Doubao Video Generation Assistant
Paste the following prompt into your AI chat to install this skill:
Install @user_1707615b/t-h-i-n-g-o-2026video into your AI assistant following https://skillhub.cn/install/skillhub.md.
About this skill
Problem
When integrating Doubao Seedance video generation, the main pain points are not just “can it generate?” but how to poll asynchronous tasks, save MP4 files reliably, distinguish OpenAI-compatible relay formats from the official Ark format, and handle image/video URL constraints. Hand-written calls easily break due to malformed base_url values, wrong API paths, expired temporary links, or interactive setup failures in non-interactive environments.
Workflow and Capabilities
This skill wraps the calling process in a local script: it reads base_url, api_key, and model from config.json, builds parameters such as prompt, ratio, and duration, and submits the task via python scripts/gen.py. The script polls the asynchronous task for up to 600 seconds by default, then downloads the video to a local directory and returns the MP4 path. Key capabilities include:
- Text-to-video: generate 5-second or 10-second videos from Chinese or English prompts, with ratios such as
16:9,9:16, and1:1. - Image-to-video: use a publicly accessible first-frame image URL as a generation reference.
- Video extension/editing: provide a reference video URL to request extension or modification, subject to relay support.
- Local saving: download MP4 by default, or return only the URL with
--no-download.
Boundaries
This skill targets Thingo’s OpenAI-compatible format and defaults to doubao-seedance-2-0-260128. It does not upload local files; reference images/videos must use http(s):// URLs. Generated links usually expire in about 24 hours, insufficient balance causes quota errors, and custom resolution parameters are not supported. If the task is actually image generation, an image-generation skill is a better fit.
Use Cases
- Generate a 5-second 16:9 video from a prompt, poll the async task, and save the MP4 locally.
- Use a provided first-frame image URL to continue a short video with a specified ratio and duration.
- Extend or edit a reference video from its URL, then download the resulting MP4 locally.
- Compare different prompt, ratio, and duration outputs by generating multiple local videos via Thingo.
Best For
- Backend engineers integrating Doubao video through an OpenAI-compatible API to reduce format adaptation.
- Product demo owners turning text descriptions or first-frame images into deliverable MP4 files.
- Script maintainers using Thingo who need stable task polling and MP4 download workflows.
- Video-content operators generating multiple candidates by specified ratio and duration.
Related Skills
Convert Chinese or natural-language requests into paste-ready English image prompts for ChatGPT's web UI, covering generation, editing, multi-image references, and exact text without performing image generation.
A creative AI image workflow for style transfer, reference-based creation, scene replacement, series expansion, material conversion, era shifts, composition rework, lineart conversion, and cartoonization.
Translates star-inspired football memories and fan resonance into compliant original IP poster concepts while avoiding real names, likenesses, official badges, and event logos.
Extracts real web page colors, typography, components, layouts, and interaction styles with a Playwright script to produce a verifiable design guideline.