Hi Comfy community! As many of you might already know, OmniGen2 is natively supported in ComfyUI!
Developed by VectorSpaceLab team, OmniGen2 is a 7B parameter unified multimodal model that combines text-to-image generation, image editing, and multi-image composition in one powerful architecture.
Model Highlights
Text-to-Image Generation: Create high-quality images from text prompts
Instruction-guided Editing: Make precise edits with natural language commands
Multi-Image Composition: Combine elements from multiple images seamlessly
Text in Images: Generate clear text content within images
Visual Understanding: Powered by Qwen-VL-2.5 for superior image analysis
Get Started
Update ComfyUI or ComfyUI desktop
Visit our documentation
Follow the guide in our documentation to download models, workflows, and then run them.
Example Outputs
Text-to-Image



Image Editing

Multiple input
Check our documentation for more details: docs.comfy.org/tutorials/image/omnigen/omnigen2
As always, enjoy creating!
Both of the sample workflows are running fine on a Mac.
runs like a charm on a 4090, and results seem a bit better than flux kontext