<a target="_blank" rel="ugc" href="https://my.bio/diffusiondudes"><img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/2b56a416-6ad5-4f3f-8334-08e0a9d735e3/width=525/2b56a416-6ad5-4f3f-8334-08e0a9d735e3.jpeg" /></a><a target="_blank" rel="ugc" href="https://my.bio/diffusiondudes">Visit my Gumroad/Ko-fi for exclusive access to all 60+ of my LoRAs! ▼▼▼▼▼</a>I'll be releasing 1-2 LoRAs on Civit every month, with early access on my stores, and over 60 exclusive models 💪 Members can get full access to my entire archive plus my private TG for my steamiest content!-This LoRA is designed to produce photorealistic bodypaint on semi-nude to completely nude people. Please note though, I used almost all men in my training data. Because of that, it will be difficult to perfectly reproduce female genetalia without inpainting/extra prompting. You can make it work, however. Just expect a lot of penises. I recommend you use tags for genetalia. And if you work with a model that isn't trained on penises, you'll get results like the ones I've posted, they're almost there but always a lil funky. For models that work well with this, check out my <a target="_blank" rel="ugc" href="https://civarchive.com/articles/1406">Male Model Masterlist</a>.It seems to work well at a weight of 1, but can go down as well. I also recommend using the trigger token to increase the impact, such as (bodypaint:1.15)Use general terms like "superhero bodypaint" or specify characters, use concepts like country flags or simple colors, it seems to be somewhat flexible. Not everything is promised though. I'm sure you can think of designs I haven't even begun to test!Backgrounds can be adjusted, default will put them in festival/group street settings mostly.When I retrain it, I will definitely include more gender diversity, and make an effort to caption the colors/concepts in more detail. Consider this unfinished, and any and all feedback and results would be appreciated. Also of note, I haven't tested this on anime/2.5D too much, but it probably will work.This is trained on the 1.5 base SD model, at 768px.I am not responsible for your use of this resource, which is not for commercial use and is for research purposes only.

Files with SHA256: 8b219ac1a11ec426b4f83fc72a3ff5dc28954d6bc4c99f72594415d862ff9312