So there are LoRA that take whatever that’s generated and make it look like doves or make it like manga and things like that. So instead of prompt engineering the style, one can just imagine the style. And because getting the source materials for this kind of style transfer is actually quite easy because everybody got a phone with a pretty good camera, you just go and take pictures of before and after transformations, and you can easily curate your LoRA for that kind of style curation and so the community around stable diffusion very quickly discovered that those LoRA stack. So you can append like three, four different LoRA, which are trained by different communities, but taken together, they will make the eyes actually be in the right place, the limbs, the right counts. [laughter]