by mful on 11/28/22, 6:33 PM with 2 comments
Doing some research, I discovered StyleGAN and similar technologies. I found they were amazing at certain tasks, like turning photos to impressionist paintings, but were too limited to create the stuff I was most interested in. So I moved on.
Fast forward to today, the recent advances in generative imagery made it seem like the time might finally be right for this idea. So with a kick in the tuchus from fellow HN'er lowe0292, we decided to build it.
We use DALL-E 2 for generating images and SharpJS for the under-the-hood image manipulation. We use a service to upscale the images before printing, but will eventually run our own setup with Real-ESRGAN or similar.
Feedback very much appreciated :)
by lowe0292 on 11/28/22, 7:09 PM