Doesn't Laura look like she's wearing a pink nun's dress?
Anyway, I bring you another To Aru character. This anime has a lot of female characters with beautiful designs (and half are nuns), seriously, there are about 50 and not even half of them have a model here in CIvitai, that's why I upload one from time to time. Today I brought you Laura.
She doesn't do anything, but she looked very sexy all wet in the bathtub.
My ko-fi page if you want a commission or support me with a little bit:
It helps me to keep bringing you Loras almost daily c:
Description
FAQ
Comments (6)
I saw some images of her now and realized that the sleeves of the dress are wide, so, you can try putting "wide sleeves", my bad
Hello you Next Movie One Piece Film: Red of nami pony loar model
Hello, I want to learn how to make pony lora. I have learned how to make sd 1.5 lora.
What should I pay attention to when comparing it with sd1.5?
I would like to ask which training base mask you use?
sd 1.5 is 512x512, what resolution does pony require? And the number of training images is also 20?
Hi, I'm glad you are interested in learning.
Well, Pony training has a couple of differences, the most notable is the amount of steps. While in SD1.5 it was around 2000 to 3000 steps here it is much less. In my case my PONY Loras go between 150 and 300 steps, other people train with more, 500, 700, 900, even 1500 I've seen some, but I don't recommend to go that high, besides it takes much longer to get done, sometimes the Loras get “overcooked” with so many steps.
As for the dimensions of the images, it doesn't necessarily have to be 512x512, it can be any, you just have to make a good dataset and raise the resolution if possible (I do it sometimes with the upscaler of Stabble diffusion). As for the amount of images, it all depends on the character and what you have to work with. I have made Loras with only 4 images due to lack of material, I have made others with 200, but what I usually find are datasets of about 25 images, if you have more, that's good, if you have a little less, that's fine too, but it is not necessary to have so many images to make a Lora, at least when we are talking about characters.
As for the base model, it is PDXL V6.
I hope you found it useful.
Thank you, your answer is of great help to me, I will find some character training materials to try!!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.