>>1548 >>1549

>A lot fo epople here made loRAs with just 20 images and i love how they work

Really? I've been living under the impression I'd need at least a few hundred pics to get workable results. (some people even told me I'd need at least 400) >.<

>PS. I do not have pussy stabbing images, but i have generic stabbing images

Thanks, these are great, too! :)

>BTW you can train a model on civitai.

Yeah I've not yet checked their training tools out really. I uninformedly assumed it would be basicly the same deal as with colab, to be honest.

>or if yo uprefer you can just share the dataset and I will train it.

that would be awesome! :D

Do you have a Pixiv or another way for me to give you a link in private? (or are you the guy who started the reentry, by chance? If so we may have already had contact on pixiv.)

>PS. I love the models you produced so far

Thank you!

P.S. if anyone here can read chinese could you please tell me if the plaque roughly says what it's supposed to say or if I got trolled by google translate, again? xD