doubao-video-generation

name	doubao-video-generation
description	集成豆包视频生成API，支持文生视频、图生视频、有声视频生成、任务查询等功能。使用时当用户需要调用豆包seedance模型生成视频，或者需要查询视频生成任务结果时触发。

Doubao Video Generation

Overview

This skill provides integration with the Doubao Seedance video generation API, enabling text-to-video, image-to-video, audio-enabled video generation, and async task status querying. It supports multiple models from Doubao with different capabilities.

Supported Capabilities

1. Create Video Generation Task

Supports multiple generation modes:

• Text-to-Video: Generate video from text prompt
• Image-to-Video (First Frame): Generate video starting from a provided first frame image
• Image-to-Video (First + Last Frame): Generate video transitioning between two provided frames (supported by select models)
• Audio-enabled Video: Generate video with automatically matched environmental audio (supported by doubao-seedance-1-5-pro-251215)
• Draft Mode: Fast low-resolution preview before generating high-quality final video
• Offline Flex Tier: Lower cost batch processing for non-urgent tasks

2. Query Video Generation Task

Check the status of an async generation task and retrieve the finished video URL when complete.

Configuration

Before using this API, create a .env file in the project root directory:

DOUBBAO_API_KEY=your_api_key_here
DOUBBAO_VIDEO_API_BASE=https://api.cozex.cn/v1/videos/generations
DOUBBAO_VIDEO_QUERY_BASE=https://api.cozex.cn/v1/videos/generations

Or copy from .env.example:

cp .env.example .env
# Edit .env and add your API key

The script will automatically load configuration from the .env file.

Usage

Basic Text-to-Video

python scripts/generate-video.py --prompt "A cat playing piano" --resolution 1080p

Image-to-Video (First Frame)

python scripts/generate-video.py --prompt "A cat walking" --images "https://example.com/image.jpg"

Image-to-Video (First + Last Frame)

python scripts/generate-video.py --prompt "Transition between scenes" --images "https://example.com/start.jpg,https://example.com/end.jpg"

Audio-enabled Video

python scripts/generate-video.py --prompt "Forest scene with birds" --generate-audio --model doubao-seedance-1-5-pro-251215

Draft Mode (Quick Preview)

python scripts/generate-video.py --prompt "Test video" --draft --resolution 480p

Query Task Status

python scripts/generate-video.py --task-id "task-xxx"

Poll Until Completion

python scripts/generate-video.py --task-id "task-xxx" --poll --poll-interval 10

Save Response to File

python scripts/generate-video.py --prompt "A sunset" --output result.json

Command Line Arguments

Argument	Required	Default	Description
--model	No	doubao-seedance-1-5-pro-251215	Model name
--prompt	Yes*	-	Text prompt for video generation
--resolution	No	1080p	Video resolution (480p/720p/1080p)
--ratio	No	16:9	Aspect ratio
--duration	No	12	Duration in seconds (2-12)
--images	No	-	Comma-separated image URLs
--fps	No	-	Frame rate
--seed	No	-1	Random seed
--watermark	No	False	Add watermark
--camera-fixed	No	False	Fixed camera
--generate-audio	No	False	Generate audio (model-specific)
--draft	No	False	Draft mode
--draft-task-id	No	-	Draft task ID for final generation
--service-tier	No	-	Service tier (flex for offline)
--task-id	No	-	Task ID to query
--poll	No	False	Poll until completion
--poll-interval	No	10	Poll interval in seconds
--output	No	-	Output file for JSON response

*Required unless --task-id or --draft-task-id is provided

Model Selection

Model	Resolution	Text to Video	Image to Video (1 frame)	Image to Video (2 frames)	Audio
doubao-seedance-1-0-pro-250528	480p/720p/1080p	✅	✅	✅	❌
doubao-seedance-1-0-pro-fast-251015	480p/720p/1080p	✅	✅	❌	❌
doubao-seedance-1-5-pro-251215	480p/720p/1080p	✅	✅	✅	✅

Resources

references/

• api-doc.md - Complete API documentation with request/response examples and full parameter reference