Update V2 Portraitfocus: as the title says, minimal tagging also. normal v2 should be more flexible.
Update V2: Expanded Dataset and added cogvlm captioning. Still some occasional bad seeds, but works pretty well generally.
Description
FAQ
Comments (5)
I merged it into my actress face LoRA very precisely without changing her facial features.
Could you tell me how the dataset works? Was your datasets face blurred or cropped to keep the focus solely on the ballgag?
It's just a (some would say unnecessarily) big dataset with a lot of variety.
So the ballgag is consistent, but the girls are different, right? How many images did you use in total? Im using OSTRIS that can handle large datasets. Maybe you’re using Ostris too? It seems like it, judging from the output size file of your LoRA
@digitalyouneed dim/alpha settings determine the file size. Dataset was somewhere around 200 for V2 Portraitfocus and over 400 for v2 if I remember correctly.
thanks so much for share your knowledge.. very helpfull appreciate that
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.