This is a a Textual Inversion Embedding to create an abstract style with a lot of detail, but still recognizable content.
Works with the 768x768 versions of Stable Diffusion 2.0 and 2.1
To use it put the painted_abstract.pt file in in your embeddings folder and use painted_abstract as promt
I recommend a cfg below 10 and maybe even a bit lower for 2.1, it gets more blocky the higher the cfg
usually I use 7 for 2.0 and 5 for 2.1
I also recommend using an anime upscaler like RealESRGAN_x4plus_anime_6B
For the examples I used different samplers and both 2.0 and 2.1, generated at 768x768 and upscaeld x4 with RealESRGAN_x4plus_anime_6B, otherwise untouched
Inference Providers
NEW
This model is not currently available via any of the supported Inference Providers.
The model cannot be deployed to the HF Inference API:
The model has no library tag.