You know, just a quick couple of online Photoshop classes.... and a good spell checker.... AI assisted content can be amazing, but relying too much on the AI, especially in a comic context where consistency between frames is a necessity, results in rather mediocre content.
Thinking about it, One rather easy way to make this kind of thing more consistent between frames is to generate the background separately , and get a model to generate the subjects with a monocolor background to chroma out. Chroma the character, or manually cut them out if the model wasn't able to make a uniform color background, and plop the subjects onto the background, maybe adjusting some of the contrast and color temps so they don't clash. Use to run a tumblr ask blog that took a similar approach, but instead I 3D modeled the backgrounds so I could get multiple angles without having to redraw the BG.