most of the models doesnt follow prompt nor refereence
most of the models doesnt follow fkin prompt, even its all clear as day not they follow reference. expecially pollo models 1.6 and 2.0. 2.0 do realistic image good but it stretches characters horisontally or vertically. also non of the available models a doesnt understand when i say "full character view" or "full body portret", models are too stupid to understand simple sentence.. for now best results are from wan 2.7, most realistic pictures, follows prompt atleast 70%, but it also doesnt understand when i write the above sentences. to make clothes shiny, glossy or whatever, most of the models ignore that part of the prompt