ComfyUI-MixMod

ComfyUI-MixMod provides a powerful way to combine multiple models during sampling.

VRAM requirement: Fitting two or multiple models. Around 12gb is minimum for two sdxl. Around 16gb vram for sdxl+pixartsigma with a Q3 t5xxl encoder.

Features

• Mix multiple models during sampling
• SD1.5+SDXL (use https://huggingface.co/ostris/sdxl-sd1-vae-lora to align the latents)
• SDXL+Pixart sigma for increased prompt adherence, needs GGUF and ExtraModels for Comfyui
• Scheduling everything
• Multiple experimental modes.

Please share cool workflows you find under discussion, this stuff is really experimental and still needs discoveries.

Installation

1. Clone this repository into your ComfyUI's custom_nodes directory:

cd ComfyUI/custom_nodes
git clone https://github.com/Kantsche/comfyui-mixmod.git

2. Restart ComfyUI if it's already running.

Tips

• Different models excel at different aspects - try mixing a detail-focused model with a composition-focused one
• For FFT modes, base models work well for low frequencies, style models for high frequencies
• Experiment with weights and CFG values to find the best balance
• Schedule models to activate at different sampling steps for creative control

Compatibility

• Tested with SD 1.5, SDXL and Pixart Sigma
• Works with different model architectures (base, inpainting, etc.)
• Only tested on Windows

Example workflow with pixart sigma:

It improves the prompt adherence of sdxl in general prompts:

Example with Ponyv6 and NoobAI:

Only Pony, Pony+Noob, Only Noob image

If you want to support the development and some finetunes: https://ko-fi.com/lodestonerock

Credits

Created by Kantsche