Tarshka Style版本v1.0 (ID: 525774)

This model is trained on my drawing style. It's not very good because I am not very good but there is something very fun about making Stable Diffusion draw something in my style. I don't think anyone else particularly wants to do that, but I want it here as a backup.

Notes:

It likes to do characters on white backgrounds and will often put borders on the sides when it does do a background. It likes black and white with a little bit of color, because a lot of the training data is Inktober drawings that have a little bit of accent color.

It's not good at hands but neither am I, just in different ways.

It is very partial to a particular type of clothing because I draw that a lot on my characters. Ironically, I'm using AI to help me come up with more interesting clothing design now.

It likes to do people and struggles with most other things. It tends to make people have big pointy ears. If you prompt for animals, it will usually make them into people. Also it will do four arms sometimes, because there's a bunch of four-armed people in the training data.

It's just not very good in general. Results vary wildly by model and prompt. I have no idea how to optimally use it. I trained it following a tutorial, maybe one day I will try again to see if I can make it better but when I tried with CivitAI's lora trainer it didn't work as well.

描述:

Original version I trained following a tutorial in summer 2023. I don't remember the training details and didn't really understand them at the time anyway.

训练词语: tarshkastyle