Ever heard of someone making a LORA based on their own art? Well, I decided to test that. I already used my LORA for several months, so I now want to upload it to see what will users do with my style.
Some of the images I trained have the same pose or lacked shading, so reduce or avoid it:, add the tags to negative:
flat color, facing viewer, straight-onOr if you're gonna refine or upscale the image, first put the LORA around 0.6-0.8, then put it to 0.9-1.1 when upscaling it or refining it.
I also not drawing not much these days, but I need to add more drawings or shade existing ones, if my PDA allows it.
Description
This first version has some drawings of mine with repetitive front-facing poses, or lack shading. Since training on CivtAI as a free user isn't easy, with Knights of New Order being very repetitive and boring, I might end up training somewhere else.