Seedance API Guide
33 sections · 2 reusable examples with full prompts
Seedance API Integration Guide
Use Seedream 5.0-lite for image generation, then use Seedance 2.0, Seedance 2.0 Fast, or Seedance 1.5 Pro for video generation tasks.
Seedream 5.0-lite
Synchronous image generation from text prompts and optional reference images.
Asset Management
Create asset groups, upload reusable media assets, and retrieve asset metadata for reference-driven workflows.
Seedance Video Generation
Shared task API for Seedance 2.0, Seedance 2.0 Fast, and Seedance 1.5 Pro.
Image-to-Video Production Workflow
A compact workflow for preparing a Seedream reference image or existing asset input, then converting it into a Seedance video task.
Prepare Reference Input seedream or asset
Choose either a newly generated Seedream reference image or an existing uploaded asset ID before creating the video task. You can also skip reference inputs and generate a video directly from text.
Generate Reference Image seedream-5-0-260128
Create the first visual reference when you do not already have approved source media.
Prompt
Create Asset asset management
If you already have an image, video, or audio file, create an asset first and use the returned asset ID as an input for video generation.
Create Video Task seedance-2.0 / fast / 1.5-pro
Submit the Seedance task using the generated image URL, the asset ID, or a text-only description, then set model, ratio, duration, resolution, and audio options. Note: Seedance may block direct nudity terms; use Grok to optimize the prompt before submitting.
Prompt
Seedance Model Guide
Practical notes summarized from the local model catalog behind /docs/models, focused on the models used in this API guide.
Seedream 5.0-lite
Text-to-image and reference-image generation for creating first frames and visual concepts.
Use cases
Prompt-only image generation for concept frames.
Single reference image plus prompt for guided image creation.
Multi-image reference input for consistent visual direction.
First-frame creation before Seedance video generation.
Inputs and outputs
Inputs: text and optional image references.
Outputs: generated image URLs.
Supports text-to-image and image-to-image workflows.
Supports watermark control through the model catalog.
Practical notes
Use a clear subject, scene, composition, lighting, and style order.
Keep the first frame clean and readable for downstream video generation.
For video use, avoid cropped hands, extreme poses, and cluttered backgrounds.
Per-image pricing in the local catalog is $0.035 / image.
Seedance 2.0
Professional multimodal video generation model for quality-focused production.
Modes
Text to video.
First-frame image to video.
First-and-last-frame image to video.
Multimodal reference generation with image, video, audio, and text inputs.
Video modify and video extend workflows.
Controls
Durations: 4 to 15 seconds.
Resolutions: 480p, 720p, 1080p, and 4K.
Ratios: 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, and adaptive.
Supports audio generation and watermark control.
Frame rate: 24 fps.
When to choose it
Use when visual detail, camera control, and character consistency matter.
Best for advertising, film-style content, and production-ready social assets.
Choose 1080p only when the extra quality is worth the cost and latency.
Use reference assets to preserve tone, material, and shot language.
Seedance 2.0 Fast
Faster multimodal video generation with the same core mode family as Seedance 2.0.
Strengths
Optimized for faster generation and iteration speed.
Supports text, image, video, audio, and reference-driven creation.
Works for video editing and video extension.
Useful for testing multiple prompt or reference variations.
Limits
Resolutions are limited to 480p and 720p.
Supports the same 4 to 15 second duration range.
Supports audio generation and watermark control.
Frame rate: 24 fps.
When to choose it
Use for previews, draft review, and high-throughput creative iteration.
Use when delivery speed matters more than maximum output resolution.
Upgrade the final selected direction to Seedance 2.0 if 1080p is required.
Keep prompts concise to benefit from faster turnaround.
Seedance 1.5 Pro
Video generation model for prompt-only, first-frame, and first-last-frame workflows.
Modes
Text to video.
First-frame image to video.
First-and-last-frame image to video.
Input types: image and text.
Output type: video.
Controls
Durations: 4 to 12 seconds in the local catalog.
Resolutions: 480p, 720p, and 1080p.
Ratios follow the shared Seedance video ratio set.
Supports silent video or synchronized audio output.
When to choose it
Use when compatibility with the Seedance 1.5 surface is needed.
Use for simpler text-to-video or first-frame workflows.
Use when multimodal reference, video modify, or video extend is not required.
Prefer Seedance 2.0 for the broadest modern multimodal workflow coverage.
FAQ
Reserved for project-specific questions, troubleshooting notes, and model selection guidance.
Input image may contain a real person
Try adding the asset through Create Asset, then generate the video with the asset ID.
Output video may be related to copyright restrictions
Your uploaded material contains too much copyrighted content. We recommend uploading the source video clip or image to Claude, Gemini, or ChatGPT, asking it to analyze the material and generate a prompt based on it, then using that prompt for generation instead.
Output audio may contain sensitive information
Turn off the generate audio switch, then generate again.
More questions
Open the Helpful Tips documentation for additional troubleshooting and workflow guidance.