Inspiration

This project is an art turing test and a experiment on understanding the relationships between art, value and human bias. It's a known effect that people value art made by a computer less than art created by a human. I wanted to know how people would value art if they didn't know who created it. After which, using blockchain technology, we can create NFTs of the generated art and see how much they sell for. Because I was not able to complete my hack due to time constraints.

How we built it

The GAN that i'm using is called StyleGAN2 which is a model that released by Nvidia. Training the GAN takes some time, so I used a pretrained model that I found on Google Colab and the notebook is available in my repo. StyleGAN2 can present some amazingly accurate results.

The website is written in React

Challenges we ran into

The backend was not completed in time regrettably.

What's next for Project Finesse

If I win one of the prizes, I would use the money on the following things: - collecting/scraping a separate dataset with abstract, contemporary art - paying for a Google Colab Pro account to support training StyleGAN2 on the aforementioned dataset - finishing the firebase backend and making it production-ready - improving the mobile responsiveness of the website

Built With

Share this project:

Updates