I came across this style on Pinterest the other day and I gathered all the image data I could find on the web. I removed watermarks and any other artifacts. I trained it twice. Once in my pc and once on site. Both results where failures. Maybe because the style was too abstract? I don't know.

Then I decided to add another painterly style from the countless here on civitai. I adjusted the weight and I came with something a lot less abstract but 'easier' to train. Then I used 'image to image' of the original training data with this combination of styles. The generated images where used to train this style.

As always, test it out and let me know what you think.

Activation Prompt: T1th1, abstract, fantasy, painterly, scenery, surreal

Negative Prompt: keychain, lowres, worst aesthetic, (worst quality, bad quality, low quality:1.2), bad anatomy, old, signature, watermark, artist name, 3d

Weight: 1.8

Sampler: Euler a - Automatic

Recommended Base Models: WAI-NSFW-illustrious-SDXL V10

Make sure to post your favorite generations on the Gallery by simply clicking +Add Post button.