Skip to content

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

English | 中文


ComfyUI-UniGen Node

🎬 ComfyUI Custom Node for Multi-Cloud Video & Image Generation (VolcEngine + DashScope + OOpii)

Custom nodes for video/image generation via cloud APIs, supporting VolcEngine Ark (Seedance video / Seedream image), DashScope (Wan 3.0 video), and OOpii NewAPI gateway (GPT-Image class image models) model series.

English | 中文


✨ Features

Video Generation

  • Text-to-Video: Generate videos from text descriptions
  • Image-to-Video: Supports single image (first frame), dual images (start/end frames), and multi-image (reference) modes
  • Reference Audio: Pass audio as voice reference (Seedance 2.0)
  • Parallel Generation: Generate multiple seeds with the same prompt (multi-roll mode)

Image Generation

  • Text-to-Image: Generate images from text descriptions, up to 4K resolution
  • Image-to-Image: Generate new images from reference images + prompts
  • Multi-Reference: Supports up to 10 reference images, or drag-and-drop any number of assets via Load Reference Images
  • OOpii GPT-Image Models: all-image-2 / gpt-image-2 / grok-imagine / monkey-image series, with original/2K/4K resolution presets (scaled from the reference image, auto-normalized to GPT-Image constraints)

General

  • Multi-Model Support: Seedance 2.0/1.5/1.0 (video), Seedream 5.0/4.5/4.0/3.1 (image)
  • Full Parameters: Resolution, aspect ratio, audio generation, locked camera, service tier, etc.
  • Auto Model List: Automatically fetches available models when a 404 error occurs

📦 Installation

Method 1: Clone to custom_nodes

cd ComfyUI/custom_nodes
git clone https://github.com/gzsiang/ComfyUI-UniGen.git

Method 2: Manual Install

  1. Download this repository
  2. Extract to ComfyUI/custom_nodes/ComfyUI-UniGen/
  3. Restart ComfyUI

🔑 Getting an API Key

  1. Visit VolcEngine Console
  2. Go to the "Ark" service
  3. Create an API Key
  4. Enable model access:
    • Video generation: Seedance series
    • Image generation: Seedream series

OOpii (NewAPI-compatible gateway)

  1. Get an API Key from OOpii
  2. Fill the key into the 🔑 OOpii API Config node (auto-encrypted). Base URL defaults to https://www.oopii.cc, any other NewAPI gateway URL works too
  3. Image models include all-image-2 / gpt-image-2 / monkey-image-pro, etc.

🚀 Usage

Nodes

Node Name Function Inputs
🎬 VolcEngine Text-to-Video Text to video API Key, model, prompt, parameters
🎬 VolcEngine Image-to-Video Image to video API Key, model, image, prompt, parameters
🖼️ VolcEngine Image Generation Text/image to image API Key, model, prompt, parameters
🔑 OOpii API Config OOpii/NewAPI gateway credentials API Key, Base URL
🖼️ OOpii Image Generation Text/image to image (GPT-Image class models) Model, prompt, resolution preset
🖼️ Load Reference Images Drag-and-drop multiple reference images (generic) File drop

Image-to-Video Modes

Automatically determined by input image count:

Image Count Mode Description
1 image First-frame mode Uses image as video start frame
2 images Start/end frame mode First image = start, second = end
3-9 images Multi-reference mode Used as style/content reference

Reference Audio

  • When audio is provided, the generated video will use the audio's voice
  • Note: Must provide at least 1 image when using reference audio

⚙️ Parameters

Video Generation Parameters

Parameter Type Range Default Description
API Key PASSWORD - - VolcEngine Ark API Key (password input)
Model Dropdown - doubao-seedance-2-0-260128 Video generation model
Prompt String - - Video content description
Concurrency Int 1-5 1 Number of parallel generations per prompt
Video Duration (s) Int 4-15 5 Video length
Resolution Dropdown 480p/720p 720p Video resolution
Aspect Ratio Dropdown Various 16:9 (text-to-video) / adaptive (image-to-video) Video aspect ratio
Watermark Boolean - True Whether to add watermark
Random Seed Int -1~∞ -1 -1 for random
Poll Interval Int 3-30 10 Task status check interval (seconds)
Max Wait Int 60-1800 600 Maximum wait time (seconds)

Seedance 2.0 Specific Parameters

Parameter Type Default Description
Generate Audio Boolean True Whether to generate video audio
Locked Camera Boolean False Lock camera perspective
Return Last Frame Boolean False Also return the last frame image
Service Tier Dropdown default default (standard) / flex (offline, ~50% cheaper)

Image Generation Parameters

Parameter Type Range Default Description
API Key PASSWORD - - VolcEngine Ark API Key (password input)
Model Dropdown - doubao-seedream-4-5-251128 Image generation model
Prompt String - - Image content description
Reference Image Image - - Optional, for image-to-image
Concurrency Int 1-5 1 Number of parallel generations per prompt
Image Size Dropdown Various 1024x1024 Output image size
Service Tier Dropdown - default default (standard) / flex (offline, ~50% cheaper)
Watermark Boolean - False Whether to add watermark
Random Seed Int -1~∞ -1 -1 for random

OOpii Image Generation Parameters

Parameter Type Default Description
Model Dropdown all-image-2 OOpii image models (all-image-2 / gpt-image-2 / monkey-image-pro / grok-imagine, etc.)
Prompt String - Image content description
Concurrency Int 1 Number of parallel generations per prompt
Resolution Dropdown original original = reference image resolution; 2K/4K = scaled from the reference aspect ratio (long side 2048/3840); 16:9 when no reference
Reference Image REF_IMG_BATCH / IMAGE - Load Reference Images output or IMAGE tensor; all images are sent to the model (auto-normalized to GPT-Image constraints)
Random Seed Int -1 -1 for random

📋 Supported Models

Video Generation - Seedance Series

Seedance 2.0 (Recommended)

  • doubao-seedance-2-0-260128 - Latest version, supports multi-modal references

Seedance 1.5

  • doubao-seedance-1-5-pro-251215

Seedance 1.0

  • doubao-seedance-1-0-pro-250528
  • doubao-seedance-1-0-pro
  • doubao-seedance-1-0-pro-fast
  • doubao-seedance-1-0-lite

Image Generation - Seedream Series

  • doubao-seedream-5-0
  • doubao-seedream-4-5-251128
  • doubao-seedream-4-0-250828
  • doubao-seedream-3-1-250312

Image Generation - OOpii (NewAPI Gateway)

  • all-image-2 / all-image-2-DD / all-image-2-H - reverse gpt-image-2 channel
  • gpt-image-2 - OpenAI GPT-Image official channel
  • monkey-image-pro / monkey-image-flash 2
  • grok-imagine-image-pro / grok-imagine-image-2.0
  • mj_* - Midjourney series (imagine/upscale/variation, etc.)

📁 Workflow Examples

The repository includes example workflow files:

  • workflows/火山API测试.json - Basic test workflow (text-to-video + image-to-video)

Import: ComfyUI interface → Load → select the json file


🔄 Output Format

Video Nodes

Port Type Description
Video Frames IMAGE Video frame sequence, can connect to VHS or other nodes for preview/save
Audio AUDIO Video audio, can connect to audio output nodes
Video Info STRING JSON metadata (fps, frame count, duration, etc.)

Image Nodes

Port Type Description
Image IMAGE Generated image sequence
Generation Info STRING JSON metadata (size, seed, etc.)

⚠️ Notes

  1. API Key Security: Do not share your API Key when sharing workflows
  2. Image Limits: Image-to-video supports up to 9 images, image generation supports up to 10 reference images
  3. Resolution Limits: Seedance 2.0 max 720p, older versions max 1080p; Seedream max 4K
  4. Audio Reference: Must provide an image to use reference audio
  5. Service Tier: flex mode uses offline inference, ~50% cheaper but may have longer wait times
  6. Concurrency: Max 5 concurrent requests. With >1, multiple videos/images are generated simultaneously — watch API costs
  7. GPT-Image Size Constraints: The OOpii image node auto-normalizes sizes (16-multiples, 655,360~8,294,400 total pixels, ratio ≤3:1). original follows the first reference image resolution; 2K/4K scale from the reference ratio. Output resolution is determined only by the resolution preset, not the reference image. Reference images should be ≥1024px on the long side and ≤1.5MB

📄 License

MIT License


🔗 Related Links

About

No description, website, or topics provided.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages