photoshop of his comic and my machine learning model comparing the two

I made a dataset of Bizarro by Dan Piraro comics from a images on google. and created a image generation machine learning using RunwayML (paid like $35 bucks)

Here's the results:

a animation of the latent vector space

Well it kinda looks like his stuff...So there's that.

More generated images

example output

Examples of his awesome comic  

I like how the machine learning model picked up on the beige color that he uses! And the Salvador Dalí influence, which is more a result of how the machine learning training works but its funny to see if pop up  in the images. (smiley face emoji)

The take away:

There is a lot of room for improvement, id like it to create text that sounds like his stuff, and combine that with some more realistic images for a improved result.

For  a first try at machine learning id say this is a awesome success!.

Please support his awesome comics by going to

Bizzaro Tip jar


Bizarro's website


by oran collins
oranbusiness [symbol for email here] gmail [ascii number 46 here] com