NFTArt is an app created with the intent of generating art based of different assets the users provides.
This is still in development, many issues exist, but they'll be fixed soon :)
A trait that is present in only 1% of the generated NFTs doesn't always equals to a rare trait : Take for example, a collection that contains 3 different types of assets :
- Hats
- Glasses
- Head color
If we create 100 different Hats and all of them have equals chances to be generated, it'll mean each of them will have, on average, a frequency of 1%. This does not depend on the collection size. To solve this problem, we use the frequency over the most frequent asset :
| Hat | Frequency |
|---|---|
| 1 | 25% |
| 2 | 20.5% |
| 3 | 20% |
| 4 | 3% |
| 5 | 1% |
| 6 | 0.5% |
| 7 | 30% |
In this example, we can all agree that the rarest feature will be Hat 6. But, how rare is it ? To evaluate we use the frequency over the frequency of the most frequent feature :
HAT 6 Rarity : 0.5 / 30 = 0,01666666667
HAT 7 Rarity : 30 / 30 = 1
We can of course later normalize it by subtracting the result to 1 giving us an higher value meaning an higher rarity trait :
HAT 6 Normalized Rarity : 1 - (0.5 / 30) = 0,9833333333
HAT 7 Normalized Rarity : 1 - (30 / 30) = 0
Knowing all this sorcery maths, It'll be easier to make the rarity score. To apply this to the generation we can make sliders that all add up to 100% and make it so the generator randomly but accurately selects a trait for the generation.
We can also add a rarity trait representing the sum of the normalized trait rarities
- Install dependencies :
npm installORyarn install - Launch the app :
npm run devORyarn dev - Open the app in your browser :
http://localhost:3000
OR Visit : https://nft-art-gen.vercel.app/