V1 - Trained on 35 images from various subreddits.
V2 - Trained on 325 images from various subreddits.
Images generated with Realistic Vision V2.
For prompts you're good to go with pretty much the basics:
full body portrait of a nude tattooed european woman on a public street <lora:flat_chest_v2:1>, RAW photo, (high detailed skin:1.2), 8k uhd, dslr, soft lighting, high quality, film grain, Fujifilm XT3Description
FAQ
Comments (4)
Looks like its getting close
This looks pretty good so far. I'll have to check it out later. Thank you for your work you've done so far! :-)
What checkpoint did you use as model to train and which one did you use to generate the content? Your lora looks pretty well trained especially for its small size. Did you use negative prompts? If not then you really did a nice job!
I'm curious why make this lora if we can get even flatter chests just by prompting? Not dissing your work, I just might be ignorant