Collection of GPT-image-generator 2.0 issues, bugs, and work-around tips (check first post)

I still cannot really test images, because the results give me headaches… But the new methods are somewhat better when it comes to anatomical details, but probably still not perfect. (Diffuser models have even more topology problems.) Hands are especially difficult. Another problem that can be tested is tube-like objects, for example snakes or tentacles. Diffusers often turn a snake into a ring, or octopus tentacles into rings or free-floating structures… (I still have to find out, I hope they fix the bug, otherwise it will become tedious.) With the bug it will be especially difficult to make hands right, when they are small in the image.