I was wrong (OpenAI's image gen is a game changer)

so that look wasnt april fool joke

love how theo changed his appearance using AI for the sponsor promotion 😊

bro went from looking like a 80's homosexual to a 2000s emo boy

We got Emo Theo before GTA VI

Theo grew his moustache back for the sponsors

If you click on the generated image, you get access to the "select" tool, which allows you to get the AI to focus on updating a specific part of the image.

theo got regenerated

1. go to that break finally 2. also buy some bright curtains while you're out

I’m not a professional coder, I’m a visual artist who uses code sometimes and I think OpenAI is now doing what they should have worked towards all along. I remember when neural style transfer first came out and was very slow, I was working on a custom video glitcher which used open cv and vgg16 to basically call tool calls on different parts of the image and do both algorithmic and neural edits combined. Then stable diffusion came along and it was fun for a while, but just expecting latent space to spit out a result is boring and out of our control. Making tool calls to preprocess the image or data on a high level and then using neural networks for the low level seems the most logical way to make a good end result. I used to feed my p5js outputs into stable diffusion. But honestly I’m bored of ai and also lot of people are abusing it to make disgusting shock content. But yeah this new open ai stuff is wild gotta respect the engineering. Lot of ai enthusiasts and companies only care about having the largest data driven latent space, I think we need more engineering and architecture in these systems. Especially with the rise of vibe coding and the loss of low level programming language skills that will come in the future generations of programmers.

Published this video 1 day earlier than planned, right?