Become a sponsor to Diego Porres
Hi! I'm Diego Porres, a Ph.D. student at the CVC, Barcelona. My research focuses on end-to-end driving and domain adaptation, but my passion also lies in generative art 🎨 and GANs 🤖!
As such, ever since late 2019, I've focused on delving deep into the code of one of the most popular GANs, StyleGAN, modifying it to fit my needs. However, I've noticed that the official releases lack the freedom many programmers (and non-programmers alike) need. This is why I've started to share some of the code I use on my machine locally for my art, so long as the code reaches a user-friendly stage. Here is a small preview of some art I've generated using StyleGAN3:
I'll continue working on my StyleGAN3-fun repository, which has a lot of work left! Your sponsorship will ensure that I can focus on continuous updates of my code, releasing it as Colab notebooks for everyone to use, and new additions and cool projects! But of course, a lot of fun code happens locally, so I'll ensure that I can dedicate time to both Colab notebooks and local code!
2 sponsors have funded PDillis’s work.
With this, I'll be able to concentrate a larger part of my time on improving my code, as well as preparing my own algorithms to share. In the same vein, I'll be able to invest in hardware to make tests for live performance with the visual-reactive code (https://youtu.be/KoEAkPnE-zA), as well as share videos and tutorials on YouTube!
Featured work
-
PDillis/stylegan3-fun
Modifications of the official PyTorch implementation of StyleGAN3. Let's easily generate images and videos with StyleGAN2/2-ADA/3!
Python 240 -
PDillis/stylegan2-fun
Slight modifications to the official StyleGAN2 implementation
Python 50 -
PDillis/stylegan2-ada-pytorch
StyleGAN2-ADA - Official PyTorch implementation
Python 24 -
PDillis/dcgan-pytorch
Let's go back to where it all began, let's go back to DCGAN.
Python 6 -
PDillis/earthview
Earth View is a collection of the most beautiful and striking landscapes found in Google Earth.
Python 7 -
PDillis/alis
Aligning Latent and Image Spaces to Connect the Unconnectable
Jupyter Notebook 3