Seedance 2.5 ComfyUI

ComfyUI custom nodes for the Seedance 2.5 API, delivered through MuAPI.

This pack follows the same workflow pattern as seedance2-comfyui: add one API Key node, connect it to a generation node, then connect video_url to Save Video and first_frame to Preview Image.

📺 Video Tutorial

How to Access Seedance 2.5 API (Step-by-Step Guide) — a full walkthrough of getting an API key and making your first Seedance 2.5 call via MuAPI.

Related Projects

• MuAPI ComfyUI integration docs — Integration guidance for running this node pack through MuAPI; create an access key, explore the Seedance 2.5 landing page, or try Seedance 2.5 text-to-video.
• Seedance-2.5-API — Python wrapper for the same Seedance 2.5 API.
• seedance2-comfyui — Related Seedance ComfyUI nodes and workflows.
• seedance-2.5-mcp — Focused MCP server for driving the 720p and 480p Seedance 2.5 Preview routes.
• awesome-seedance-2.5-api-prompts — Curated Seedance 2.5 prompts, camera controls, and API examples.
• n8n-nodes-seedance2 — Automate related Seedance generation workflows in n8n.
• muapi-comfyui — Broader ComfyUI nodes for MuAPI’s image, video, and audio catalog.
• seedance-2-mcp — MCP server for generating Seedance 2 videos through MuAPI.
• awesome-seedance-motion-control-api — Seedance 2 & 2.5 Motion Control API guide — recast a performance video with new performers while preserving motion, camera, and location

Included nodes

Node	What it does
Seedance 2.5 API Key	Stores one MuAPI key for the workflow.
Text-to-Video	Generates a 720p or 480p clip from a prompt.
Image-to-Video	Animates one input image using the 2.5 image_url contract.
First/Last Frame	Interpolates between exactly two input images.
Omni Reference	Combines up to 20 images, 6 videos, and 6 audio references.
Spicy Text/Image-to-Video	720p relaxed-moderation siblings exposed by the API.
Consistent Character	Creates a reusable character sheet from 1–3 reference images.
Consistent Video	Anchors Omni Reference on a character sheet.
Extend	Uses MuAPI’s Seedance-family extension endpoint.
Save Video	Downloads a URL to ComfyUI/output and returns decoded frames.

Installation

Copy this directory into ComfyUI/custom_nodes/seedance2.5-comfyui, then install the Python dependencies in the same environment ComfyUI uses:

pip install -r ComfyUI/custom_nodes/seedance2.5-comfyui/requirements.txt

Restart ComfyUI. The example workflows can be loaded with File → Load.

Included examples are Seedance25_T2V_Example.json, Seedance25_FirstLastFrame_Example.json, and Seedance25_ConsistentCharacter_Example.json.

API key

Use the 🔑 Seedance 2.5 API Key node and connect its output to the generation nodes. Alternatively, leave the node field empty and configure one of these:

export MUAPI_API_KEY="your_muapi_key"

or configure the MuAPI CLI:

muapi auth configure --api-key YOUR_KEY

The node also reads ~/.muapi/config.json when it contains an api_key field.

Quick workflow

1. Add 🔑 Seedance 2.5 API Key.
2. Add 🌱 Seedance 2.5 Text-to-Video.
3. Choose 720p or 480p, aspect ratio, duration, and output format.
4. Connect video_url to 🌱 Seedance 2.5 Save Video.
5. Connect first_frame to ComfyUI’s Preview Image node.
6. Queue the prompt.

duration accepts 4–30 seconds. seed = -1 uses the provider’s default random behavior. generate_audio, camera_fixed, output_format, and the optional character ID are passed through to the API.

Reference inputs

Image-to-Video

Connect one IMAGE. The node uploads it and sends it as image_url to the 2.5 endpoint.

First/Last Frame

Connect first_image and last_image. The node uploads both in order and sends:

{
  "images_list": ["first-frame-url", "last-frame-url"]
}

Omni Reference

Use @image1, @video1, and @audio1 in the prompt. Connect media slots from 1 upward so the prompt numbering matches the order of the uploaded references. Video and audio fields accept a URL, an absolute local path, or a filename from ComfyUI/input.

The 2.5 limits documented by the API are 20 images, 6 videos, and 6 audio clips. If a local video/audio file is selected, it is uploaded through /upload_file before generation.

Consistent character

LoadImage → Seedance 2.5 Consistent Character → Consistent Video → Save Video

The character node returns sheet_image, sheet_url, and character_id. Connect sheet_image to Consistent Video for an image anchor, or connect character_id to a text/image/omni node and reference it inline as @character:<id>.

API endpoints

The generation nodes use the documented 2.5 endpoints:

Capability	720p	480p
Text-to-Video	seedance-2.5-text-to-video	seedance-2.5-text-to-video-480p
Image-to-Video	seedance-2.5-image-to-video	seedance-2.5-image-to-video-480p
First/Last Frame	seedance-2.5-first-last-frame	seedance-2.5-first-last-frame-480p
Omni Reference	seedance-2.5-omni-reference	seedance-2.5-omni-reference-480p

All requests use x-api-key, poll predictions/{request_id}/result, and wait for completion before returning the URL and first frame. Seedance 2.5 currently exposes 480p and 720p tiers; this pack does not advertise 1080p or 4K controls.

The upstream API repository documents video extension and editing through Seedance 2.0-compatible endpoints. The included Extend node therefore targets seedance-v2.0-extend intentionally.

Requirements

Python 3.8+, requests, Pillow, numpy, and opencv-python. ComfyUI supplies the installed PyTorch runtime used for IMAGE tensors.