05/15/26: V3 released - see version notes and all that.
Tl;dr - this model should make fur fluffier, scales scalier, feathers feathery-er...? No prompting needed, just keep it on mid-strength. Cool.
Works very well with KFT Requiem Fur a Dream and Tiger Milk 🐯🌿 checkpoints!
My highest rated model (KFT Furropean Extreme [PDXL]), now on Illustrious/NoobAI! Built from the same dataset as the PDXL LoRA, this model delivers some of the fluffiest fur, scaliest scales, and most feathery feathers available*.
This should work well with most NoobAI/Illustrious checkpoints to enhance furry features/environments, but the scene is changing/updating very quickly so this should be considered more experimental as it matures.
V3
Recommended Strength: 0.4 - 0.8 (the higher the strength, generally the more realistic it'll be)
Tags: N/A
V2
Recommended Strength: 0.3 - 0.6 (depending on type of character and checkpoint used: experiment!)
Tags: N/A
V1
Recommended Strength: 0.4 - 0.8 (depending on type of character and checkpoint used: experiment!)
Sampler: DPM++ 2M Karras / DPM++ 3M Karras (GPU seeded) (Euler gives a less 'fluffy' fur)
Tags: "kftenhance" (optional tag - may enhance certain features), "furry"
Negative Tags: "human", "hands", "fingers", "hair"
12/06/25: V2 released! See version notes for details.
*statement not verified for accuracy; offer not valid in Hawaii and Quebec.
🐶🐺🦊🦝🐱🦁🐯🐴🫎🦄🐮🐭🐰🐻🐻❄️🐼🦏🦫🦇🦦🦘🦅🦉🦚🦜🐦⬛🐊🐲🐍🦖🦑
Description
After almost exactly a year since putting out V1, here's V2 with a better dataset and improved captioning. Should work well with most checkpoints (see examples) and with other LoRAs to give better fluff, feathers, and scales.
Trained with a larger dataset and tested to find it better using a higher epoch with lower weight (0.3-0.6) depending on the checkpoint and intended results.
No tags specifically required.
FAQ
Comments (1)
I just want to point out that if you use KFT Requiem For A Dream and set this LORA to negative one you get some absolutely insane, stylized results and it's amazing
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.