>>2771734Nah nah nah, if you have more than 11 images set the iterations for each image to 100, if you have less to 150. You tag each image individually and you descrive what's in there, if it's a girl type it. You don't need to necessarily add the token to the description but you can, and if when you use the LoRA it makes the image fried just don't use it, if you're training on a girl every girl you generate will look like her anyways, just specify hair and eye color and if they have some unique clothing.
Training for character or styles is done the same way.
