Part two of a series of artifical intelligence assisted (Stable Diffusion!) photos from the world renowned Steampunk Circus.

Still 002 - AI assisted image generation. A close up, humanish circus performer with creature like features. Light shines through the fine white and black hairs and long face resembling a wallaby yet possesive of the strangest, gear-like eyes. The depth of the image is pronounced with a fast blur exclusive of the fine facial features.

I hope you enjoy. These come from a 1.5 model dreamboothed or similar in some manner. Some version of a version of a circus. It’s interesting using the various tuning methods and feeding it it’s own goods. So much more approachable these days to experiment though. I can’t keep track of it all sometimes.

It takes a surprising amount of GPU geebees to get these rendered. I’ve got a reasonable workflow now that doesn’t result in OOM almost every time so that’s a plus.

Tooling

Stable Diffusion
Deforum
Google FILM
ffmpeg
Colab
& Python.

I’m weaning myself off of Premiere slowly. Still used it here a bit but mostly ffmpeg and soon enough I’ll be moving to #moviepy, I think \0/, for easier stitching and sequencing during rendering with the models in the loop to some extent. I love #deforum so it’s mostly grown on that and looking to just put it in that notebook, just a better interface. I know about RunwayML very much so. I used it extensively at one time.

Perpetually preposterous! Subscribe, like and share. Much appreciated.

Let me know if you would like to see more of this type of stuff. Also, I’m posting videos that aren’t directly related to the circus on my personal YT at @johnnypeck Artificial intelligence, machine learning and robots fascinate. An amalgamation of automatons. An Amalgamaton!

Aiart is art with some new tooling added. Now art is more accessible to more people. Isn’t that wonderful? Creativity reigns.

Set 2 of ?