https://github.com/k-l-lambda/diffusers-web
A web front-end wrapper of HuggingFace/Diffusers
https://github.com/k-l-lambda/diffusers-web
Last synced: 4 months ago
JSON representation
A web front-end wrapper of HuggingFace/Diffusers
- Host: GitHub
- URL: https://github.com/k-l-lambda/diffusers-web
- Owner: k-l-lambda
- Created: 2022-09-09T08:15:18.000Z (almost 4 years ago)
- Default Branch: master
- Last Pushed: 2025-04-27T11:42:27.000Z (about 1 year ago)
- Last Synced: 2025-04-27T12:39:02.006Z (about 1 year ago)
- Language: Python
- Homepage:
- Size: 22.1 MB
- Stars: 1
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- Changelog: changelog.txt
Awesome Lists containing this project
README
# Diffusers Web
This project is a web based image generation tool, a front-end wrapper for [HuggingFace's diffusers](https://github.com/huggingface/diffusers).
## Usage
Run the web server:
```.bash
python ./main.py
```
If this works, navigate to *http://localhost:8157* in your browser.
## Features
### Text to image
### Image transform
Navigate to *http://localhost:8157/converter*.
### Inpainting and outpainting
Before run web server, config the diffuser model to an inpainting model. E.g. add an environment variable in `.env.local`:
```
DIFFUSER_MODEL_PATH=runwayml/stable-diffusion-inpainting
```
Then navigate to *http://localhost:8157/painter*.
Tips: image copy in a web page other than *localhost* requires *https* protocol, config `SSL_CONTEXT="'adhoc'"` in `.env.local` to achieve this.
## Requirements
### Hardware
A GPU with CUDA capability is required for good performence. The more video memory the better, 24GB is recommended.
### Python Libraries
Install dependency libraries with pip, reference to [requirements.txt](./requirements.txt).
### Authorization
You may need a [HuggingFace access token](https://huggingface.co/settings/tokens) to download pretrained models for the first running.
## Local Environment Variables
Edit local environment variables in file *.env.local*.
Variable Name | Default Value | Description
:-- | :-- | :--
**HF_TOKEN** | | Your HuggingFace access token. If a local config path provided, this can be ignored.
**DIFFUSER_MODEL_PATH** | stabilityai/stable-diffusion-2 | This can be a local model config path.
**TEXTGEN_MODEL_PATH** | k-l-lambda/clip-text-generator | The random painting description generator model path. This can be a local model config path.
**HTTP_HOST** | 127.0.0.1 | Use `0.0.0.0` for network access.
**HTTP_PORT** | 8157 |
**SSL_CONTEXT** | None | Use `'adhoc'` for https server.
**DEVICE** | | `cuda` or *None*
**TEXT_DEVICE_INDEX** | 0 | This can be greater than 0 if **CUDA_VISIBLE_DEVICES** has more than 1 gpu specified.