OmniGen2 Multimodal Image Editor
OmniGen2 is an open-source model that both understands and generates images — text-to-image, instruction-based editing, and multi-image composition — running efficiently on 6GB+ VRAM in ComfyUI.
✓ 100% free — download the workflow and run it locally.Preview
Requirements
- GPU: NVIDIA RTX 30XX/40XX/50XX (FP16)
- VRAM: 6 GB minimum
- OS: Windows
- Storage: 30 GB+ free
Model downloads
- omnigen2_fp16.safetensors (diffusion) https://huggingface.co/Comfy-Org/Omnigen2_ComfyUI_repackaged/tree/main/split_files/diffusion_models
- qwen_2.5_vl_fp16.safetensors (text encoder) https://huggingface.co/Comfy-Org/Omnigen2_ComfyUI_repackaged/tree/main/split_files/text_encoders
- ae.safetensors (VAE) https://huggingface.co/Comfy-Org/Omnigen2_ComfyUI_repackaged/tree/main/split_files/vae
Custom nodes
- ComfyUI Manager https://github.com/Comfy-Org/ComfyUI-Manager
Setup steps
- Install ComfyUI Manager and place the three models in their folders.
- Load the workflow below into ComfyUI.
- Write a detailed, instruction-based prompt describing the edit or generation.
- Set Image CFG between 2.0–4.0 to control likeness, then run.
Notes & tips
- Under 1 minute on an RTX 4090.
- Project GitHub: https://github.com/VectorSpaceLab/OmniGen2
Want the 1-click version?
Skip the manual model downloads and node installs — get a ready-to-run installer, or unlock all 75+ with Local Lab Pro.