Trained using kohya_ss on ~30 images of Priyanka Chopra, mostly from the show Quantico.
Trigger is "prix" or "prix woman"; the most recent version does best on close-up through to cowboy shots using a LORA strength of 1. For full-body images, using inpaint for improved facial detail is recommended, as well as giving the prix trigger word higher emphasis.
Additional note:
If "indian" or "south asian" are used in the prompt SD tends to generate a bindi; adding a negative prompt of "bindi, pottu, forehead dot" mitigates this.
Description
Retrained from scratch on a better-selected dataset than v. 1.0. Works very well with SD 1.5 base model, but I noticed the model has overfit on smiles in a disconcerting way because I didn't tag lipstick colours in the dataset captions.
(This should be fixed in 2.1 when I can retrain the model overnight.)
This version does best on close-up through to cowboy shots using a LORA strength of 1. For full-body images, using inpaint for improved facial detail is recommended, as well as giving the prix trigger word higher emphasis.
FAQ
Comments (2)
v2 - Still not there yet. Getting closer.
I agree with the others, v2 is better but still not there yet
would you mind sharing the training data? I could rate and perhaps give you some hints on selection since I've made a couple of trainings myself :)
if not, that's fine too, I'm tempted to train her myself to check if she is one of those more difficult people to capture (some faces are easier than others, that is my experience)
Looks like we don't have an active mirror for this file right now.
CivArchive is a community-maintained index — we catalog mirrors that volunteers upload to HuggingFace, torrents, and other public hosts. Looks like no one has uploaded a copy of this file yet.
Some files do get recovered over time through contributions. If you're looking for this one, feel free to ask in Discord, or help preserve it if you have a copy.
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.