Edit an existing video from a text prompt with Seedance 2.5 Video Edit 480p — rewrite lighting, style, or environment. Try free — pay per generation.
About this model
Seedance 2.5 Video Edit 480p is the video-editing endpoint in the Seedance 2.5 family. It edits an input video from a natural-language prompt — the reference video drives subject identity, composition, and motion, while the model rewrites lighting, style, weather, environment, or specific elements as instructed. Output resolution is 480p.
Cost analysis
| Provider | Cost | Notes |
|---|---|---|
| muapiapp | $0.11/sec (480p) | Billed on the combined duration of the input video plus the generated output, at 65% of the base per-second rate. |
Billed on the combined duration of the input video plus the generated output, at 65% of the base per-second rate.
* Competitor pricing is estimated based on similar model architectures and usage tiers.
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt | string | Instructions for the edit — lighting, style, weather, environment, or specific elements to change. Automatically prefixed with "Edit the input video." | Change the lighting to golden hour and add light fog to the environment. |
| Video | string | URL of the video to edit. Videos longer than 30s are trimmed to 30s. | https://cdn.muapi.ai/assets/seedance-2.5-video-edit-in.mp4 |
| Reference Images | array | Reference image URLs guiding subject identity or style. Up to 30 images. | undefined |
| Reference Audio | array | Reference audio URLs guiding audio generation. Up to 10 files. | undefined |
| Duration | int | The duration of the generated video in seconds. | 5 |
| Aspect Ratio | Enum (7 options) | Aspect ratio of the output video. | 16:9 |
| Generate Audio | boolean | Whether to generate new audio. If false, the input video's original audio is preserved. | true |
| Seed | int | Random seed for reproducible generation. Use -1 for random. | 42 |
Instructions for the edit — lighting, style, weather, environment, or specific elements to change. Automatically prefixed with "Edit the input video."
Change the lighting to golden hour and add light fog to the environment.URL of the video to edit. Videos longer than 30s are trimmed to 30s.
https://cdn.muapi.ai/assets/seedance-2.5-video-edit-in.mp4Reference image URLs guiding subject identity or style. Up to 30 images.
undefinedReference audio URLs guiding audio generation. Up to 10 files.
undefinedThe duration of the generated video in seconds.
5Aspect ratio of the output video.
16:9Whether to generate new audio. If false, the input video's original audio is preserved.
trueRandom seed for reproducible generation. Use -1 for random.
42Developer documentation
Provide the Source Video
video URL. Videos longer than 30s are trimmed to 30s.Write Edit Instructions
prompt — lighting, style, weather, environment, or specific elements.Optional References
reference_images (up to 30) to guide subject identity or style, reference_audios (up to 10) to guide audio generation.Configure Output
duration (4–30s), aspect_ratio, generate_audio, and optionally seed.Submit via the API
curl -X POST https://api.muapi.ai/api/v1/seedance-2.5-video-edit-480p \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "Change the lighting to golden hour and add light fog to the environment.",
"video": "https://example.com/input.mp4",
"duration": 5
}'
GET /api/v1/predictions/{request_id}/result until status is completed, then retrieve the video URL from output.Frequently asked
It edits an existing video from a text prompt — the input video drives subject identity, composition, and motion, while the prompt controls lighting, style, weather, environment, or specific elements.
Yes — the reference video drives composition and motion; the prompt only controls what's rewritten (lighting, style, environment, specific elements).
Billed at 65% of the base per-second rate ($0.11/sec at 480p) on the combined duration of the input video plus the generated output, since a video reference is always present.
Videos longer than 30 seconds are automatically trimmed to 30 seconds before editing.
Yes — set `generate_audio` to `false` to keep the input video's original audio instead of generating new audio.