<p>Although version 1 of this LoRA seems very capable, I noticed it had trouble producing some types of images in the correct style. I suspect this was because of how I had written the labels for the training images (I was lazy, labeling training data is tedious).<br />This version was trained with much better labels, so hopefully it should have a better idea about what constitutes the style this is supposed to emulate.</p>

Files with SHA256: e903c96c9bce1b3cad04536d9a323df77147a1c18fb2e2c03c25eac16530214b