AnimeGAN is an open-source neural network that converts images and videos into an anime look using GAN (generative adversarial networks). The project is hosted on GitHub and is built with PyTorch.
What it does
AnimeGAN is designed for anime-style stylization of visual content. The code has been public since 2019, and an improved release, AnimeGANv2, was introduced in 2021.
- Turn photos into anime-style illustrations
- Stylize video frames into an anime look
- Use styles inspired by well-known directors
- Generate outputs at different resolutions
- Run a lightweight GAN architecture
AnimeGANv2 reduces the generator size to 8.17 MB, which helps make it one of the faster options for image processing.
How to use
There’s no official website or web app. You run AnimeGAN locally via the GitHub code.
- Install Python 3.6+
- Clone/download the GitHub repository
- Install PyTorch and required dependencies
- Add test images or videos
- Run the provided test script (test.py) and set file paths
AnimeGAN is free for non-commercial use and includes English-language documentation.
