Testing Neta Lumina using HiRes FIX seems to have enhanced the outcome, particularly in terms of color and line definition, although it might just be the usual post hiresfix effect as well.

I’m not sure, so I’ll stick with my beloved SDXL, which has controllable and accurate dan booru tags and controlable and accurate Loras, until a new model arrives that is both directional and factual, yet also creative and expressive, similar to NovelAI but in a local format.
Additionally, I’m a GPUlet and Vramlet, and this process took me 873 seconds. Lol.