A Frankensteined merge of too many models (a mash of Illustrious, Noob and plain XL models) with a ton of added LoRAs, just to try and get a general purpose model that can do almost anything. It’s Vpred, so treat it like one I guess. I’ll add more sample images when I do more testing. It’s my first real model attempt, so go easy on me 😂
Works best with 20+ steps, a CFG between 5.5 and 7, and Clip Skip 2 from what I tested. Sampler tends to depend on what you want (DPM samplers work better for realism and detail while Euler A works better for toon stuff with less detail). It uses e621/booru/rule 34 tags primarily. (I think it can also do natural language but I haven’t really tested).
Description
Can be finicky and probably needs some serious prompting sometimes, but it can usually do the job. I plan to make it better eventually, especially because it’s hard to get good backgrounds right now.





