Generative Avatars

Article
Preston Brown

2 months ago by prestonbrown

Gemerative Avatars are pretty popular in the crypto space. Most peple know about Cryptopunks and the famous $11.8 million sale.
Cryptopunk

I have become interested in hash-generated images with complete sets, a unique image for every hash. A good example for looking at this is MonsterID
MonsterID

MonsterID takes a hash of text, usually a username or email address, and assembles layers of sub-images and colors to make a unique avatar.
Unique Avatar

This means that for a given hash, MonsterID needs to pick from a selection of items together, and combine them for the complete image, take this chart of images:
chart of images

In this example alone, there are:

  • 12 unique head pieces
  • 20 eyes
  • 17 mouths
  • 14 back ornaments
  • 20 body types
  • 18 leg types

Altogether thats:

12 x 20 x 17 x 14 x 20 x 18 = 20,563,200

Over 20 million unique monsters. That’s not a ton of work for a huge selection of potential avatars. The power of this method also continues to scale. If you add another layer with just two options, you have doubled your total number of possible combinations to 61,689,600.

I wonder how many layers and combinations you’d need for every Bitcoin address? 🤔