Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/AgaMiko/pixel_character_generator
Generating retro pixel game characters with Generative Adversarial Networks. Dataset "TinyHero" included.
https://github.com/AgaMiko/pixel_character_generator
dataset dataset-generation dcgan-pytorch deep-learning generative-adversarial-network
Last synced: 3 days ago
JSON representation
Generating retro pixel game characters with Generative Adversarial Networks. Dataset "TinyHero" included.
- Host: GitHub
- URL: https://github.com/AgaMiko/pixel_character_generator
- Owner: AgaMiko
- Created: 2020-04-28T10:01:08.000Z (over 4 years ago)
- Default Branch: master
- Last Pushed: 2020-05-24T12:21:40.000Z (over 4 years ago)
- Last Synced: 2024-11-01T02:21:37.606Z (8 days ago)
- Topics: dataset, dataset-generation, dcgan-pytorch, deep-learning, generative-adversarial-network
- Language: Jupyter Notebook
- Homepage:
- Size: 20.3 MB
- Stars: 121
- Watchers: 4
- Forks: 10
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
- awesome-generative-ai - AgaMiko/pixel_character_generator
README
# pixel_character_generator
Generating retro pixel game characters with Generative Adversarial Networks. Dataset "TinyHero" included.
   
   
   
   
   
   
# Dataset TinyHero
Dataset TinyHero includes 64x64 retro-pixel character. All characters were generated with [Universal LPC spritesheet by makrohn](https://github.com/makrohn/Universal-LPC-spritesheet/tree/7040e2fe85d2cb1e8154ec5fce382589d369bdb8). Each character in the dataset was randomly generated including: sex, body type, skin color and equipment with LPC spritesheet with 4 different angles view.
| Image sixe | Dataset size | Source | Download |
|------------|---------------|-----------------|----------|
| 64x64 | 3648 images | [LPC Spritesheet](https://github.com/makrohn/Universal-LPC-spritesheet/tree/7040e2fe85d2cb1e8154ec5fce382589d369bdb8) | [data.zip](data.zip)|
| | 912 per class | | |
According to the rules of the LPC all art submissions were dual licensed under both GNU GPL 3.0 and CC-BY-SA 3.0. Further work produced in this repository is licensed under the same terms.
CC-BY-SA 3.0:
http://creativecommons.org/licenses/by-sa/3.0/
See the file: cc-by-sa-3.0.txt
GNU GPL 3.0:
http://www.gnu.org/licenses/gpl-3.0.html
See the file: gpl-3.0.txt
   
   
# Pixel Character Generator - DCGAN
Based on the DCGAN pytorch tutorial: https://pytorch.org/tutorials/beginner/dcgan_faces_tutorial.html
* Experimented with latent size (input for Generator) and feature map sizes
* Added soft and noisy labels
* Added Wasserstein loss which is a good solution for mode collapse
* Wasserstein loss - https://developers.google.com/machine-learning/gan/loss
* mode collapse - https://machinelearningmastery.com/practical-guide-to-gan-failure-modes/
* Added dropout in both generator and discriminator
* [DCGAN Tutorial](https://pytorch.org/tutorials/beginner/dcgan_faces_tutorial.html)
* [Final notebook with modified DCGAN](https://github.com/AgaMiko/pixel_character_generator/blob/master/notebooks/3_DCGAN.ipynb)
### Example results
# Conditional DCGAN
Conditional DCGAN that generates a pixel character seen from selected angle.
* different learning rate for discriminator and generator
* soft labels
* added classification loss to the discriminator. Discriminator have to guess fake/real but also the character angle
* generator is conditioned with embedding from trainable look-up table that gives the info about the character view angle
* [notebook with modified Conditional DCGAN](https://github.com/AgaMiko/pixel_character_generator/blob/master/notebooks/4_Conditional_DCGAN.ipynb)
# DC Autoencoder
Deep convolutional autoencoder. This autoencoder have the same architecture as DCGAN above. The only difference is the additional fully-connected layer at the top of the encoder, which projects output from convolutional layer to selected latent size.
* embedding size = 40 is enough for a good-quality reconstruction
* autoencoder have great denoising properties
* easier and more stable to train then GAN's
* [notebook with Deep Convolutional Autoencoder](notebooks/5_DC_Autoencoder.ipynb)