I would prefer if you guys don't use this model for Commercial Use.

the lora you see is inspired by Babes models created by alexds9. I used the babes models to generate images for the dataset to create these lora models.

check out my pixiv for ai models and ai art (specific for anime): https://www.pixiv.net/en/users/34413372

tip me buzz if you want to support me so that I can make more lora model.

v4: I don't recommend using cartoon or anime model. This lora was trained for realistic model. I used images from sdxl babes model. this specific version (v4) is more Asian focus compared with my other model like v3 that focus on western. I used very few images to train this model.

The model images that I used to train the lora: https://civarchive.com/models/215877/babes-xl?modelVersionId=243278

v3:

epoch 1 and epoch 6

these are babes 1,1 style, but more cartoony. both epoch 1 and 6 are flexible, but epoch 1 have a little different style than epoch 6. I used babes 1.1 (sd1.5) and use images from that to train this lora.

I recommend start from 648x976, 976x768, 768x768

v2:

different style than the v1. This one is inspired by sdxl babes model made by alexds9.

this model is more bias for crop-top and buttons so add the crop-top and buttons in negative if you don't get what you want. Also add hands, fingers, 5 fingers in negative for better hands. v1 didn't need score_9, score_8_up, score_7_up etc. but the v2 need it more for better results.

v1:

I really like the Babes 1.1 model created by alexds9. The model art style is what inspired me to make this lora. in babes 1.1.

I recommend image size 1152x768, 1064x648, 968x968, 648x976, 976x648

I would recommend avoid using the score_9, score_8_up, score_7_up prompt and score_6, score_5, score_4 in negative prompt. It will effect the style. You also need to add furry, pony, mammals etc. in negative if you get furry art. add also 3d in negative.

I would recommend to use 0.7-0.8. when using the lora. The lora get worse if you go higher. it will easily get artifacts and artifact from upscaling with img2img and highres.fix. keep it low. If you want to get the style write lips or glossy lips. if you still struggle than write this prompt ''glossy skin, shiny skin''