seedance-video-generation_skill
- Python
2.5k
GitHub Stars
6
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill seedance-video-generation- _meta.json659 B
- CHANGELOG.md873 B
- how_to_send_video_via_feishu_app.md4.9 KB
- LICENSE.txt11.1 KB
- seedance.py13.0 KB
- SKILL.md14.3 KB
Overview
This skill enables AI video generation using ByteDance Seedance models via the Volcengine Ark API. It supports text-to-video, several image-to-video modes (first frame, first+last frame, reference images), draft previews, and task management. The default model is Seedance 1.5 Pro which includes synchronized audio support. Use the included Python CLI for robust execution and automatic handling of local images and retries.
How this skill works
The skill creates asynchronous generation tasks against the Ark API, uploading prompt text and optional images (as URLs or base64 data URLs) and returning a task ID. It polls task status until completion, exposes the resulting video URL (download immediately; links expire in 24 hours), and supports operations to list, wait for, cancel, or delete tasks. Optional parameters let you control model, aspect ratio, duration, resolution, draft mode, audio, seeds, and service tier.
When to use it
- Generate short videos from natural-language prompts (text-to-video).
- Create videos using an existing image as the first frame or first+last frame (image-to-video).
- Use reference-image mode to anchor composition or style (Seedance 1.0 Lite I2V).
- Create a low-cost draft preview before committing to a final render (Seedance 1.5 Pro).
- Manage and monitor generation tasks: check status, list history, cancel, or delete tasks.
Best practices
- Always set ARK_API_KEY in your environment before calling the API and show the task ID to users.
- Default to Seedance 1.5 Pro and to 720p/16:9/5s with audio for text-to-video unless specific requirements exist.
- Convert local image files to base64 data URLs before uploading; respect image format, size, and aspect ratio limits.
- Use draft mode for quick previews and only render final videos after review.
- Poll every 15 seconds for status and download the resulting video immediately because URLs expire in 24 hours.
Example use cases
- Create a 5s product teaser from a short prompt and download the MP4 to share with stakeholders.
- Animate a portrait: provide a first-frame photo and a prompt to produce a short motion clip.
- Produce a sequence of consecutive videos by returning the last frame from one task and using it as the first frame of the next.
- Run a cheap 480p draft to validate style, then convert the draft task into a final 720p/1080p render.
- Batch-manage tasks: create multiple tasks, poll for completion, list succeeded tasks, and delete history older than 7 days.
FAQ
Seedance 1.5 Pro supports synchronized audio and draft mode. Other models may not support audio or drafts.
What image formats and sizes are supported?
Supported formats include jpeg, png, webp, bmp, tiff, gif (and heic/heif for 1.5 Pro). Dimensions must be 300–6000 px per side, aspect ratio 0.4–2.5, and max file size 30 MB.
How long do task results stay available?
Video URLs expire in 24 hours and task history is kept for 7 days. Download outputs immediately and record task IDs for later checks.
How do I generate a final video from a draft?
Create a draft task (draft: true), wait until it succeeds, then create a new task passing the draft_task id to produce the final higher-resolution video.
