Please make it easier to delete/reject unusable results and refine results

It is strange that getting a specific number of things is such a challenge.
You would think that numbers would be pretty straightforward for AI to
understand.



Kayte

I want to be able to delete the files from the generator view. It’s
unnecessarily cumbersome to have to keep going back to the file view to
delete something, especially because in the file view, you only see
thumbnails, so it’s harder to see which ones to delete. When you click on an
image to see it closer, it takes you back into generator view.



It would be helpful if you could add the little check boxes to batch delete
from the generator view, or at least add the option to permanently delete
from the ellipses on the right of each batch of 4 AND on the top left drop
down for each image.



I was using reference photos to generate images, but the AI doesn’t seem to
have any concept of numbers. Even thought the reference photos for both
style and composition had three women, and I asked for three/3 women, it
kept giving me 4 or more. I frequently have this problem. Also, they often
had three arms, or just one; hands, as usual, were garbled; and clothes were
sometimes part one thing, part another – like one leg in shorts and one in
long jeans, or half one shirt, half another, or not resembling clothes, or
humans at all. The more times I generate, the worse it gets.



I am using the thumbs up and down. When I got something close that was a ¾
view of three women, I asked for the same exact women in the same room from
a wider angle showing the whole person including feet. It would not generate
a wider view. Later, with a different prompt, it gave me the whole person,
but 4 or more people, not the same women. Sometimes it seemed to understand
when I was focused on one image, and set it as a composition and style
image, and asked for the same women, it gave me the same women, and other
times, it changed them to completely different women. When I tried to
generate more ideas, it seemed to keep the same composition with slight
variations to the faces and clothing, and more and more errors and
distortions.



Three hours of trying didn’t really get me anything close to what I was
asking for, but at least it generated some parts I could comp in together in
Photoshop.



Kayte