Seedance 2.5
Seedance 2.5 is ByteDance's high-fidelity video generation model. This skill supports text-to-video, image animation, keyframe transitions, multimodal references, video editing, video extension, and character workflows.
Prerequisites
- MUAPI_API_KEY: You must have an API key from muapi.ai. Set it as an environment variable.
Usage Guide
You can select standard, Intl, or Spicy route variants and 480p, 720p, upscaled 1080p, or upscaled 4K tiers. Durations range from 4 to 30 seconds.
Text-to-Video (T2V)
Generate a video from a descriptive text prompt.
python3.11 skills/seedance-v2/seedance_cli.py t2v --prompt "A cinematic slow-motion shot of a cyberpunk city in the rain" --variant intl --resolution 1080p --wait
Image-to-Video (I2V)
Animate one static image.
python3.11 skills/seedance-v2/seedance_cli.py i2v --image_url "https://example.com/image.jpg" --prompt "Make the clouds move slowly" --wait
First & Last Frame
Generate a transition between exactly two images.
python3.11 skills/seedance-v2/seedance_cli.py first-last --images "https://example.com/start.jpg" "https://example.com/end.jpg" --prompt "Smoothly transition from day to night" --wait
Video Rendering & Editing
Edit an existing video or apply styles.
python3.11 skills/seedance-v2/seedance_cli.py edit --video "https://example.com/video.mp4" --prompt "Turn the sunny afternoon into a rainy blue-hour scene" --wait
Omni-Reference Generation
Condition videos on any combination of images, videos, and audio.
python3.11 skills/seedance-v2/seedance_cli.py omni --prompt "The character @character:ID dances in the rain" --images "https://example.com/bg.jpg" --videos "https://example.com/motion.mp4" --wait
Character Consistency
Create a reusable character sheet for consistent generation.
python3.11 skills/seedance-v2/seedance_cli.py character --images "https://example.com/face.jpg" --outfit "Cyberpunk neon jacket" --name "Neo" --wait
Watermark Removal
Remove MuAPI watermarks from generated videos.
python3.11 skills/seedance-v2/seedance_cli.py watermark-remover --video_url "https://api.muapi.ai/..." --wait
Extending Videos
Continue an existing video from its final frame.
python3.11 skills/seedance-v2/seedance_cli.py extend --video "https://example.com/video.mp4" --prompt "Continue the camera move into the city" --wait
Any route
Use the generate command when you need an exact route slug, including any
Intl or Spicy resolution variant.
python3.11 skills/seedance-v2/seedance_cli.py generate --endpoint seedance-2.5-spicy-video-edit-4k --video "https://example.com/video.mp4" --prompt "Change the lighting to a neon night scene" --wait
Tips for Best Results
- Be Descriptive: Detailed prompts result in better video quality and more accurate motion.
- Wait for Completion: Use the
--waitflag to receive the final video URL directly. Without it, you will receive arequest_idwhich you can check later using thestatuscommand. - Aspect Ratios: Use
16:9for horizontal videos and9:16for vertical content (like TikTok or Reels). - Resolution: Select the route tier with
--resolution; 1080p and 4K are upscaled from the 720p base render. - Variants: Use
--variant intlfor the international route or--variant spicyfor the relaxed-content-safety route.
Commands Reference
| Command | Arguments | Description |
|---|---|---|
t2v |
--prompt, --variant, --resolution, --aspect_ratio, --duration, --seed, --wait |
Text to Video |
i2v |
--image_url, --prompt, --variant, --resolution, --aspect_ratio, --duration, --seed, --wait |
Image to Video |
first-last |
--images (exactly 2), --prompt, --variant, --resolution, --aspect_ratio, --duration, --seed, --wait |
First & Last Frame |
omni |
--prompt, --images, --videos, --audios, --variant, --resolution, --aspect_ratio, --duration, --seed, --wait |
Omni Reference |
character |
--images, --outfit, --name, --wait |
Create Character Sheet |
edit |
--prompt, --video, --variant, --resolution, --reference-images, --reference-audios, --no-audio, --wait |
Video Edit |
watermark-remover |
--video_url, --wait |
Remove Watermark |
extend |
--prompt, --video, --last-image, --variant, --resolution, --no-audio, --wait |
Video Extend |
generate |
--endpoint, --prompt, workflow-specific input flags, --wait |
Any current Seedance 2.5 route |
status |
--request_id |
Check Task Status |