PCTResearch Wiki

You've loaded an old revision of the document! If you save it, you will create a new version with this data.

{{navi>navigationmenu}}
<HTML>
<style>
  .dokuwiki {
    font-size: 1.200em;  /*it was 0.875 in design.less*/
}
</style>
</HTML>

====== Stable Diffusion Version 2 - Run locally. Abandon if this doesn't work. ======

- [[stable-diffusion#Windows11 Installation|Windows11 Installation]]

===== Windows11 Installation =====

- [[stable-diffusion#Prerequisites|Prerequisites]]
    - [[stable-diffusion#git|git]]
    - [[stable-diffusion#python3.10.6|python3.10.6]]
    - [[stable-diffusion#Anaconda3|Anaconda3]]
  - [[stable-diffusion#Make conda environment|Make conda environment]]
  - [[stable-diffusion#Download & Installation|Download & Installation]]
    - [[stable-diffusion#Installing stablediffusion|Installing stablediffusion]]
    - [[stable-diffusion#Installing NVIDIA CUDA Toolkit|Installing NVIDIA CUDA Toolkit]]
    - [[stable-diffusion#Installing PyTorch|Installing PyTorch]]
    - [[stable-diffusion#Installing xFormers|Installing xFormers]]
    - [[stable-diffusion#Installing diffusers|Installing diffusers]]
    - [[stable-diffusion#about triton|about triton]]
  - [[stable-diffusion#Operation Confirmation|Operation Confirmation]]
    - [[stable-diffusion#scripts/txt2img.py|scripts/txt2img.py]]
    - [[stable-diffusion#scripts/img2img.py|scripts/img2img.py]]
  - [[stable-diffusion#Installing stable-diffusion-webui|Installing stable-diffusion-webui]]
  - [[stable-diffusion#Corresponding Windows System Environments|Corresponding Windows System Environments]]
  - [[stable-diffusion#Trouble shooting|Trouble shooting]]
    - [[stable-diffusion#Out of memory|Out of memory]]
    - [[stable-diffusion#venv folder|venv folder]]
    - [[stable-diffusion#Locate files in the right places|Locate files in the right places]]

==== Prerequisites ====

=== git ===
<code HTML5>
<HTML>
<table>
    <thead>
        <tr>
            <th>Official site</th>
            <th>Installation Guide</th>
        </tr>
    </thead>
    <tbody>
        <tr>
            <td><a href="https://gitforwindows.org/">Git for Windows</a></td>
            <td><a href="https://pctresearch.is-best.net/wiki/doku.php?id=Git-for-Windows-Installation"<tt>Git for Windows Installation</tt></a></td>
        </tr>
    </tbody>
    <tbody>
        <tr>
            <td><a href="https://tortoisegit.org/download/">TortoiseGit</a>(optional)</td>
            <td><a href="https://pctresearch.is-best.net/wiki/doku.php?id=tortoisegit-installation"<tt>TortoiseGit Installation</tt></a></td>
        </tr>
    </tbody>
</table>
</HTML>
</code>
=== python3.10.6 ===

python3.10.6 (3.10.0, 3.11.4 has also been confirmed to work)

[[https://www.python.org/|python.org]]

:!: python3.10.6 was recommended, nevertheless 2023/09/03 Operation has been confirmed in python3.11.4

[[python-installation|Python Installation]]

=== Anaconda3 ===

[[https://www.anaconda.com/download|Anaconda]]

[[anaconda-installation|Anaconda Installation]]

==== Make conda environment ====

Make conda environment for stable diffusion

:!: to activate conda environment, you must use ''Anaconda Prompt''.

Here we use environment name ''webui_env''.
<code Dos>
E:\> conda create --name webui_env python==3.10.6
</code>
Location of envs are
<code Dos>
C:\Users\%USERNAME%\anaconda3\envs
</code>

activate env
<code Dos>
E:\> conda activate webui_env
</code>
for reference
<code Dos>
(webui_env) E:> conda list
# packages in environment at C:\Users\kan\anaconda3\envs\webui_env:
#
# Name                    Version                   Build  Channel
bzip2                     1.0.8                he774522_0
ca-certificates           2023.05.30           haa95532_0
libffi                    3.4.4                hd77b12b_0
openssl                   1.1.1v               h2bbff1b_0
pip                       23.2.1          py310haa95532_0
python                    3.10.0               h96c0403_3
setuptools                68.0.0          py310haa95532_0
sqlite                    3.41.2               h2bbff1b_0
tk                        8.6.12               h2bbff1b_0
tzdata                    2023c                h04d1e81_0
vc                        14.2                 h21ff451_1
vs2015_runtime            14.27.29016          h5e58377_2
wheel                     0.38.4          py310haa95532_0
xz                        5.4.2                h8cc25b3_0
zlib                      1.2.13               h8cc25b3_0

(webui_env) E:\stable_diffusion>pip list
Package    Version
---------- -------
pip        23.2.1
setuptools 68.0.0
wheel      0.38.4
</code>
and maybe followings are needed.
<code Bash>
> conda config --append channels conda-forge
> conda config --append channels nvidia
</code>

==== download & installation ====

=== installing stablediffusion ===
installing stablediffusion using requirements.txt
<code Dos>
(webui_env) E:\> mkdir stable_diffusion
(webui_env) E:\> cd stable_diffusion
(webui_env) E:\stable_diffusion> git clone https://github.com/Stability-AI/stablediffusion.git
(webui_env) E:\stable_diffusion> cd stablediffusion
(webui_env) E:\stable_diffusion\stablediffusion> pip install -r requirements.txt
</code>

:!: if you got following error 
      ...
      running build_rust
      error: can't find Rust compiler
      ...
You need to Install Rust from
[[https://www.rust-lang.org/tools/install|rust-lang.org]]

=== Installing NVIDIA CUDA Toolkit ===

[[https://developer.nvidia.com/cuda-downloads|NVIDIA CUDA Toolkit downloads]]

maybe 11.8 is recommended

[[https://developer.nvidia.com/cuda-11-8-0-download-archive|CUDA Toolkit 11.8 Downloads]]

and execute following command
<code Dos>
(webui_env) E:\stable_diffusion\stablediffusion> conda install cudatoolkit
</code>

=== Installing PyTorch ===

There are two method.
  - Method 1: Installing PyTorch with Conda
  - Method 2: Installing PyTorch with Pip

Both methods are straightforward and can be done.

[[https://pytorch.org/get-started/locally|pytorch.org]] says Anaconda is our recommended package manager since it installs all dependencies.

But we choice pip install.

It says
<code Dos>
> pip3 install --pre torch torchvision torchaudio --index-url https://download.pytorch.org/whl/nightly/cu121
</code>
But we execute more precisely,
<code Bash>
(webui_env) E:\stable_diffusion\stablediffusion> pip install --pre torch==2.1.0.dev20230817+cu121 torchvision==0.16.0.dev20230818+cu121 torchaudio==2.1.0.dev20230818+cu121 --index-url https://download.pytorch.org/whl/nightly/cu121
</code>
:!: This is one of the most important checkpoints. At this point torch.cuda should be available. Check it!
<code Dos>
(webui_env) E:\stable_diffusion\stablediffusion> python
>>> import torch
>>> print(torch.__version__)
2.1.0.dev20230817+cu121
>>> torch.cuda.is_available()
True
>>> exit()
</code>
for reference
<code Dos>
(webui_env) E:\stable_diffusion\stablediffusion> conda list | findstr torch
open-clip-torch           2.7.0                    pypi_0    pypi
pytorch-lightning         1.4.2                    pypi_0    pypi
torch                     2.1.0.dev20230817+cu121          pypi_0    pypi
torchaudio                2.1.0.dev20230818+cu121          pypi_0    pypi
torchmetrics              0.6.0                    pypi_0    pypi
torchvision               0.16.0.dev20230818+cu121          pypi_0    pypi
</code>

=== Installing xFormers ===
:!: You don't need to do this, if you only use with webui.
<code Dos>
(webui_env) E:\stable_diffusion\stablediffusion> pip install git+https://github.com/facebookresearch/xformers
Successfully installed xformers-0.0.22+cfad52d.d2023090
</code>
Upon successful installation, the code will automatically default to memory efficient attention for the self- and cross-attention layers in the U-Net and autoencoder.

=== Installing diffusers ===
<code Dos>
(webui_env) E:\stable_diffusion\stablediffusion> pip install git+https://github.com/huggingface/transformers
      Successfully uninstalled transformers-4.19.2
Successfully installed safetensors-0.3.3 transformers-4.33.0.dev0
</code>

=== about triton ===

Set Windows environment ''XFORMERS_FORCE_DISABLE_TRITON=1''

I Installied from vcpkg but xformers can't recognize triton.

==== Operation Confirmation ====

=== scripts/txt2img.py ===

download ''v2-1_768-ema-pruned.ckpt''
from [[https://huggingface.co/stabilityai/stable-diffusion-2-1/blob/main/v2-1_768-ema-pruned.ckpt|stabilityai/stable-diffusion-2-1]]
<code Bash>
(webui_env) E:\stable_diffusion\stablediffusion> python scripts/txt2img.py --prompt "a professional photograph of an astronaut riding a horse" --ckpt "E:\difwork\SD2.1-v\v2-1_768-ema-pruned.ckpt" --config "E:\stable_diffusion\stablediffusion\configs\stable-diffusion\v2-inference-v.yaml" --H 768 --W 768 --device cuda
</code>
<note warning>
[WARNING]
txt2img.py makes 9 images. So it takes so much time and large memory. [[stable-diffusion#Out of memory|''Trouble shooting of Out of memory'']]. If you can't fix this error,skip this.
</note>

=== scripts/img2img.py ===

we need to download [[https://huggingface.co/stabilityai/stable-diffusion-2-1-base/blob/main/v2-1_512-ema-pruned.ckpt|SD2.1-base\v2-1_512-ema-pruned.ckpt]]
<code Bash>
(webui_env) E:\stable_diffusion\stablediffusion> python scripts/img2img.py --prompt "To create a commentary on modern technology and classical art, use an art style like Pop Art that often addresses cultural themes. Clearly mention the juxtaposition of elements. Pop Art painting of a modern smartphone with classic art pieces appearing on the screen." --init-img "E:\difwork\SD2.1-base\sample.png" --strength 0.8 --ckpt "E:\difwork\SD2.1-base\v2-1_512-ema-pruned.ckpt"
outputs/img2img-samples
Enjoy.
</code>

==== installing stable-diffusion-webui ====

[[https://github.com/AUTOMATIC1111/stable-diffusion-webui|stable-diffusion-webui]]

<code Bash>
(webui_env) E:\stable_diffusion\stablediffusion> cd ..
(webui_env) E:\stable_diffusion> git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
(webui_env) E:\stable_diffusion> cd stable-diffusion-webui
</code>

Edit the webui-user.bat to include ''--xformers'' (''--reinstall-xformers'') for the command line arguments as well as
the desired package version. Then run the ''webui-user.bat'' as normal.

''webui-user.bat'':
<code Bash>
@echo off

set PYTHON=
set GIT=
set VENV_DIR=
set XFORMERS_PACKAGE=xformers==0.0.20
set COMMANDLINE_ARGS=--xformers

call webui.bat
</code>
<note important>
[IMPORTANT]
We need to install xformers for webui.And It should be 0.0.20 not 0.0.22 at current condition.
</note>

then execute ''webui-user.bat''
<code Bash>
(webui_env) E:\stable_diffusion\stable-diffusion-webui> webui-user.bat
              Running on local URL:  http://127.0.0.1:7860
Applying attention optimization: xformers... done.
...
</code>
** Gooooal!!! **

for reference
<code Bash>
(webui_env) E:\webui\stable-diffusion-webui>python
>>> import torch
>>> print(torch.__version__)
2.1.0.dev20230817+cu121
>>> torch.cuda.is_available()
True
>>> exit()
</code>
<code Bash>
(webui_env) E:\webui\stable-diffusion-webui>conda list | findstr torch
open-clip-torch           2.7.0                    pypi_0    pypi
pytorch-lightning         1.4.2                    pypi_0    pypi
torch                     2.1.0.dev20230817+cu121          pypi_0    pypi
torchaudio                2.1.0.dev20230818+cu121          pypi_0    pypi
torchmetrics              0.6.0                    pypi_0    pypi
torchvision               0.16.0.dev20230818+cu121          pypi_0    pypi
</code>

==== Corresponding Windows System Environment ====

followings are checked by programs
<code Bash>
CUDA_HOME = C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2
CUDA_PATH = C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8
PYTORCH_CUDA_ALLOC_CONF = garbage_collection_threshold:0.6,max_split_size_mb:4096
PYTORCH_NO_CUDA_MEMORY_CACHING = 1
XFORMERS_FORCE_DISABLE_TRITON = 1
</code>
maybe not correnponding
<code Bash>
CUDA_PATH_V11_8 = C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.8
CUDA_PATH_V12_2 = C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.2
</code>

==== Trouble shooting ====

=== Out of memory ===

If you got error like following,
<code Bash>
torch.cuda.OutOfMemoryError: CUDA out of memory.
</code>
edit windows system environment variables
<code Bash>
set PYTORCH_NO_CUDA_MEMORY_CACHING=1

set PYTORCH_CUDA_ALLOC_CONF=garbage_collection_threshold:0.6,max_split_size_mb:32
set PYTORCH_CUDA_ALLOC_CONF=garbage_collection_threshold:0.6,max_split_size_mb:1024
set PYTORCH_CUDA_ALLOC_CONF=garbage_collection_threshold:0.6,max_split_size_mb:4096
</code>

<note>
About ''max_split_size_mb''

''max_split_size_mb'' prevents the native allocator from splitting blocks larger than this size (in MB).

Default value is unlimited.So easily cause Out of Memory.

So,set ''max_split_size_mb'' to a maximum integer value smaller than the minimum size of the video memory request when Out of Memory occurs.

That can ensure the feasibility of running large images while maximizing performance.
</note>

==== venv folder ====

If you change some environments and use the same directory, you should delete the ''venv'' folder completely.
<code Bash>
stable-diffusion-webui\venv
</code>
webui will create new venv again.

=== Locate files in the right places ===

''.ckpt'', ''.safetensors'', ''.pt''', ''.bin'', ''.pth''

There are many types of extension files that correspond to webui.

You must place these files in the correct locations.

Otherwise, it will cause many warnings or even errors.

See [[https://github.com/jianwu13/PCTResearch/wiki/Webui-Tips#webui.folder|Webui-Tips]]

Edit summary

Note: By editing this page you agree to license your content under the following license: CC Attribution-Share Alike 4.0 International