Kunfeng AI
Back to creation

Help Center

Seedance 2.5

Seedance 2.5 video generation parameters, reference-material limits, and delivery scope.

Updated
August 7, 2026
Applies to
Current Kunfeng AI client

Quick answer

Seedance 2.5 video generation parameters, reference-material limits, and delivery scope.

Capabilities and limits

Type
Video
Outputs per generation
1
Aspect ratios
adaptive1:14:33:416:99:1621:9
Final clarity
720p1080p4K
Duration
4-30 s
Output audio
Optional
Reference modes
Omni referenceFirst and last frame
Total references
50
Image references
30
Image formats
JPEGPNGWebPBMPTIFFGIFHEICHEIF
Image size per file
< 30 MB
Image dimensions
>300 x 300 px - <6000 x 6000 px
Reference image ratio
>0.4 - <2.5
Video references
10
Video formats
MP4MOV
Video size per file
<= 200 MB
Video dimensions
300 x 300 px - 6000 x 6000 px
Reference image ratio
0.4 - 2.5
Video pixels
409600-8295044 px
Video duration
2-30 s
Combined duration
<= 30 s
Video frame rate
24-60 fps
Audio references
10
Audio formats
WAVMP3
Audio size per file
<= 15 MB
Audio duration
2-30 s
Combined duration
<= 30 s
First and last frames
A first-frame image is required; one last-frame image is optional
Combination rule
Audio-only reference is supported
Combination rule
Reference images or videos containing a real person's face are not supported and will be rejected by the model service

Video model · Current Kunfeng AI client

Contact support

Scan with WeChat to contact support

WeChat support QR code