Are you tired of your old NFTs but not really into selling them? What about modifying them with our tool?
What it does
We made a protocol to modify existing NFTs with the help of neural networks using different filters (adding animation and movements, changing style, background, and many others)
How we built it
We developed the smart contract that uses the create2 mechanism that provides the technology to allow one NFT to own another. As well we trained a neural network to perform animation transfer of motion from source NFT to another NFT. Later other types of neural networks transformation types can be added and applied
Challenges we ran into
It was quite challenging to run a virtual machine with GPU to perform neural network inference. As well we faced some difficulties with the frontend part (deployment and interaction with a smart contract)
Accomplishments that we're proud of
We succeeded in training a neural network to generate good-quality gifs and created some funny NFTs.
What we learned
How to work with Polygon How to mint NFT tokens How to create proxy-contract How to train and deploy neural networks How to run GPU virtual machine How to use Flask
What's next for DO[NFT] - Dynamic Ownable NFT Collection
We will try to release our project in mainnet Polygon and start to attract NFT creators