The goal was to create a model that provides higher percentage of images that i'd like to save. It's a hit and miss with SD model alone. So, to reduce the amount of images in the trash, here's an attempt to fix it with mixing SD1.5 with other models like f222. Mixed with auto1111 merge tool. The model can generate 512x768 perfectly most of the time.

This model can create beautiful women, also it is kind of a all-in-one, because it can generate many other things as well like landscape and sci-fi without a problem. Artist names affect the image like SD model alone. Can create nsfw content if asked.

For best results use VAE:

https://huggingface.co/stabilityai/sd-vae-ft-mse-original/tree/main