Awesome
PGGAN-tensorflow
The tensorflow implementation of PROGRESSIVE GROWING OF GANS FOR IMPROVED QUALITY, STABILITY, AND VARIATION.
The generative process of PG-GAN
<p align="center"> <img src="/images/figure.png"> </p>Differences with the original paper.
- Recently, just generate 64x64 and 128x128 pixels samples.
Setup
Prerequisites
- TensorFlow >= 1.4
- python 2.7 or 3
Getting Started
- Clone this repo:
git clone https://github.com/zhangqianhui/progressive_growing_of_gans_tensorflow.git
cd progressive_growing_of_gans_tensorflow
- Download the CelebA dataset
You can download the CelebA dataset and unzip CelebA into a directory. Noted that this directory don't contain the sub-directory.
-
The method for creating CelebA-HQ can be found on Method
-
Train the model on CelebA dataset
python main.py --path=your celeba data-path --celeba=True
- Train the model on CelebA-HQ dataset
python main.py --path=your celeba-hq data-path --celeba=False
Results on celebA dataset
Here is the generated 64x64 results(Left: generated; Right: Real):
<p align="center"> <img src="/images/sample.png"> </p>Here is the generated 128x128 results(Left: generated; Right: Real):
<p align="center"> <img src="/images/sample_128.png"> </p>Results on CelebA-HQ dataset
Here is the generated 64x64 results(Left: Real; Right: Generated):
<p align="center"> <img src="/images/hs_sample_64.jpg"> </p>Here is the generated 128x128 results(Left: Real; Right: Generated):
<p align="center"> <img src="/images/hs_sample_128.jpg"> </p>Issue
If you find some bugs, Thanks for your issue to propose it.