MBW testing Part 2
Continuing with a new prompt. Focusing on the model's ability to retain the goals of one of my test prompts, 'sorceress in evil forest', as the base. The link shows my previous tests with about 350 generations of roughly the same prompt across many different models. If a model said it focused on a specific style, I might enhance the prompt to bring out those traits.
The images in this post are from attempts at using this APv1.4 as Alpha, Real Dream Classic 16 as Beta with MBW weights:
0.05,0.05,0.05,0.05,0.05,0.05,0.05,0.05,0.05,0.07,0.09,0.12,0.14,0.16,0.18,0.20,0.22,0.23,0.24,0.24,0.25,0.25,0.25,0.25,0.25,0.25and RCv1.6 as gamma with MBW weights:
0.125,0.125,0.125,0.125,0.125,0.125,0.12,0.12,0.115,0.11,0.10,0.09,0.08,0.07,0.06,0.05,0.05,0.05,0.05,0.05,0.05,0.05,0.05,0.05,0.05,0.05My goal in mixing RC back in is to help it with the way I switch models mid generation to adjust creativity towards the realism that the AP series is meant to have.
I think the images turned out decent. I can see improvements in the faces being pulled towards realism, especially when adetailer zooms in. Gonna call that good, make the pruned version and start uploading. After that I'll start on merging it back into a new version for Recursed Canvas v1.7.