ogkalu2/Merge-Stable-Diffusion-models-without-distortion
Adaptation of the merging method described in the paper - Git Re-Basin: Merging Models modulo Permutation Symmetries (https://arxiv.org/abs/2209.04836) for Stable Diffusion
This tool helps AI artists and Stable Diffusion enthusiasts combine two different Stable Diffusion models to create a new, blended model. You provide two existing model files, and it produces a single, merged model that combines their characteristics without distorting the outputs. This allows you to achieve unique artistic styles or capabilities that weren't present in either original model.
148 stars. No commits in the last 6 months.
Use this if you want to creatively blend the strengths of two Stable Diffusion models into a new one, ensuring the merged model generates high-quality images without visual artifacts often caused by simpler merging methods.
Not ideal if you need to merge models of vastly different architectures (beyond minor version differences) or if you are looking for advanced features like pruning or specific CLIP fixes, as these are not fully supported.
Stars
148
Forks
22
Language
Python
License
MIT
Category
Last pushed
Apr 30, 2024
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/ogkalu2/Merge-Stable-Diffusion-models-without-distortion"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
siliconflow/onediff
OneDiff: An out-of-the-box acceleration library for diffusion models.
wooyeolbaek/attention-map-diffusers
🚀 Cross attention map tools for huggingface/diffusers
jina-ai/discoart
🪩 Create Disco Diffusion artworks in one line
chengzeyi/stable-fast
https://wavespeed.ai/ Best inference performance optimization framework for HuggingFace...
hkproj/pytorch-stable-diffusion
Stable Diffusion implemented from scratch in PyTorch