Is because with AI you can jump between "dimensions" from 2D style, 2.5Dish style 3D or realistic with one single img. Not everyone can do great realistic ones mostly because model and skills for editing the img with a photo editor.
Discussion
I am not personally familiar enough with the technology to diagnose reasons why. I am sure that there are some models that do better than others, and that there are also those making such generations that are doing editing after the fact to clean up wonkiness in what they generated, while others simply lack the skill to do so.






