Hacker News new | ask | show | jobs
by kazagistar 3710 days ago
24/40

Poses, colors, and movements of people is very shaky, but I can do landscapes OK. But something about the structure of the questions bothers me... It seems like they are implying that I am observing the image, rather then generating it. Like, "how clearly can you see ____?" means I focus a bit and add it to the image, until I can see it better, but very little of the details asked about pop up unless I think about them, take the information asked for, and intentionally integrate it into the image.