Self-Explanatory is built from a few artists I like that specialize in watercolor and acrylic paintings, with their subjects being a mix of surreal anime and reference photographs. If you like big lines, textures, swirly backgrounds, and interesting eyes, Self-Explanatory might be the one for you!
Alongside the activation tag s3lfxp, try other tags like
watercolor \(medium\), acrylic paint \(medium\), canvas texture, photorealistic, surreal, abstract background
Things to keep in mind when using this LoRA:
This LoRA was trained on NoobAI V-Pred, and will have limited to no-effect on NoobAI Epsilon on Illustrious-based checkpoints. NoobAI V-Pred checkpoints often require additional configuration when generating locally, so make sure to read up if you've never used one before!
Compared to my other LoRAs, Self-Explanatory is weak. You might need to bump the weight up past 1.0, and NoobAI merges that have strong built-in styles may override most of the look of this LoRA. Check out the sample images to see checkpoints I feel work well with this LoRA.
Hi-Res Fix style upscaling tends to ruin the textures inherent to the LoRA. Either don't upscale at all, do a pure upscale with no further img2img, or suffer the consequences!
As is normal with painterly art styles, standard quality tags like masterpiece, best quality, newest tend to worsen the effect of the LoRA. Try not using them if you find the LoRA isn't very strong.
Due to the primary subject matter of the artists referenced, humans generated with the LoRA will trend female and younger. You'll have to use tags like aged up, mature female, or mature male to get older and less feminized appearances.
No NSFW/R+/suggestive training data was used, so you'll have to rely on your checkpoint to handle NSFW.