NOTE: V4 changed the quality tags. Update your prompts.
This LoRA adds quality tags inspired by NAI to create aesthetically pleasing images without artist tags. The dataset was split into 4 tag groups based on their favcount on e621.
V4 is for fluffyrock-1088-minsnr-zsnr-vpred-ema (trained on e267), which requires a corresponding yaml file. You will also need the CFG Rescale extension.
V3 is for fluffyrock vpred (trained on e160).
Usage: add the tags to the prompt, use emphasis to change the strength of the effect. Do NOT use LoRA weight.
Positive prompt: (best quality, good quality:1.2)
Negative prompt: (worst quality, bad quality:1.2)Description
Retrained from scratch for more consistent tag names. The quality tags now follow the pattern: best/high/normal/low/worst quality.
masterpiece was causing issues with some mixed models by forcing a painterly style
FAQ
Comments (1)
This is likely beyond the scope of your interests, but would you be willing to do this with other models that are a bit closer to the SD 1.5 model merges that are more common? Even with the same dataset is probably fine, especially if it's with another furry model. I intend to use the LoRA Bock Weight extension to limit to the middle blocks to primarily influence composition in models other than Fluffyrock, like Photon and EpicRealism.