The exact same dataset as used for ZootPhotomaxxer XL (captioned with Florence-2 Large rich captions and also WD-VIT-V3 Booru tags), but trained on SD 1.5 (still at native 1024px, like all my SD 1.5 Loras).

Recommended strength is around 0.6 to 0.7. Using 'raw, photo, realistic' anywhere in your positive prompt will significantly boost the effect the Lora has on the image. All showcase images were generated directly with Base SD 1.5 and this Lora at 0.65 strength.

This Lora is most useful either during hi-res fix or for generating directly at higher than standard SD 1.5 resolutions.