Stable Diffusion WebUI Forge - Classic

Stable Diffusion WebUI Forge is a platform on top of the original Stable Diffusion WebUI by AUTOMATIC1111, to make development easier, optimize resource management, speed up inference, and study experimental features.
The name "Forge" is inspired by "Minecraft Forge". This project aims to become the Forge of Stable Diffusion WebUI.

- lllyasviel
^{(paraphrased)}

"Classic" mainly serves as an archive for the "previous" version of Forge, which was built on Gradio 3.41.2 before the major changes (see the original announcement) were introduced. Additionally, this fork is focused exclusively on SD1 and SDXL checkpoints, having various optimizations implemented, with the main goal of being the lightest WebUI without any bloatwares.

How to Install

Features [Apr. 30]

Most base features of the original Automatic1111 Webui should still function

New Features

Support v-pred SDXL checkpoints (eg. NoobAI)
Support uv package manager
- requires uv
- drastically speed up installation
- see Commandline
Support SageAttention
- requires manually installing the triton package
  - how to install
- requires RTX 30 +
- ~10% speed up
- see Commandline
Support FlashAttention
- requires manually installing the flash-attn package
  - how to install
- ~10% speed up
Support fast fp16_accumulation
- requires PyTorch 2.7.0 +
- ~25% speed up
- see Commandline
Support fast cublas operation (CublasLinear)
- requires manually installing the cublas_ops package
  - how to install
- ~25% speed up
- enable in Settings
Support fast fp8 operation (torch._scaled_mm)
- requires RTX 40 +
- ~10% speed up; reduce quality
- enable in Settings

The fp16_accumulation and cublas operation achieve the same speed up; if you already install/update to torch==2.7.0, you do not need to go for cublas_ops

The fp16_accumulation and cublas operation require fp16 precision, thus is not compatible with the fp8 operation

Implement RescaleCFG
- reduce burnt colors; mainly for v-pred checkpoints
Implement MaHiRo
- alternative CFG calculation
- graph
Implement diskcache
- (backported from Automatic1111 Webui upstream)
Implement skip_early_cond
- (backported from Automatic1111 Webui upstream)
Update spandrel
- support most modern Upscaler architecture
Add pillow-heif package
- support .avif and .heif formats
Automatic row split for X/Y/Z Plot
Add an option to disable Refiner
Add an option to disable ExtraNetworks Tree View
Support Union / ProMax ControlNet
- I just made them always show up in the dropdown

Removed Features

SD2
Alt-Diffusion
Instruct-Pix2Pix
Hypernetworks
SVD
Z123
CLIP Interrogator
Deepbooru Interrogator
Textual Inversion Training
Checkpoint Merging
LDSR
Most built-in Extensions
Some built-in Scripts
The test scripts
Photopea and openpose_editor (ControlNet)
Unix .sh launch scripts
- You can still use this WebUI by copying a launch script from another working WebUI; I just don't want to maintain them...

Optimizations

[Freedom] Natively integrate the SD1 and SDXL logics
- no longer git clone any repository on fresh install
- no more random hacks and monkey patches
Fix memory leak when switching checkpoints
Clean up the ldm_patched (ie. comfy) folder
Remove unused cmd_args
Remove unused shared_options
Remove unused args_parser
Remove legacy codes
Remove duplicated upscaler codes
- put every upscaler inside the ESRGAN folder
Improve color correction
Improve code logics
Improve hash caching
Improve error logs
- no longer prints TypeError: 'NoneType' object is not iterable
Improve setting descriptions
Check for Extension updates in parallel
Moved embeddings folder into models folder
ControlNet Rewrite
- change Units to gr.Tab
- remove multi-inputs, as they are "misleading"
- change visible toggle to interactive toggle; now the UI will no longer jump around
- improved Presets application
Run text encoder on CPU by default
Fix pydantic Errors
Fix Soft Inpainting
Lint & Format most of the Python and JavaScript codes
Update to Pillow 11
- faster image processing
Update protobuf
- faster insightface loading
Update to latest PyTorch
- torch==2.7.0+cu128
- xformers==0.0.30
No longer install open-clip twice
Update certain packages to newer versions
Update recommended Python to 3.11.9
many more... :tm:

Commandline

These flags can be added after the set COMMANDLINE_ARGS= line in the webui-user.bat (separate each flag with space)

A1111 built-in

--no-download-sd-model: Do not download a default checkpoint
- can be removed after you download some checkpoints of your choice
--xformers: Install the xformers package to speed up generation
- Currently, torch==2.7.0 does not support xformers yet
--port: Specify a server port to use
- defaults to 7860
--api: Enable API access

Once you have successfully launched the WebUI, you can add the following flags to bypass some validation steps in order to improve the Startup time
- --skip-prepare-environment
- --skip-install
- --skip-python-version-check
- --skip-torch-cuda-test
- --skip-version-check

Remove them if you are installing an Extension, as those also block Extension from installing requirements

by. Forge

For RTX 30 and above, you can add the following flags to slightly increase the performance; but in rare occurrences, they may cause OutOfMemory errors or even crash the WebUI; and in certain configurations, they may even lower the speed instead
- --cuda-malloc
- --cuda-stream
- --pin-shared-memory

by. Classic

--uv: Replace the python -m pip calls with uv pip to massively speed up package installation
- requires uv to be installed first (see Installation)
--uv-symlink: Same as above; but additionally pass --link-mode symlink to the commands
- significantly reduces installation size (~7 GB to ~100 MB)

Using symlink means it will directly access the packages from the cache folders; refrain from clearing the cache when setting this option

--fast-fp16: Enable the allow_fp16_accumulation option
- requires PyTorch 2.7.0 +
--sage: Install the sageattention package to speed up generation
- requires triton
- requires RTX 30 +
- only affects SDXL

--xformers is still recommended even if you already have --sage, as sageattention does not speed up VAE while xformers does

--model-ref: Points to a central models folder that contains all your models
- said folder should contain subfolders like Stable-diffusion, Lora, VAE, ESRGAN, etc.

This simply replaces the models folder, rather than adding on top of it

Installation

Install git

Clone the Repo

git clone https://github.com/Haoming02/sd-webui-forge-classic

Setup Python

Recommended Method

Install uv

Set up venv

cd sd-webui-forge-classic
uv venv venv --python 3.11 --seed

Add the --uv flag to webui-user.bat

Standard Method

Install Python 3.11.9
- Remember to enable Add Python to PATH

(Optional) Configure Commandline
Launch the WebUI via webui-user.bat
During the first launch, it will automatically install all the requirements
Once the installation is finished, the WebUI will start in a browser automatically

Install cublas

Expand

Ensure the WebUI can properly launch already, by following the installation steps first
Open the console in the WebUI directory
```
cd sd-webui-forge-classic
```
Start the virtual environment
```
venv\scripts\activate
```
Create a new folder
```
mkdir repo
cd repo
```

Clone the repo

git clone https://github.com/aredden/torch-cublas-hgemm
cd torch-cublas-hgemm

Install the library
```
pip install -e . --no-build-isolation
```
- If you installed uv, use uv pip install instead
- The installation takes a few minutes

Install triton

Expand

Ensure the WebUI can properly launch already, by following the installation steps first
Open the console in the WebUI directory
```
cd sd-webui-forge-classic
```
Start the virtual environment
```
venv\scripts\activate
```
Install the library
- Windows
```
pip install triton-windows
```
- Linux
```
pip install triton
```
- If you installed uv, use uv pip install instead

Install flash-attn

Expand

Ensure the WebUI can properly launch already, by following the installation steps first
Open the console in the WebUI directory
```
cd sd-webui-forge-classic
```
Start the virtual environment
```
venv\scripts\activate
```
Install the library
- Windows
  - Download the pre-built .whl package from https://github.com/kingbri1/flash-attention/releases
```
pip install flash_attn...win...whl
```
- Linux
  - Download the pre-built .whl package from https://github.com/Dao-AILab/flash-attention/releases
```
pip install flash_attn...linux...whl
```
- If you installed uv, use uv pip install instead
- Important: Download the correct .whl for your Python and PyTorch version

Install sageattention 2

If you only use SDXL, then 1.x is already enough; 2.x simply has partial support for SD1 checkpoints

Expand

Ensure the WebUI can properly launch already, by following the installation steps first
Open the console in the WebUI directory
```
cd sd-webui-forge-classic
```
Start the virtual environment
```
venv\scripts\activate
```
Create a new folder
```
mkdir repo
cd repo
```

Clone the repo

git clone https://github.com/thu-ml/SageAttention
cd SageAttention

Install the library
```
pip install -e . --no-build-isolation
```
- If you installed uv, use uv pip install instead
- The installation takes a few minutes

Install older PyTorch

Read this if your GPU does not support the latest PyTorch

Expand

Navigate to the WebUI directory
Edit the webui-user.bat file
Add a new line to specify an older version:

set TORCH_COMMAND=pip install torch==2.1.2 torchvision==0.16.2 --extra-index-url https://download.pytorch.org/whl/cu121

Attention

The --xformers and --sage args are only responsible for installing the packages, not whether its respective attention is used; This also means you can remove them once they are successfully installed

Forge Classic tries to import the packages and automatically choose the first available attention function in the following order:

SageAttention
FlashAttention
xformers
PyTorch
Basic

The VAE only checks for xformers

In my experience, the speed of each attention function for SDXL is ranked in the following order:

SageAttention ≥ FlashAttention > xformers > PyTorch >> Basic

SageAttention is based on quantization, so its quality might be slightly worse than others

Issues & Requests

Issues about removed features will simply be ignored
Issues regarding installation will be ignored if it's obviously user-error
Feature Request not related to performance or optimization will simply be ignored
- For cutting edge features, check out reForge instead
- Non-Windows platforms will not be supported, as I cannot verify nor maintain them

Special thanks to AUTOMATIC1111, lllyasviel, and comfyanonymous, kijai,
along with the rest of the contributors,
for their invaluable efforts in the open-source image generation community