https://github.com/woctezuma/steam-stylegan
Train a StyleGAN model on Colaboratory to generate Steam banners.
https://github.com/woctezuma/steam-stylegan
colab colab-notebook colaboratory gan generative-adversarial-network google-colab google-colab-notebook google-colaboratory steam steam-api steam-data steam-game steam-games steam-gan steam-pics steam-store style-gan stylegan stylegan-model
Last synced: 22 days ago
JSON representation
Train a StyleGAN model on Colaboratory to generate Steam banners.
- Host: GitHub
- URL: https://github.com/woctezuma/steam-stylegan
- Owner: woctezuma
- Created: 2019-06-01T21:22:05.000Z (over 6 years ago)
- Default Branch: master
- Last Pushed: 2020-10-10T12:18:16.000Z (about 5 years ago)
- Last Synced: 2025-08-02T00:45:08.680Z (3 months ago)
- Topics: colab, colab-notebook, colaboratory, gan, generative-adversarial-network, google-colab, google-colab-notebook, google-colaboratory, steam, steam-api, steam-data, steam-game, steam-games, steam-gan, steam-pics, steam-store, style-gan, stylegan, stylegan-model
- Language: Jupyter Notebook
- Homepage: https://colab.research.google.com/
- Size: 17.6 KB
- Stars: 5
- Watchers: 2
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Steam StyleGAN
The goal of this [Google Colab](https://colab.research.google.com/) notebook is to capture the distribution of Steam banners and sample with a StyleGAN.
## Usage
- Acquire the data, e.g. as a snapshot called `128x128.zip` in [another of my repositories](https://github.com/woctezuma/download-steam-banners-data),
- Follow the instructions to edit `train.py` in the [official StyleGAN Github repository](https://github.com/NVlabs/stylegan),
- Run [`StyleGAN.ipynb`][StyleGAN] to train a StyleGAN.
[![Open In Colab][colab-badge]][StyleGAN]
- To resume training from a checkpoint, you will have to edit `training/training_loop.py`.
NB: You might have to edit `metrics/frechet_inception_distance.py` to retrieve the network `inception_v3_features.pkl` locally if it cannot be downloaded from Google Colab.
## Results
The dataset consists of 31,723 Steam banners with RGB channels and resized from 460x215 to 128x128 resolution.
Pre-processed data, as `.tfrecords` files, can be downloaded from [Google Drive](https://drive.google.com/open?id=1CZxtfwbCmrDqIlSvi_3BTxtaLAAIRp-o).
A StyleGAN model was trained on 3,524,000 images, with a decreasing mini-batch size, which is about 111 epochs.
A checkpoint of the network can be downloaded from [Google Drive](https://drive.google.com/open?id=1BQr7lFiHkx_WFmiyqIcd1m6XAFJNZFOh).
Caveat: training was manually stopped after roughly 1 day, using 1 Tesla K80 GPU in the cloud.
Based on the [expected training times](https://github.com/NVlabs/stylegan#training-networks) for 1024x1024, 512x512 and 256x256 images, 9 days of computation time might be required to get the best results for 128x128 images.
### Generated Steam banners
Results obtained with different numbers of images seen during training are shown [on the Wiki](https://github.com/woctezuma/steam-stylegan/wiki).
A grid of generated Steam banners after 3,524 kimg:
### Real Steam banners
A grid of real Steam banners:
## References
- StyleGAN2:
- [StyleGAN2](https://github.com/NVlabs/stylegan2)
- [Steam-StyleGAN2](https://github.com/woctezuma/steam-stylegan2)
- StyleGAN:
- [StyleGAN1](https://github.com/NVlabs/stylegan)
- DCGAN:
- [Steam-DCGAN](https://github.com/woctezuma/google-colab)
[StyleGAN]:
[colab-badge]: