Trained at 1280 resolution with nearly 300 images. Only booru-style tags were used for this version. This version can be a bit difficult to use on its own and may require some work searching for good seeds. Lower the weight a bit and/or mix with other LoRA if it's too much trouble.