Trigger word: BWWM
Basic prompt: 1girl,1boy,dark-skinned female,interracial,very dark skin
Weight: 1
CFG Scale/Steps: 8/30
I had issue with creating Black Women/White Men pics, because checkpoints and LoRAs either have bias toward fair skin characters or change everyone/men skin color to dark, so i made this LoRA to help me fix this, hopefully it will help others too.
It works best with YogiriMix, however i did some small test on different models and LoRAs, it seemed to either work with them pretty well or they needed a little bit of tweaking strength of prompts.
Description
FAQ
Comments (12)
Very underrated concept.
Agree. When I will get better at understanding machine learning and how diffusion works, i will try to make checkpoint that will be biased towards making BWWM hentai
@Alan_Rex I'd say you should fine-tune it on NSFW artworks of only BWWM, I'm no expert but looks like after the tuning any model in itself will be biased on the last training dataset, making it looking sort of "overfit" even when it's not. I say this because I happened to merge models that seemed to grasp very well a concept and after the merge they totally forgot.
Love the idea but sadly isn't working with InvokeAI. :( Not sure why. I'm replicating all of the settings I can. I'm a newbie.
Hot hot and hot, much better then that trashy blacked content
Need a lora for white man having a threesome with a white and black girl
BWC stay winnin doe
In your experiments, did it ever work with realistic models/checkpoints?
Please please please make an XL version :)
You are a legend. I've been struggling with this.



