Wrong Prompts?
Often times the genitals are not correctly distributed among the portrayed characters, am I using the wrong prompts? E.g. using terms like "the man (left) with brown hair" or something has... but it often still can´t recognize what or who I mean. Is this normal?
Also posing is even more fiddly to get right, even simple ones. Using the 11.3 model
I have found that it's a good idea to specify, e.g. 'to the left of the image', otherwise the AI models may sometimes take it as meaning from the subject's perspective.
In somewhat rare cases it may not correctly identify the gender of a subject, e.g. if a face is obscured or if a woman has short hair or is wearing a jacket or business suit, etc, so you may need to specify 'person' along with identifying information such as which side of the image they are on, or what colour outfit they are wearing. e.g. 'Have the person to the left of the image wearing a pink satin dress and have the person to the right of the image wearing blue jeans and trainers'.
For what it's worth, I don't have a lot of luck when it comes to posing subjects either. It can be pretty hit or miss.
e.g. I am often having to specify that a subject uses just the one hand for a gesture, such as waving or giving a thumbs up, otherwise the subject will likely use both hands.