AI General Thread

Viewing single post

Started by Legend, Dec 05, 2022, 04:35 AM

previous topic - next topic

Legend

AI apparently has issues understanding hamburger without cheese.  
It's mostly just bad at understanding the "without" part. Use a negative weight on cheese instead and it works just fine the majority of the time (but it did stick a straw in my hamburger)

Quote
Much harder to remove cheese, lettuce, onion, and tomato all at the same time.
Quote
In general though it is cool to find combinations that it struggles with. Yeah these are under water hamburgers like I asked for, but the hamburger itself does not look like it's underwater.
Quote
It's trivial to fix this stuff with loras (you give it a few example pictures and then the ai can learn a new concept) but they're a strong sign that these image models have a ways to go. I'm impressed though with the top bun being reflected by the surface of the water.