I've been messing with AI for quite some time now, but not at the level you guys seem to operate. I'm not even proficient with prompts as I just describe what I want to see in normal language.

I used Midjourney when it was introduced and I really liked mixing styles together to get something unique.

So 1st question is, is there a benchmark or a list somewhere that evaluate what mainstream models are good at base level for image gen? Like GPT could be good for portrait picture, another model could be good for scenery, another for anime...

2nd question, I'd like to build a small pipeline where I can output a consistant style across different types of subjects (characters, backgrounds, items...). I'm not sure where to start as what you guys seem to do is pretty technical, and also I have an AMD GPU (9070) which I think would block ways to get into it.
When I say pipeline, it would be a fixed pipeline requiring almost no tweaks even if it impacts negatively the result of the pictures slightly (if that's possible as of now).

Any recommendations?
Pic related is a 3D model I generated for Tabletop Simulator where I managed to somewhat create a "pipeline" to obtain consistant results creating several models.