Basically, I am working on Linux. I sometimes check to make sure that this code works in Windows.
I am not familiar with Windows or Anaconda. If you are familiar with Anaconda, do it your way.
Install Python 3.10 from Windows Store.
Run Installer for Windows.bat
.
If you want to reinstall, delete venv
folder and run it again.
If you have Anaconda installed, make sure you have added python
to PATH.
If python
command opens Windows Store, see https://stackoverflow.com/questions/58754860/cmd-opens-windows-store-when-i-type-python.
Run Run waifu2x GUI.bat
.
Run Run waifu2x Web.bat
.
Run Open Prompt.bat
.
NOTE: ImageMagick(wand) is only required for training and benchmark.
See Install ImageMagick on Windows.
NOTE: libraqm is only required for synthetic training data generation.
Download libraqm‑0.7.1.dll.zip
from https://www.lfd.uci.edu/~gohlke/pythonlibs/#pillow .
git clone https://github.com/nagadomi/nunif.git
cd nunif
If you want to use the dev
branch, execute the following command.
git clone https://github.com/nagadomi/nunif.git -b dev
or
git fetch --all
git checkout -b dev origin/dev
python -m venv venv
.\venv\Scripts\activate
pip install -r requirements-torch.txt
pip install -r requirements.txt
If you want to use GUI,
pip install -r requirements-gui.txt
Download pre-trained models.
python -m waifu2x.download_models
Generate waifu2x/web/public_html
python -m waifu2x.web.webgen
Start the web server.
python -m waifu2x.web
If you got ImportError: cannot import name '_imagingcms' from 'PIL'
error, upgrade the pillow package.
pip install --upgrade pillow
This seems to be a problem with the pillow that is installed by conda by default.
See also waifu2x README.md.