Pokemon Incubator

Train

1. Save training images to folder training-data

2. Run training script

python train.py --epochs=100 --batch-size=32 \
    --sample-interval=50 --load-saved=True --method=wgangp

Example Results

Credits

The code wgan and wgangp are taken and modified from Keras-GAN.

Why not fork?

It's a great repo to fork and start with, but there are too many GAN implementations that we are not going to use. So we copied the code we need and modified them for our training sets.

utils is taken and modified from Pokemon_GAM.

Why not fork?

Again, a great repo that taught us a lot, but we wanted to try different algorithms to train our data.