padeyao4/AnimeGANv3-Python

A command-line tool that transforms photos into an anime look using ONNX Runtime trained models of AnimeGANv3.

AnimeGANv3-Python

A command-line tool that transforms photos into an anime look using ONNX Runtime trained models of AnimeGANv3.
Based on the inference process of AnimeGANv3.exe.

Install

It is written in Python, so it is cross-platform.
However, it consumes a large amount of CPU and memory because inference is performed on the CPU.

Requirements: Python 3.10 / pip / pipenv

# clone code
git clone https://github.com/tsukumijima/AnimeGANv3-Python.git
cd AnimeGANv3-Python

# install pipenv
pip install pipenv

# run pipenv sync
## Windows (PowerShell)
$env:PIPENV_VENV_IN_PROJECT="true"; pipenv sync
## Linux
PIPENV_VENV_IN_PROJECT="true" pipenv sync

Usage

# run AnimeGANv3-Python.py inside pipenv
pipenv run python AnimeGANv3-Python.py C:/path/to/input_images C:/path/to/output_images --onnx-model-type H40

usage: AnimeGANv3-Python.py [-h] [--onnx-model-type {H40,H50,H64}] InputDirPath OutputDirPath

positional arguments:
  InputDirPath          Image directory path of input source
  OutputDirPath         Image directory path of output destination

options:
  -h, --help            show this help message and exit
  --onnx-model-type {H40,H50,H64}
                        onnx model type (H40, H50, H64)

docker run -v $pwd/inputs:/inputs -v $pwd/outputs:/outputs guojiank/animeganv3

Examples

⬅ original photo | transform into anime look ➡

License

MIT License