Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/lightdiffusion/lightdiffusion
Fastest Stable Diffusion GUI, Pipeline, with only one python script, with the least number of lines and in the least complex way.
https://github.com/lightdiffusion/lightdiffusion
ai artificial-intelligence deep-learning diffusion generative-ai gui image-generation pytorch stable-diffusion text2image torch txt2img
Last synced: 8 days ago
JSON representation
Fastest Stable Diffusion GUI, Pipeline, with only one python script, with the least number of lines and in the least complex way.
- Host: GitHub
- URL: https://github.com/lightdiffusion/lightdiffusion
- Owner: LightDiffusion
- License: gpl-3.0
- Created: 2024-12-13T07:53:47.000Z (9 days ago)
- Default Branch: main
- Last Pushed: 2024-12-13T07:55:10.000Z (9 days ago)
- Last Synced: 2024-12-13T09:31:01.653Z (9 days ago)
- Topics: ai, artificial-intelligence, deep-learning, diffusion, generative-ai, gui, image-generation, pytorch, stable-diffusion, text2image, torch, txt2img
- Language: Python
- Homepage: https://aatrick.github.io/LightDiffusion/
- Size: 33.8 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# LightDiffusion
The purpose of this repository and project is to remake the famous **stable-diffusion** in only **one python script**, with the
**least number of lines** and in the **least complex way**, while being the **fastest diffusion algorithm**. It's made by retro-engineering Stable-Diffusion, ComfyUI and
SDWebUI.
There's also a LightDiffusion version using Flux Dev available in releases and Flux branch.
For more information, check the [documentation](https://aatrick.github.io/LightDiffusion/)
The project grew up with time, leaving the purpose of only being the fastest and simplest SD1.5 image generator, if you seek the **original version of about 3000 lines of python**, you can still find it at [LightDiffusion-og](https://github.com/Aatrick/LightDiffusion-og) or in the releases [here](https://github.com/Aatrick/LightDiffusion/releases/tag/V1.0).
Also check out the **Refactored version for improved simplicity and assembling SD and Flux** at [LightDiffusion-Next](https://github.com/Aatrick/LightDiffusion-Next)
## Features
- Original **Txt2Img**, **Img2Img** (R-ERSGAN4x+ UltimateSDUpscaling DPM++ 2M)
- **Attention** syntax
- **Hires-Fix** (euler ancestral normal)
- Nvidia, amd and Intel gpus and cpu
- **Xformers and Pytorch** optimization
- **Stable-Fast** implementation offering a 70% speedup at the cost of pre inference model optimization windup time
- **FP16 and FP32** precision support
- **Saved state** in between starts
- **GUI**
- DPM Adaptive Karras
- **Clip Skip**
- **LoRa and textual inversion** (embeddings) support
- **Automatic Prompt-Enhancing** with llama3.1 (ollama)
- **Discord bot** integration with the installation of [Boubou](https://github.com/Aatrick/Boubou) and the usage of LightDiffusion's pipeline
- **Automatic Face and body Detailer** (based on [Impact pack](https://github.com/ltdrdata/ComfyUI-Impact-Pack))
- **TAESD** image previewing at the cost of a 10% performance hit
## Installation
To install, please download a release or clone this repo and execute the run.bat file in a cmd and you should be good to go.
#### From Source
Else install the python dependencies by writing `pip install -r requirements.txt`
After doing that, add your SD1/1.5 safetensors model to the checkpoints directory and you should be good to go.
### Stable-Fast
To use the stable-fast optimization refer to this [guide](https://github.com/chengzeyi/stable-fast?tab=readme-ov-file#installation) to install from the wheel file.
### Ollama
To use the Prompt enhancer refer to this [guide](https://github.com/ollama/ollama?tab=readme-ov-file) to install and run those commands
`ollama run llama3.1`
`pip install ollama`
### Discord
To use LightDiffusion in discord, refer to the installation guide in this [repo](https://github.com/Aatrick/Boubou)