Just getting the number of fingers and to a lesser degree arms and legs correct would solve 80% of the issues I encounter
Right. And imagine being able to refine an image with the instruction to move the left arm down a little, and lift the chin up just a tad. Today that's impossible.
Re:Ten Years of Image Synthesis
by fjkdlsjflkds on Wednesday November 30, 2022 @12:57AM
(#33797468)
Are you sure it is impossible?
https://arxiv-sanity-lite.com/?q=diffusion+image+editing&ran...
Very cool, thanks.
Re:Ten Years of Image Synthesis
by theemathas on Wednesday November 30, 2022 @03:15AM
(#33798245)
Seems really close to reality https://prompt-to-prompt.github.io/
Legs, arms, and heads correctly oriented too.
Re:Ten Years of Image Synthesis
by theemathas on Wednesday November 30, 2022 @03:10AM
(#33798206)
Proper hands are already here https://www.reddit.com/r/StableDiffusion/comments/z8gje0/har...
Some of those hands are ok/good, but even the second image in that collection has messed-up hands.
After a few dozen prompts, you quickly learn how the image generation is simultaneously amazing and dumb. The dumb part is the next 10 years of improvements.
This is like Google 1999.