This is Pony XL version of Nurse Luka which I made it before and previosly, it was trained under standard SDXL. Just use same existing datasets, but the prompt tags had been reworked to get it easier to train. How to use this LORA:<ol><li>Use the keyword nurse luka to activate the LORA. </li><li>Strength should be at least 0.7 to 1.0, depending on what kind of the model.</li><li>Use scale-dot product for cross-attention optimization (or just don't use it you have a GPU with plenty of VRAM including FP8 feature activated). </li><li>Usage may vary between models, additional prompt keyword may be needed such as pink hair, nurse uniform etc. </li></ol>Please remember when you want to submit the image into this LORA page, make sure that your image generation is really, really, really high quality. I don't want people submitting ugly hell images and trying to make it as an example from it.

Files with SHA256: faf31e6078b4f94d05986e9afd0369993b8b3fb12a32949f654b32399053c0c9