-
-
Notifications
You must be signed in to change notification settings - Fork 439
Installation
Tip
These instructions assume that you already have Git and Python installed and available in your user PATH
.
If you do not have both Git and Python installed, follow the instructions in Install Python and Git to set those up first.
Start terminal
Launch your preferred system terminal and navigate to the directory where you want to install SD.Next.
- This should be a directory which your user account has read/write/execute access to.
- Installing SD.Next in a directory which requires admin permissions may cause it to not launch properly.
- Installing SD.Next with superuser/admin permissions is not recommended.
Clone SD.Next
Clone the repository by running following command in your desired location and then navigate into the cloned directory.
git clone https://github.com/vladmandic/automatic <optional directory name>
Important
Decide on appropriate compute backend for your system ahead of time as that will determine which libraries are installed on your system
--use-cuda Use nVidia CUDA backend (autodetected by default)
--use-rocm Use AMD ROCm backend (autodetected by default)
--use-ipex Use Intel OneAPI XPU backend (autodetected by default)
--use-openvino Use Intel OpenVINO backend
--use-zluda Use ZLUDA
--use-directml Use DirectML
Note
nVidia CUIDA, AMD ROCm and Intel OneAPI XPU are autodetected when available
all other compute backends require explicit selection on first startup
For platform specific information, check out
WSL | Intel Arc | DirectML | OpenVINO | ONNX & Olive | ZLUDA | AMD ROCm | MacOS | nVidia
Run the appropriate launcher for your OS to start the web interface:
- Windows:
webui.bat --debug --use-xxx
or.\webui.ps1 --debug --use-xxx
- Linux & Mac:
./webui.sh --debug --use-xxx
Now wait for few minutes to let the server install all required libraries.
The server is finished launching when the console shows an entry for "Startup time".
Tip
If you don't want to use built-in venv
support and prefer to run SD.Next in your own environment such as Docker container, Conda environment or any other virtual environment, you can skip venv
create/activate and launch SD.Next directly (command line flags noted above still apply):
python launch.py --debug
Tip
For the initial setup and future tech support, it is advisable to include the --debug
option which provides more detailed logging information.
Tip
All command line options can also be set via env variable
For example --debug
is same as set SD_DEBUG=true
Tip
For improved memory utilization on Linux, see Malloc
- Start the web interface
Once the web interface starts running, you can access it by opening your web browser and navigating to the address listed in the console next to "Local URL." For most users, this should behttp://localhost:7860/
.
You will see a brief loading screen, then you should be taken to theText
tab. - Adjust paths
You may want to adjust these settings in theSystem
:Settings
tab:- If you already have models, LoRAs, Embeddings, LyCORIS, etc. set your paths in the
System Paths
page now - Pay special attention to the
Folder with Huggingface models
andFolder for Huggingface Cache
as they can grow to significant size - You can use
Base path
to set a common root for all paths
- If you already have models, LoRAs, Embeddings, LyCORIS, etc. set your paths in the
- Set your desired look & feel
You can change the theme in theUser Interface
section. - Save your settings
If you changed any settings in the previous step, clickApply settings
to save those settings to your config file. This will also apply some defaults from built-in extensions. - Restart server
Click
Restart server
to re-launch the SD.Next server with the updated settings.
Note
SD.Next supports Python versions 3.9.x
up to 3.12.3
However, not all compute backends exist on Python 3.12 as they may be based on older torch
versions
Recommended version is latest service release of Python 3.11.x
Python versions higher than 3.12.3
are not supported
- Download Git for Windows from the following link: Git for Windows
- Run the downloaded
.exe
file and follow the installation wizard. - During the installation process, make sure to check the box for
"Use Git from the Windows Command line and also from 3rd-party-software" to add Git to your system's PATH. - Complete the installation by following the on-screen instructions.
- Download Python for Windows from the following link: Python for Windows
- Run the downloaded
.exe
file and follow the installation wizard. - On the "Customize Python" screen, make sure to check the box for "Add Python to PATH."
- Continue the installation by following the prompts.
- Once the installation is complete, you can open the command prompt and verify that Python is installed
by executingpython --version
andpip --version
to check the Python and Pip versions respectively.
- Download Git for macOS from the following link: Git for macOS
- Open the downloaded
.pkg
file and follow the installation instructions. - During the installation process, make sure to check the box for "Install Git Bash" to have a command-line Git interface.
- Complete the installation by following the prompts.
See these instructions for Python on MacOS (and an explanation why it's unique).