Miami Buildathon Spotlight: AI Outfitters' Winning Solution Using OWL Protocol Contracts and AI-Powered ChatGPT
The Miami Buildathon, held in March, was a platform that united talented teams with diverse skills in development, art, and business. As one of the event's sponsors, Owl Protocol actively engaged with participants in their pursuit of solutions for the following challenge: build a dynamic NFT project that interacts with Owl Protocol's on-chain DNA data standard. The goal was to leverage on-chain storage to manage and store dynamic NFT metadata directly on the blockchain.
On-chain DNA involves encoding NFT metadata into a 'dna' bytes data structure stored on the NFT token contract. The decoding schema is stored on IPFS, and its hash is added to the NFT smart contract's baseURI. This approach offers increased transparency, improved data efficiency, and enhanced accessibility and interoperability for developers and users.
This blog post will provide the following:
- An overview of AI Outfitters’ winning solution.
- Delving into its features.
- Technical implementation.
- The role Owl Protocol’s contracts play in enabling it.
AI Outfitters' Motivation
Breathing Life into NFTs
One of the primary issues with current NFTs is their static nature. The team recognized this limitation and sought to bridge the gap by infusing NFTs with dynamic content. The goal was to transform NFTs into interactive, personalized entities that reflect their owners' unique preferences and experiences.
The Solution
AI Outfitters devised a solution that utilizes Owl Protocol contracts, AI-powered ChatGPT, and MidJourney. Here's an overview of their approach:
Dressing NFTs Based on Location: AI Outfitters employed ChatGPT to predict clothing styles based on location data. The Outfit suggestion is sent to an artist who could create the clothing aligned with the base NFT, and the background outputs are sent to be generated in Midjourney.
Generating Dynamic Layers On-Chain: To address the challenge of aligning developed image layers with NFT PFPs, AI Outfitters used TopDownDna.sol contract. This allowed them to dynamically style NFTs by generating and applying different layers of clothing and accessories on-chain.
Owl Protocol's Key Features Utilized
Specific features of the ERC721TopDownDna.sol smart contract were used to enhance their NFT customization solution.
1️⃣ TopDown: The TopDown feature enables an NFT to own other NFTs on-chain and exposes all owned NFT data to the owner. This functionality seamlessly incorporates additional NFT layers and accessories, expanding users' customization options.
2️⃣ DNA: Owl Protocol's DNA feature enables the standard encoding of NFTs on-chain using an off-chain schema. This encoding mechanism ensures efficient storage and retrieval of NFT data, optimizing the performance and scalability of any project.
Technical Implementation
Firstly, they generated layers through AI tools and uploaded them to IPFS. In our current version, a developer must still generate the JSON schema of the NFT’s dynamic traits. A developer primarily uses the OWL Protocol NFT SDK to import the libraries NFTGenerativeTraitEnum and NFTGenerativeTraitImage.
The traits included an enum for “Vibe” and image-based traits for “Background,” “Body,” “Glass,” and “Hats” as shown below. Optionally a developer can specify the “probabilities” of each trait’s attribute, which is just the weighted probability corresponding to the index of each attribute.
import { NFTGenerativeTraitEnum, NFTGenerativeTraitImage } from '@owlprotocol/nft-sdk';
export const traitEnumVibe: NFTGenerativeTraitEnum = {
name: 'Vibe',
type: 'enum',
options: ['Chill', 'Boring', 'Eccentric'],
probabilities: [70, 20, 10],
};
export const traitImageBg: NFTGenerativeTraitImage = {
name: 'Background',
type: 'image',
image_type: 'png',
options: [
{
value: 'Dunes',
image_url: 'ipfs://QmfSABDaq7V2WKrdTnK3ofnnbucax4e5jBcztXqL34zsrL/bg-dunes.png',
},
{
value: 'Downtown',
image_url: 'ipfs://QmfSABDaq7V2WKrdTnK3ofnnbucax4e5jBcztXqL34zsrL/bg-downtown.png',
},
],
probabilities: [5, 5],
};
export const traitImageBody: NFTGenerativeTraitImage = {
name: 'Body',
type: 'image',
image_type: 'png',
options: [
{
value: 'Common',
image_url: 'ipfs://QmPwdFi3y79iDtG59CE1kEB395KemBe2xBZu1uVs7Xphvr/body-base.png',
},
{
value: 'Albino',
image_url: 'ipfs://QmPwdFi3y79iDtG59CE1kEB395KemBe2xBZu1uVs7Xphvr/body-base.png',
},
],
probabilities: [2, 8],
};
export const traitImageGlass: NFTGenerativeTraitImage = {
name: 'Glass',
type: 'image',
image_type: 'png',
options: [
{
value: 'gucci',
image_url: 'ipfs://QmPwdFi3y79iDtG59CE1kEB395KemBe2xBZu1uVs7Xphvr/body-glass.png',
},
{
value: 'prada',
image_url: 'ipfs://QmPwdFi3y79iDtG59CE1kEB395KemBe2xBZu1uVs7Xphvr/body-glass2.png',
},
],
probabilities: [5, 5],
};
export const traitImageHats: NFTGenerativeTraitImage = {
name: 'Hats',
type: 'image',
image_type: 'png',
options: [
{
value: 'Cap',
image_url: 'ipfs://QmPwdFi3y79iDtG59CE1kEB395KemBe2xBZu1uVs7Xphvr/h“For the hackathon, we used Midjourney to generate NFT backgrounds and ChatGPT to develop ideas for closing styles based on location data. But we still needed a designer to create the assets to ensure they would be properly positioned when layering on top of the base NFT. As a future improvement, we would like to go fully autonomous on utilizing ChatGPT and Midjourney to generate the assets, and Chainlink functions or GPS to accurately determine the user’s location and suggest appropriate clothing options for their NFT.“ — mention Anna Kazannik, member of the team.
We are thrilled to have sponsored and participated in the first Miami Buildathon, an incredible event showcasing developers' and creators' exceptional skills and creativity. Engaging with such a talented community has been truly inspiring for us.
Congrats again to AI Outfitters' team: Anna Kazannik, Anto Joseph, Krishna Priya, and Sheena Johns.
If you want to explore Owl Protocol’s contracts or build exciting products on our platform, we encourage you to check out our documentation here. Our team is ready to support you and answer any questions.
