AI VIDEO MODEL

SEEDANCE 2.5

COMING SOON

30 seconds of cinema in one pass.
Native 4K, sound included.

WHAT'S COMING

SEEDANCE 2.5
AT A GLANCE

DURATION

30s

ONE SHOT, ONE PASS

Native 30-second clips in a single pass. No stitching, double Seedance 2.0's clip length.

RESOLUTION

4K

NATIVE 4K

Crisp at every frame, in any aspect ratio from vertical 9:16 to widescreen 21:9.

REFERENCES

50

FACES, PRODUCTS, PLACES

Up to 50 multimodal reference inputs, locked on-model. Seedance 2.0 takes 12.

ADHERENCE

+20%

THE SHOT YOU WRITE

The shot you write is the shot you get. About 20% better prompt adherence than Seedance 2.0.

01/ NATIVE AUDIO

SOUND IS BORN
WITH THE SHOT

Footsteps land, engines rumble, the score swells on cue. Audio is generated in the same pass as the pixels, never glued on after.


02/ REFERENCES

DRIVE THE FILM WITH
50 REFERENCES

Up to 50 multimodal inputs. A few frames of a face, a product, a location set the direction for the full 30 seconds.


03/ BEYOND 10-BIT

10-BIT IN.
16-BIT ACES OUT.

Seedance tops out at 10-bit. Run the clip through SDR to HDR on Beeble and finish with a true 16-bit ACES2065-1 EXR sequence. Blown highlights rebuilt, crushed shadows cleaned, ready for the grade.

FOR STUDIOS

ENTERPRISE READY

Seedance on Beeble is licensed directly from ByteDance. Built for studios that answer to legal.

LICENSED DIRECTLY FROM BYTEDANCE

We license Seedance directly from ByteDance. No middle layer between your footage and the model.

LICENSED LIKENESS

Hold the rights to a real actor's likeness? On the Enterprise Plan you can generate with their face. Cleared, contracted, documented.

NO-TRAINING GUARANTEE

Your uploads and generations never train the model. Guaranteed in writing.

Talk to us

FREQUENTLY
ASKED
QUESTIONS

BE FIRST IN LINE

One email when Seedance 2.5 goes live. Nothing else.