It's an out-of-the-box animegan2 that is modified from the original version.
So how to run it?
- pip install -r requirements.txt
- put the origin pictures into the samples/inputs (please notice the max shape of the picture is 600*800 in 24G GPU.)
- python3 test.py --checkpoint ./weights/face_paint_512_v2.pt
- get the converted pictures from the samples/results
More details can be seen the blog.