Is there anyway to overcome the jankness when using colab/paperspace >generate batch of images >it finishes and displays them all successfully >and the never again or img2img just stops working entirely and clicking the button does nothing >only fix is to reload the gradio link entirely
I don't know where they're coming from exactly but we obviously got several dozen people who are brand new to this stuff today and are still having their minds blown by stuff we became jaded to over the last month I'm annoyed by it too but we're probably not gonna make the thread better by sniping at them, just gotta ride out the AI eternal september
if i wanted to reverse engineer a prompt, is interrogation the way to go? i feel like describing the image the best i can isn't working out. probably because there's things like artist style that aren't immediately obvious to me
Caleb Martin
>grid consider eating shit
Hudson Jones
Basic rompt: harem, group of girls, akihiko yoshida, mark arian
Add your favored adjectives and quality control vectors. Personally, I like 'trending on artstation' and 'highly detailed'. Also add to the negatives as well; this particular group of images excluded 'big breasts' and 'busty', and included 'bikini'.
it's only been 5 days since I first downloaded it.
I experimented with different artist combo and their placement in the prompt. First the subject of the image, then the artists, then the filler details like clothing type works the best for me
Consider putting the grids your great images in separate posts so I can enjoy them more
Bentley Ward
They snipe back but only out of hurt pride.
They know that they are boring. It's important to let them know that others think so, too. Or, more abstractly, this is a general in Any Forums - Technology, discussing Stable Diffusion, not "sharing endless boring softcore general"