Seedance 1.0 Pro Text-to-Video Generator

Problem

Video generation APIs are often asynchronous: after submitting a prompt, the caller must poll task status, handle completion, and download the MP4 file. seedance 1.0-video-generator wraps the Volcengine Seedance 1.0 Pro text-to-video workflow into runnable scripts, useful for local or CI validation of text-to-video output.

How It Works

The skill reads the Volcengine api_key and model from config.json, with the default model doubao-seedance-1-0-pro-fast-251015. The main scripts include:
- seedance_1_0_pro_generator.py: submits a generation task and supports --duration, --resolution, --poll, --max-attempts, and --interval.
- check_task.py: queries a task_id and can download the video when the task succeeds.

The default flow is to create the task, poll its status, and download the file. The default video duration is 5 seconds, with support for 4-12 seconds; output is MP4 at 1080p by default. Volcengine watermarking is enabled by default and can be disabled with --no-watermark.

Boundaries

This skill is suitable for engineers who need to call the Seedance 1.0 Pro text-to-video API. It is not suitable for environments without an API key, account quota, or network access to Volcengine. Async generation typically takes 2-5 minutes, so missing configuration, invalid keys, account inference limits, and network errors should be handled.