Home 
 Map 
 ⇑ 

 ⇐   January 29th, 2023   ⇒ 

Copyright 2023 Michael Anttila

Last year I started playing around with AI image generation. Now that I have a new computer, I'm able to take it one step further and actually train the AI on my own images. I started where a lot of people start: Training a model on images of myself, and then trying to get it to generate an image of me as a character in Star Trek.

It is a bit of a fiddly business, training a neural net. There are a lot of dials and knobs, and it is easy to either undertrain or overtrain. After sorting through hundreds of horrifying abominations, I finally managed to generate a few half-decent images. This one was my favourite.

Technical Details: This photo was created using Stable Diffusion.

POTW - Photo Map - Home - Feedback

Hosted by theorem.ca