CartoonGAN

a simple PyTorch implementation of CartoonGAN (CVPR 2018)

Usage

1.prepare dataset

The folder structure is as follows

├── data
│   ├── train # train real images
│   ├── test  # test real images 
│   ├── ShinKai # cartoon images
│   |── ...
│
├── outputs # generated images
│
├── main.py # training code
├── utils.py
├── config.py # some configs

2.preprocess cartoon images

python utils.py your_cartoon_image

3.train

python main.py --cartoon_name your_cartoon_image # yon can see more arguments in config.py

Resutls

1.

input

ufotable

Kyoto Animation

2.

input

ufotable

Kyoto Animation

3.

input

ufotable

Kyoto Animation

4.

input

ufotable

Kyoto Animation

Reference

CartoonGAN-CVPR2018

pytorch-CartoonGAN by znxlwm