開源項目：AnimeGANv2

https://github.com/TachibanaYoshino/AnimeGANv2

可以將圖片和視頻轉換成漫畫風格的工具。采用的是神經風格遷移+生成對抗網絡（GAN）的組合.

測試圖片：（如有侵權，請聯系我刪除或修改）

生成漫畫風格

Requirements

• python 3.6

• tensorflow-gpu 1.15.0 (GPU 2080Ti, cuda 10.0.130, cudnn 7.6.0)

• opencv

• tqdm

• numpy

• glob

• argparse

• onnxruntime (If onnx file needs to be run.)

Usage

1. Inference

python test.py --checkpoint_dir checkpoint/generator_Hayao_weight --test_dir dataset/test/HR_photo --save_dir Hayao/HR_photo

2. Convert video to anime

python video2anime.py --video video/input/お花見.mp4 --checkpoint_dir checkpoint/generator_Hayao_weight --output video/output

3. Train

1. Download vgg19

vgg19.npy

2. Download Train/Val Photo dataset

Link

3. Do edge_smooth

python edge_smooth.py --dataset Hayao --img_size 256

4. Train

python train.py --dataset Hayao --epoch 101 --init_epoch 10

5. Extract the weights of the generator

python get_generator_ckpt.py --checkpoint_dir ../checkpoint/AnimeGANv2_Shinkai_lsgan_300_300_1_2_10_1 --style_name Shinkai