V3I skipped version 2 because the file is v3, so I don't mix up the files I already have.Well, this version is trained in Illustrious, not on a specific model like version 1.I do have to mention that ChatGPT helped me with the settings, which is kind of funny and interesting that an AI can help me do this.I recommend using a strength of 0.9 and 1.Also, if you want to make a bird girl, use the tag "Beak,bird"And if you want a character that has a snout, put the tag snout, for example Krystal, you can do it without a snout and with a snoutV1This lora allows you to use the Thousandfoldfeathers styleFor some reason,It is difficult for the LoRa to understand the position of the camera, even if you want it to be a frontal image, from the waist up, she ignores what you want and maybe makes her kneel on the floor,but if it happens, forcing it a little will fix it.The LoRa is trained on the NTR MIX v21 model.

Thousandfoldfeathers Style

-Reduced the number of images.-Improved tags.-Tested new training settings.

Files with SHA256: f3cc103cd7aece869660c6387c3ccf85bf414f5058f32acb682503e107532391