Art? ART!

Inspiration

When I was researching the modern art, I find the generative art. As a data science graduate, my first mind is that my knowledge can make changes in this area. Due to the $432,000 (£337,000) sell of AI painting Portrait of Edmond Belamy, I noticed this can be effective and valuable way in generative art. Moreover, with the help of blockchain, the NFT will be the new type of AI generative art.

What it does

During the step 1, our art? ART! platform will bring our users the revolutionary experience of NFT Art. Users can pay some fee to generate their own style of NFT art without any SKILLS and professional knowledge. The only need is that user's upload two images, one is the image with their favorite painting style, another one is the content image which contains the thing they want to draw. After paying fee, the user can get the stylized images with the thing they want to draw. For example, user A wants van gogh to draw a car, so A can upload a car image as content image and a van gogh's painting as style image. After modeling, our platform can generate the stylized image as NFT for our user.

How we built it

Our platform will took React + Flask + Brownie as our framework. First, the data will upload to our server. After addressing images, the new image will return to our pages and also stored in IPFS and generate ERC721 token in blockchain.

Challenges we ran into

As a sole project, I need much work to do. I'm not familiar with the front end, which will take a lot of time.
Long time of waiting. Due to the Deep learning model need the GPU to computing, the model will take a long time to get a satisfied result, except we have better and more GPU to reduce the time.

Accomplishments that we're proud of

I have finished the model part, which the core part of this project and the result is well. The art platform can preprocess user's image very well, including crop, resize and return corresponding size image The NFT part can generate the ERC721 token for image, stored on IPFS and published on opensea.