This LoRA was trained on minimally captioned beefcake imagery from the Athletic Model Guild of the '40s and '50s.
Each image was tagged with either 'b33f, photo of a man, black and white' or 'b33f, photo of two men, black and white'.
Many of the images were 512x768 so generating at that size will likely have the best results.
Positive Prompt
b33f : the trigger word; has the most effect on the style of the image generated
photo of a man, photo of two men : the number of subjects desired
black and white : the color of the image; however, some of the style bled into this tag so it may force nudity, etc.
Negative Prompt
border : to help avoid the white or black border on some images
Notes
You must lower the weight of the LoRA or prompt carefully to avoid penises.
Description
FAQ
Comments (4)
Awesome!!
Thanks! For the amount of effort I put in captioning, it works better than I expected.
@spline Can't wait to play with it! I have the dataset for a 60's-80's magazine style I never got around to captioning, I'll look at yours for reference
Thank you so much for this! AMG is an important part of history and this brings a lot of great things to these tools. No pun intended.