Type | |
Stats | 657 113 |
Reviews | (144) |
Published | Apr 15, 2023 |
Base Model | |
Training | Steps: 2,000 |
Trigger Words | BW |
Hash | AutoV2 56756572A8 |
Experimental Lora I made with a dataset of simple black and white geometric shapes I made using Photoshop.
0.1 - .9: Generally has subtle influences on the outcome, making the image lean towards a more simple background, richer colors, and SFW images.
-1 - -.1: Does the opposite, making more complex backgrounds, and subdued colors, and pushes for a more NSFW image despite not having explicit tags in the prompt.
1.1 - 2: Cuases the outcome to be more abstract, making geometric shapes very visible, sometimes even causing the character to disappear. Use with caution.
Trigger word: BW
Monochrome and Greyscale also are stronger. Try giving them negative weights for vivid colors, for example: (monochrome, greyscale:-.3),
Some models react to this lora more than others, best to experiment and see what makes the most interesting thing to you.
If you like this one, you might also like my other Lora that is similar in concept here: https://civitai.com/models/44924/bwheartabstract