Since this is so popular decided to put this one for Chroma too to help spread the love.
This was trained 3200 steps on around 200 iphone photos and selfies.
Rank: 32
Speed: 5e-5
Trained on: v48-detail-calibrated
Workfow is included in the images though you need to update the name of the lora.
The workflow i cobbled together from various things I've seen various people do. Any WF improvement suggestion is more than welcome.
Description
FAQ
Comments (2)
How to train? which tool? some guide available?
Great LoRA! Any chance you'd be kind enough to also share your image/caption pairs as a zip? I'd love to use your data set for regularization on some of my own LoRAs.


