes! We've been actively integrating SwitchLight 2.0 into our pipeline. It's even more artist-friendly, and the AOV generation and UI responsiveness have seen massive improvements.

What's been truly valuable is the support from Hoon Kim and his team at Beeble. They've been listening to our feedback and implementing the suggestions they find useful. That level of responsiveness has not only made the tool better, but has made using it a great experience.

---

Learn more about [SwitchLight 2.0](../research/switchlight-2-0-is-here), or relight your own footage in the browser with [3D Relight](https://beeble.ai/features/3d-relight).

Courtesy of [Boxel Studio](https://boxelstudio.com/beeble-switchlight/)


# Remove.bg for Video: What to Use Instead (2026)
Source: https://beeble.ai/compare/remove-bg-for-video
Published: 2026-08-14

remove.bg stops at images. Its video answer is Canva, which exports MP4 with no alpha channel. Beeble Background Remover gives you real alpha at up to 4K.

remove.bg is the default name in background removal, and for images it deserves it. Upload a photo, download a clean cutout, done. Video is a different story: remove.bg has never processed it. The official video page sends you to Canva's editor, and that route ends in an MP4 with no alpha channel. If you searched for "remove.bg for video", what you actually want is Beeble [Background Remover](/features/background-remover).

Nothing here argues against remove.bg for single images. It stays good at that. The problem is video: remove.bg never handled it, Unscreen (its video sibling) shut down in December 2025, and the Canva route it now recommends exports MP4, which can't carry transparency. **Beeble Background Remover** fills exactly that gap: automated background removal for video, image sequences, and images, exported as grayscale mattes, transparent PNG sequences, and an alpha-embedded 10-bit ProRes MOV at up to 4K.

## remove.bg stops at images

remove.bg processes still images, up to 50 megapixels, in seconds. That is the whole product, and it's a good one. Video lived in a separate product: Unscreen, built by the same company (Kaleido) and acquired by Canva along with remove.bg in 2021.

[Unscreen shut down on December 1, 2025](/compare/unscreen-alternative). Since then the remove.bg side has had no video path of its own. The video background removal page on remove.bg today simply hands you to Canva.

## The Canva detour, and where it dead-ends

Canva's video background remover does work, as long as you stay in Canva. Upload a clip, the AI isolates the subject, and you can put new graphics or colors behind it in the editor. For a social post that starts and ends in Canva, that's fine.

The problem is the export. Canva delivers video as MP4, and MP4 can't carry an alpha channel. The transparency only exists inside the Canva canvas. The file you download is flattened onto whatever sat behind the subject: no matte, no transparent frames, nothing you can bring into Premiere Pro, After Effects, DaVinci Resolve, or Nuke. It removes a background in an editor. It doesn't give you an alpha you can take somewhere else.

## Beeble Background Remover

Beeble [Background Remover](/features/background-remover) was built for video from the start. Drop in a video, an image sequence, or a single image, and cloud GPUs track your subject and pull the matte across every frame. No green screen required.

### Real alpha output

You get the alpha as a grayscale matte (MOV or PNG sequence), the RGBA version as transparent PNGs, and an alpha-embedded 10-bit ProRes 4444 MOV. All of it drops straight into After Effects, DaVinci Resolve, Nuke, or Premiere Pro with no re-encoding. Runs go up to 4K and 60fps.

### Two models, including a real keyer

MatAnyone, the default, pulls a clean foreground matte from ordinary footage with no controlled backdrop. If you shot on green, switch to CorridorKey, a purpose-built green-screen keyer with spill removal. Neither remove.bg nor the Canva route has an equivalent.

### Sequence-aware, not per-frame

Video is processed as a sequence, so edges stay consistent across frames instead of shimmering the way per-frame cutouts do. The result holds up in a timeline without frame-by-frame cleanup.

### Control before the run

Before the full extraction you can review a preview keyframe and refine the matte in Edit Alpha, masking exactly what to keep. Or hit Skip Review and let everything run automatically from frame one.

## remove.bg vs Beeble at a glance

| Feature | remove.bg | Beeble |
| --- | --- | --- |
| Image background removal | Yes | Yes |
| Video background removal (remove.bg's own video page routes you to Canva's editor; the remove.bg model never touches video) | No | Yes |
| Transparent (alpha) video output (the Canva route exports MP4, which cannot carry an alpha channel; Beeble exports transparent PNG sequences and a 10-bit ProRes 4444 MOV) | No | Yes |
| Grayscale matte export | No | Yes |
| Green-screen keying mode (CorridorKey, purpose-built for chroma key footage with spill removal) | No | Yes |
| Manual refinement (remove.bg offers an erase/restore brush on images; Beeble's Edit Alpha masks the subject before the full video run) | Partial | Yes |
| Max video spec (up to 3,600 frames per run on Professional, 1,800 on Creator) | No video | **4K · 60fps** |
| Free tier (remove.bg previews are free at low resolution; Beeble Background Remover ships with the Creator and Professional plans) | Yes | No |

## Choose remove.bg if

- Your work is **single still images** (product photos, headshots, thumbnails) and per-image speed and cost matter most.
- Free preview-resolution cutouts already cover your use case.

## Choose Beeble if

- Your source is **video or an image sequence**. remove.bg has no path for it.
- The output has to hold up in an **edit or composite**: transparent PNG sequences, grayscale mattes, and a 10-bit ProRes 4444 MOV instead of a flattened MP4.
- Your footage was **shot on green** and you want a keyer with spill removal, not a generic segmenter.
- You want **control over the matte**: refine it before the run instead of taking whatever comes out.

**Can remove.bg remove backgrounds from video?**

No. remove.bg processes still images only. Its official video background removal page points to Canva's editor, and its former video sibling, Unscreen, shut down on December 1, 2025.

**Does Canva's video background remover export transparent video?**

No. Canva removes the background inside its own editor but exports video as MP4, a format that cannot carry an alpha channel. The transparency exists only while you stay in Canva. For a file you can composite elsewhere, you need a tool that exports a matte or alpha-embedded footage, like Beeble Background Remover.

**What happened to Unscreen?**

Unscreen, the video background remover from the makers of remove.bg, was acquired by Canva in 2021 and shut down on December 1, 2025. Beeble Background Remover covers the same automated video workflow with cleaner mattes and lossless alpha export. See the full Unscreen comparison on this site.

**How much does Beeble Background Remover cost?**

MatAnyone costs 1 credit per image and 1 credit per 30 video frames; CorridorKey costs 2. Background Remover is available on the Creator and Professional plans, at up to 4K resolution and 60fps.


# AI Rotoscoping vs. Green Screen Keying: Which Do You Need?
Source: https://beeble.ai/compare/ai-rotoscoping-vs-green-screen-keying
Published: 2026-08-12

When to use AI rotoscoping and when to key green screen footage — and how MatAnyone 2 and CorridorKey handle each inside Beeble's Background Remover.

Every subject isolation job comes down to one question: did you control the background when you shot, or not? The answer decides which of two very different techniques you need — and which AI model should run it.

Beeble's [Background Remover](/features/background-remover) ships both. [MatAnyone 2](/models/matanyone) handles AI rotoscoping on ordinary footage; [CorridorKey](/models/corridor-key) handles green screen keying. Here's how to choose.

## What AI rotoscoping actually does

Rotoscoping is the craft of tracing a subject out of footage that was never shot for extraction — no green screen, no controlled backdrop. Done by hand, it's one of the slowest jobs in post. Done with a tool like After Effects' Roto Brush, it still means propagating a selection shot by shot and cleaning up where it drifts.

AI rotoscoping replaces that with a matting model. [MatAnyone 2](/models/matanyone), the CVPR 2026 Highlight release from NTU's S-Lab, pulls an alpha matte directly from ordinary footage and propagates memory across the shot, so the matte stays consistent through motion instead of flickering frame to frame. Wispy hair, motion blur, and semi-transparent edges survive the whole clip.

## What green screen keying does when it's done right

If you did shoot on green, keying should be the easy path — but classic chroma keyers mask pixels by color and hide spill, which is exactly where hair, glass, and motion blur fall apart.

[CorridorKey](/models/corridor-key), built by Corridor and licensed on Beeble, treats keying as an unmixing problem instead: it reconstructs the true, un-multiplied color of the subject as if the green screen had never influenced it. Transparency stays transparent, spill is removed rather than desaturated, and the output is a compositor-ready un-multiplied foreground.

## Side by side

| | AI rotoscoping (MatAnyone 2) | Green screen keying (CorridorKey) |
| --- | --- | --- |
| Footage | Any — no controlled backdrop | Shot on green screen |
| Approach | Video matting with memory propagation | Unmixing — true foreground reconstruction |
| Spill handling | Not applicable | Actively removed, not hidden |
| Credits | 1 per image / per 30 video frames | 2 per image / per 30 video frames |
| Best for | Location footage, archival, anything unplanned | Planned shoots where edge quality is everything |

## So which do you need?

- **No green screen?** [MatAnyone 2](/models/matanyone). It's the default model in Background Remover for a reason — it mattes footage that was never meant to be keyed.
- **Shot on green?** [CorridorKey](/models/corridor-key). A dedicated keyer beats a general matting model on chroma footage: cleaner edges, true colors, zero spill.
- **A mixed project?** Use both. The two models share one workspace, one pipeline, and the same exports — alpha mattes (MOV or PNG sequence) and 10-bit RGBA MOV that drop straight into After Effects, DaVinci Resolve, or Final Cut Pro.

## Coming from After Effects?

If you're replacing Roto Brush sessions or a plugin-based keyer, the workflow difference is the point: nothing to install, no local GPU, and full resolution up to 4K/60fps for as many as 3,600 frames per run on Professional. CorridorKey's After Effects and Resolve plugin still exists for local work — on Beeble, the same model runs in the cloud. For the full story of how it landed on Beeble, read the [CorridorKey launch post](/blog/corridorkey-beeble-green-screen-keying).


# Unscreen Alternative in 2026: Beeble Background Remover
Source: https://beeble.ai/compare/unscreen-alternative
Published: 2026-08-12

Unscreen shut down in 2025. Beeble Background Remover picks up where it left off — cleaner mattes, lossless RGBA export, manual control. Compared August 2026.

Unscreen shut down on December 1, 2025, after being acquired by Canva in 2021. If you relied on it for video background removal, you need a replacement. Beeble [Background Remover](/features/background-remover) picks up where Unscreen left off — and goes further. (Its still-image sibling, remove.bg, lives on but never picked up video; that gap gets its own breakdown in [remove.bg for video](/compare/remove-bg-for-video).)

This isn't a head-to-head between two live products — Unscreen is gone. The question is whether **Beeble Background Remover** actually fills the gap, and it does: the same upload-and-go, cloud-based video and image background removal, plus the production upgrades Unscreen never shipped — cleaner alpha mattes, 10-bit ProRes and 4K output, steadier edges across frames, and manual control when automatic detection misses. The short version: if your Unscreen output ever headed into an edit or a composite, Beeble is the upgrade, not just the replacement.

## What Unscreen did well

Unscreen earned its reputation for a reason. It was one of the only dedicated tools for video background removal. You uploaded a clip, it removed the background, and you got your result. No editing skills required, no expensive software licenses, no manual frame-by-frame work.

For creators, marketers, and small production teams, Unscreen made video keying accessible. It filled a real need that the big editing suites didn't address — quick, automated video background removal without the learning curve of After Effects or Nuke.

That simplicity mattered, and the gap it left is real.

## Where Unscreen fell short

Unscreen was built for convenience, not for production work. If you used it professionally, you probably ran into its limits:

- **Alpha quality** — edges were soft, with visible haloing around hair and fine details. Good enough for social media, not for compositing over complex backgrounds.
- **Lossy pro-path alpha** — the promoted route into an editor was the Pro Bundle: a separate compressed matte video you recombined with the footage yourself, and everything capped at 1080p. Workable for social cutdowns, lossy for compositing.
- **Limited controllability** — fully automatic with no way to guide the extraction when the default result wasn't right.
- **Temporal artifacts** — edges could shimmer and jitter between frames, especially around hair and fast motion.

These weren't dealbreakers for casual use. But for anyone trying to use the output in a real post-production pipeline, they added up.

## Beeble Background Remover

As of 2026, Beeble Background Remover does everything Unscreen did — cloud-based, automated video and image background removal — while addressing the limitations that kept Unscreen out of professional workflows. It runs two extraction models: MatAnyone pulls a clean matte from ordinary footage with no controlled backdrop, and CorridorKey is a purpose-built green-screen keyer with spill removal for footage that was shot on green — a choice Unscreen never gave you.

### Cleaner alpha mattes

Beeble generates alpha channels that preserve partial transparency at edges. Hair, fur, and semi-transparent materials come through with natural falloff instead of hard, haloed edges. The matte quality is the kind you can actually composite over detailed backgrounds without cleanup.

### PNG sequence export

This is the biggest practical upgrade. Beeble exports lossless PNG sequences — individual 8-bit RGBA frames with the alpha embedded — plus a grayscale matte sequence and an alpha-embedded 10-bit ProRes 4444 MOV when your pipeline prefers a single file. Unscreen could hand you PNG frames too, but its alpha route for editors was the Pro Bundle's compressed matte video, capped at 1080p. Beeble's frames drop straight into After Effects, Nuke, DaVinci Resolve, or any compositing tool — no re-encoding, no format conversion, no quality loss.

### Temporal consistency

Beeble processes video as a sequence, not as independent frames. Edges stay consistent across frames without the shimmer and flicker that plagued per-frame approaches. The result is footage you can use in a timeline without spending hours fixing frame-to-frame inconsistencies.

### Controllability

When automatic detection doesn't get it right, you can guide the extraction. Select what to keep, adjust the matte, and refine the result — rather than accepting whatever the algorithm decided and working around it in post.

## Unscreen vs Beeble at a glance

| Feature | Unscreen | Beeble |
| --- | --- | --- |
| Video background removal | Yes | Yes |
| Image background removal | Yes | Yes |
| Cloud-based processing | Yes | Yes |
| Alpha matte quality | Consumer-grade | **Pro-grade** |
| Hair and fine edge handling | Soft, haloed edges | **Clean partial transparency** |
| Lossless alpha export (Unscreen shipped PNG frames and a Pro Bundle matte, but the pro path was compressed video; Beeble exports lossless RGBA PNG sequences and a 10-bit ProRes MOV) | Partial | Yes |
| Max output resolution | Up to 1080p | **Up to 4K** |
| Manual controllability | No | Yes |
| Temporal consistency (Unscreen output could jitter at edges on fast motion; Beeble's model carries a memory of the subject across frames) | Prone to shimmer | **Sequence-aware** |
| Compositing pipeline ready (Unscreen got you there via the Pro Bundle — a compressed matte video recombined by hand in your editor) | Partial | Yes |

Unscreen's column reflects the product as it stood at shutdown.

## More than a replacement

Beeble isn't just a background removal tool. It's a VFX platform with relighting, PBR extraction, and generative video transformation. Once you've extracted your subject, you can relight it, place it in a new environment, or transform it — all within the same platform.

What you lost with Unscreen was a good, simple tool. What you're getting with Beeble is an upgrade — the same ease of use, better output quality, and a path into workflows that Unscreen was never built for.


# Clipdrop Relight vs. Beeble — AI Relighting Compared (2026)
Source: https://beeble.ai/compare/clipdrop-relight-vs-beeble-3d-relight
Published: 2026-08-11

Clipdrop Relight vs Beeble 3D Relight compared: video support, physically based lighting, light types, free tiers, and which relight AI fits your work.

AI relighting promises the same thing everywhere: fix the lighting after the shot, no studio gear required. Two of the best-known browser tools approach it very differently. Clipdrop Relight (part of the Clipdrop suite, owned by Jasper since February 2024) repaints the lighting on a still image with draggable light sources. Beeble's [3D Relight](/features/3d-relight) turns your image or video into a relightable subject inside a real-time 3D scene, built on the SwitchLight 3.0 model.

They overlap on exactly one job — relighting a still image in the browser — and diverge on everything else, so the comparison below is mostly about which job you actually have.

**Clipdrop Relight** is the faster, simpler tool for relighting a still image: drag lights onto a photo, download, done. **Beeble 3D Relight** works on video as well as stills, and relights with physically based lights in a real 3D scene — HDRI, point, spot, area, sun, and video lights, keyframeable on a timeline, with MP4 and PNG sequence export. The short version: for a quick social-post still, Clipdrop Relight does the job; for footage, or any shot where the light has to behave like real light, Beeble is the tool built for it.

| Feature | Clipdrop Relight | Beeble |
| --- | --- | --- |
| Image relighting | Yes | Yes |
| Video relighting | No | Yes |
| Lighting model (Beeble extracts PBR materials from the shot, so new lights cast real direction, falloff, and reflections) | 2D image effect | **Physically based (PBR)** |
| Light types | Draggable colored lights | **HDRI, point, spot, area, sun, video** |
| Animated / keyframed lights | No | Yes |
| 3D scene composition (GLTF models, Gaussian splats, and image backdrops sit in the same scene as your footage) | No | Yes |
| Runs in the browser (Beeble's editor needs Chrome or Edge on desktop with GPU acceleration) | Yes | Yes |
| Export | Image download | **MP4, PNG sequence** |
| Free tier (on Beeble, relighting and exporting inside the editor are free — credits are only spent generating a shot's VFX passes) | 20 images/day | 90 credits/month |

## Still images vs. video footage

This is the hard boundary. Clipdrop Relight processes still images — portraits, product shots, artwork — and there is no video support. Beeble 3D Relight was built for footage first: your clip becomes a relightable subject in the 3D scene, you scrub the timeline, keyframe lights, and export the result as an MP4 or a PNG sequence for compositing. Stills work the same way, as one-frame shots.

If you're here looking for a Clipdrop Relight alternative that handles video, this section is the whole answer.

## A 2D effect vs. physically based light

Clipdrop Relight's AI repaints the lighting onto the flat image. It's fast and often convincing, but there is no 3D scene behind it — light doesn't wrap around the subject with true direction, falloff, or reflections, and you can't move a light *behind* something.

Beeble's in-house [SwitchLight 3.0](/research/switchlight-3-0-is-here) model extracts PBR materials from your shot, so the subject sits in a 3D scene where new lights behave the way light does on set. The difference shows up where 2D relighting struggles: skin, hair, glossy surfaces, hard shadow edges. On video, the gap compounds frame by frame.

## The lighting toolkit

Clipdrop Relight's vocabulary is a set of draggable colored light sources with position and color controls. That's enough for its job, and part of why it's so quick to use.

Beeble 3D Relight gives you a light kit: HDRI environments plus point, spot, area, sun, and video lights — video lights mirror DMX and LED-wall behavior, so a playing clip can illuminate your subject. Lights animate on the timeline with keyframes, and you can build out the scene with GLTF models, Gaussian splats, and image backdrops.

## Speed and simplicity

Clipdrop Relight wins this one. Upload, drag, download — nothing to learn, no wait beyond the render, and any browser will do.

Beeble asks more of you. A shot is first processed into VFX passes (a few minutes), and the editor is desktop-only: Chrome or Edge with GPU acceleration, nothing to install. In exchange, the processed shot stays a live scene — you can relight it, reframe it, and export it again without paying for the processing twice.

If what you actually want is Clipdrop's one-shot simplicity on video, Beeble has that too: [SwitchX](/features/switchx) relights footage generatively, driven by a reference image or a prompt, as part of a full scene transform. This comparison sticks with 3D Relight because it's the direct counterpart to Clipdrop's draggable lights — you place lights, and they cast.

## Free tiers and pricing

Clipdrop's free tier covers about 20 images per day at the time of writing, with high-resolution output behind the Pro subscription.

Beeble's free Starter plan includes 90 credits per month. Credits are only spent generating a shot's VFX passes — 3 credits per second of video, 3 per image — and once a shot is processed, relighting and exporting inside the editor cost nothing. The built-in demo shots are free to light, so you can try the editor before spending anything. Starter caps uploaded clips at 5 seconds and covers non-commercial use; paid plans raise the cap and add a full commercial license.

## Choose Clipdrop Relight if

- You're relighting **single still images** for social posts or quick mockups.
- You want **zero learning curve** and an instant result.
- You're working **on a phone or tablet**, where Beeble's editor doesn't run.

## Choose Beeble 3D Relight if

- You're relighting **video**, not just stills.
- The light has to be **physically plausible** — real falloff and reflections, not a painted-on effect.
- You want a **real light kit**: HDRI, point, spot, area, sun, and video lights, keyframed on a timeline.
- Your output feeds **an edit or a compositing pipeline** — MP4 and PNG sequence export, plus the shot's PBR passes in the same toolset.

**Is Clipdrop Relight still available?**

Yes. Clipdrop has changed hands twice — built by InitML, acquired by Stability AI in 2023, then sold to Jasper in February 2024 — and the Relight tool is live as part of the Clipdrop suite, alongside background removal, cleanup, and upscaling.

**Can Clipdrop Relight relight video?**

No — it processes still images only. Beeble 3D Relight is built for footage: the clip becomes a relightable subject in a real-time 3D scene, with keyframeable lights and MP4 or PNG sequence export. Still images run through the same editor as one-frame shots.

**Is Beeble 3D Relight free to use?**

The free Starter plan includes 90 credits per month. Credits are only spent generating a shot's VFX passes (3 credits per second of video, 3 per image); relighting and exporting inside the editor are free, and the built-in demo shots cost nothing to light. Starter caps uploads at 5 seconds and covers non-commercial use.

**What do I need to run Beeble 3D Relight?**

A desktop browser — Chrome or Edge with GPU acceleration. There are no plugins or downloads; the 3D scene runs entirely in the browser.


# LTX vs. Beeble — AI SDR to HDR Conversion Compared (2026)
Source: https://beeble.ai/compare/ltx-vs-beeble-sdr-to-hdr
Published: 2026-07-29

LTX vs Beeble SDR to HDR compared: dynamic range reconstruction, ACES output, speed, resolution, and control. Tested side-by-side, July 2026.

Generative SDR-to-HDR, sometimes called inverse tone mapping (ITM) or HDR reconstruction, is a new category. Instead of tone-curve tricks, a video model infers the high-dynamic-range scene the footage came from, reconstructing plausible detail inside blown-out highlights and crushed shadows, and hands you scene-linear HDR frames a colorist can grade. There are two leading generative options for converting SDR footage to HDR: Lightricks' LTX SDR-to-HDR (an in-context LoRA for the LTX video model) and Beeble's [SDR to HDR](/features/sdr-to-hdr), built on Beeble's proprietary SwitchHDR model.

We ran the same source clips through both models and report the detailed Beeble vs. LTX results below.

Both tools produce real scene-linear HDR (EXR) from SDR footage. **LTX** is faster and outputs at higher resolution. **Beeble** reconstructs substantially more usable dynamic range in highlights and shadows, gives you control over the result with threshold masks and reconstruction prompts, and outputs ACES with longer clip support. The short version: Beeble is the better choice for footage headed into a grade or a VFX pipeline, because it reconstructs more usable dynamic range and outputs ACES. If turnaround speed and output above 1080p matter most today, LTX is the better fit.

| Feature | LTX | Beeble |
| --- | --- | --- |
| Linear HDR (EXR) output | Yes | Yes |
| Dynamic range reconstructed (side-by-side at −6 EV in Resolve, Beeble retains markedly more highlight detail) | Partial | Yes |
| ACES (AP0) color space (LTX outputs linear Rec.709/sRGB only) | No | Yes |
| Highlight / shadow threshold control | No | Yes |
| Prompt-guided reconstruction | No | Yes |
| Processing speed (LTX: minutes; Beeble: ~10 min for a max-length clip) | Yes | Partial |
| Clip length (LTX's hosted API tiers its cap by input resolution: 181 frames up to 1080p, 101 up to 1440p, 41 up to 4K. Beeble's 240 applies at its full 1080p output) | 181 frames | **240 frames** |
| Max output resolution (LTX reaches 4K only on clips of 41 frames or fewer; 4K support coming to Beeble later in 2026) | **Up to 4K** | Up to 1080p |

## Dynamic range reconstruction

The reason to run SDR footage through an HDR model at all is to rebuild the light the camera clipped, and this is where the two models differ most. One thing to be clear about up front: once a highlight clips, the original radiance is gone from the file, and no tool retrieves it, Beeble's included. Both models infer a plausible scene-linear reconstruction, informed by how light behaves in real high-dynamic-range footage. The practical question is whose reconstruction holds up when you grade into it.

In our side-by-side tests we pulled both EXR sequences down 6 stops with the HDR wheels in DaVinci Resolve. Beeble's SwitchHDR output holds recoverable detail deep into the highlights where LTX's has already flattened out. Car headlights, practical lights, bright skies: pushed down in the grade, Beeble's frames keep resolving structure while LTX's clip earlier.

How much this matters depends on how hard you plan to grade. For a light touch, both models deliver. If you are rescuing a blown-out window or relighting around a practical, the extra stops of reconstructed range decide whether the pass is usable.

## Processing speed

Lightricks' LTX is the faster of the two. A 181-frame clip processes in a couple of minutes, while a 240-frame clip on Beeble takes around ten. If you process a high volume of clips and keep what works, that difference adds up.

## Control over the HDR result

LTX's tool is a single shot: upload, preview a frame with an exposure control, download. Nothing lets you influence the reconstruction itself. What the model treats as a highlight worth rebuilding isn't up to you.

Beeble treats SDR-to-HDR as a pass you direct. You set the highlight and shadow thresholds that decide what the model treats as clipped, with a live mask overlay and per-frame scrubbing to check coverage anywhere in the clip. You can write prompts to guide what gets reconstructed inside the blown-out regions ("the sky behind the building is a sunset gradient"), or let the Suggest button draft them from your footage. You can pin a seed to reproduce a result across iterations. And before downloading anything, you preview the output in the browser as a −2 / 0 / +2 EV bracket, so you know what the grade will find before the EXRs land on disk.

For a one-off clip this is a convenience. Across a project, it means you can iterate on a shot instead of re-running and hoping.

## Color science and pipeline fit

LTX outputs linear Rec.709/sRGB EXRs. That is real linear HDR data and it works. Beeble outputs ACES (AP0) by default, the interchange standard film and VFX pipelines are built around, with FPS metadata stamped into the EXR sequence. Either way you set an input transform on import. With Beeble's frames it's the one an ACES pipeline already expects.

On clip length, Beeble processes up to 240 frames per run against LTX's 181, which at 24 fps is ten seconds versus about seven and a half. Resolution runs the other way: LTX outputs up to 4K, while Beeble tops out at 1080p today, with 4K support coming later in 2026. Note that LTX's resolution and clip length trade off against each other: on its hosted API the 181-frame cap holds up to 1080p and drops to 101 frames at 1440p and 41 at 4K. If your delivery is native 4K and upscaling isn't acceptable, that is a real constraint on Beeble today.

## Choose LTX if

- You need **turnaround in minutes** and are processing a high volume of clips.
- You need **output above 1080p** today.
- A **single-shot** workflow (upload, check exposure in the preview, download) fits how you work.

## Choose Beeble if

- You need the **most recoverable dynamic range**, with highlights and shadows that hold up under a real grade.
- Your pipeline runs on **ACES**, or you'd rather not hand-manage color from linear Rec.709/sRGB.
- You want to **steer the result** with threshold masks, reconstruction prompts, and seeds instead of re-running and hoping.
- Your clips run **longer than 181 frames**, or you want the pass alongside relighting, alpha, and VFX passes in one toolset.

**Do LTX and Beeble both output real HDR?**

Yes. Both produce linear floating-point EXR frames with reconstructed highlight and shadow detail, not a tone-mapped 8-bit approximation. Neither model retrieves data destroyed by clipping; both infer plausible detail. The difference is in how much usable range the reconstruction gives you (Beeble's holds up further into the highlights and shadows) and the color space it arrives in (Beeble outputs ACES AP0; LTX outputs linear Rec.709/sRGB).

**Can I control what the AI reconstructs?**

On Beeble's SDR to HDR (the SwitchHDR model), yes: you set highlight/shadow thresholds with a live mask preview, write prompts describing what should be reconstructed in clipped regions, and fix a seed for reproducibility. LTX's tool lets you preview a single frame and adjust exposure, but exposes no controls over the reconstruction itself.

**What resolution can each output?**

LTX outputs up to 4K and Beeble up to 1080p, with 4K support coming to Beeble later in 2026. Clip length caps per run: 240 frames on Beeble at 1080p; LTX's hosted API tiers its cap with resolution, 181 frames up to 1080p and down to 41 at 4K.