CLIP SKIP = 1. Model was trained using CLIP SKIP 1.

I suggest using this VAE.

I wanted to get started on working on a model for myself, and this is the beginning of my new journey. This model was both trained and then merged.

The main objective of this model is to have a style that you can prompt for. If you don't prompt for it, the style is mostly a basic anime style.

If you prompt for "Clay Mann" (I suggest always using it as your first tag in the prompt to make it as strong as possible), it will use that style instead. So far, I'm very happy with the results and will probably keep iterating on this model.

Other than that, just use normal Booru-style prompting.

If you want to use a LoRA, it's a good idea to use the LoRA at low weights when also prompting for the Clay Mann style due to style conflict.

You can use tags like "outline," "lineart," "grayscale," "monochrome," "flat colors," "dark," and "vivid colors" in both positive and negative prompts, and they have a strong effect on the output. For example, if you want vivid colors, you can negate "grayscale" and "monochrome," and the output will end with brighter and stronger colors. I really like the outline and grayscale outputs from this model!

Nothingv2.3 as my base (cus its one of the best models I've ever used) + a private fine tuned model.

Model is heavily biased towards MY own taste.

Please check out my Ko-Fi page if you are interested in supporting me.

LoRA commissions open! -> Fiverr