>>2802478The regional prompt extension for the auto1111 thing has help links right in the webui (Usage guide). It's not underlined, or has a question mark and is in a weird place below the title of the dropdown.
You type in the corresponding prompt BEFORE the keyword. You have to tick use base, use common, use common negative, I'm not sure how this last thing works.
ADDBASE = this is for the whole image, example, a pov photo at a dinner table, the base ratio increases how hard this goes, 1.0 is all base, 0.5 is half base half regional
ADDCOMM = this will be applied to every region, example, everyone has massive sagging breasts, so you don't have to add it every time
ADDCOL, ADDROW = columns (vertical) and rows (horizontal), example here you type in, this person has a blue dress and red hair.
The only other thing that isn't retard proof is the numbers. You actually have to think a bit. So you have horizontal/row or a vertical/column workings ticked, this are ONE. Inside of ONE you separate the PARTS of ONE with commas (,), to add another ONE you use a dotcomma (;). BUT if you have many ONEs the first comma separated number is used for the size ratio.
IT WILL SHIT THE BED AT LOW RES, AND AT OTHER STUFF TOO. Example image was only 512x768
