Show HN: an app to create images with AI

https://geni.vercel.app/

by dvrp • 4 years ago

15 10 4 years ago

geni.vercel.app

mike_r_parsons• 4 years ago

This is awesome!

dalewebb• 4 years ago

The image wouldn't generate for me

dvrp4• 4 years ago

had issues with the queuing system... fixed!

gillis• 4 years ago

What image dataset/generation library does this use?

dvrp4• 4 years ago

latent difussion models (see: https://github.com/CompVis/latent-diffusion).

we also use our own open-source library https://github.com/thegeniverse/geniverse

jollyrosso• 4 years ago

How that is different with DALLE ? https://openai.com/blog/dall-e/

dvrp4• 4 years ago

in terms of ai, this uses latent diffusion models: smaller model, does not use clip, and it was trained on noisy data.

legaloslotr• 4 years ago

Why invent your own models when DALL-E exists? Is it supposed to be better/more-accurate?

scubakid• 4 years ago

I haven't read much about how these systems work yet, so this is probably a novice question, but I'd be interested to hear more about how the algorithm handles text input and feeds it into the generator. Does the training process include a ton of tagged images or something, and then the model learns to be able to generate stuff that corresponds reasonably to those tags?