Read
(Use the tags Notice lines and heart if you want those fancy purple motion lines.)
I didn’t like any of the other LoRas, so I made my own- and it turned out spectacular. I also added new poses, concepts, and characters. This is probably my best LoRa yet.
If you enjoy my work and can spare some buzz, I'd deeply appreciate it.
Models To Use:
Waillustrious NSFW v11.0 – The best overall choice for this LoRa.
Noobai-Cyberfix-V2/Perpendicular-Cyberfix-V2 (VPred) – Very creative, though not based on Illustrious, so images might sometimes look washed out. However, it can be more accurate.
Hassaku 1.3 Style A – A solid option.
Negative: Use the negatives from my example images and adjust them based on what you’re trying to generate.
Important Info:
1- Check my example images and their prompts to learn how to prompt the LoRa better
2- It’s often better to avoid quality tags.
3- Change the reoslution based on what you are trying to gen, resolution vastly changes the output, i really like 1408x896.
4- try out img2img! its really great, the LoRa can gen great outputs without img2img but it can get some amazing hard-to-hit ones with it.
5- Share any good gens you get here! they might get used for a V2, and you get buzz off the likes.
New Tags: (Check the example images for prompts)
cooperative buttjob, symmetrical docking
penis on asscheek(this has three variations standing, on stomach and bent over, only standing works unless you img2img, or maybe use high strength on the tag, i might fix it in a v2 if i collect some more training data)
penis under asscheek (does not work unless you img2img on very low denoising)
New Characters: Maria Robotnik, Riley Andersen, Gabi Braun, Videl
Description
FAQ
Comments (7)
Wow, based on the gallery alone, it seems like you did a tremendous job with this model. Well done!
Thank you! Yea, this was carefully compiled over the course of 5 days, 200 image dataset all high quality, you dont know how many times i rechecked the tags... 🥴
I'm glad it turned out so well, im compiling for a V2 right now, will take a while though since im carefully selecting AI outputs to make sure nothing low quality sneaks through
And, thank you a lot for the 500 buzz tip!
@ShocksOnDeck Of course, it's nice to see really well done LoRA. I actually have some datasets that might interest you... feel free to reach out, and we can chat.
@waifuliberator hello, deeply sorry for the late reply i didnt see your notification, yea sure you can hit me up and ill take a look at em, no promises though😅 got a few projects of my own lined up, specifically one with an absolutely colossal 800 img dataset
Awesome style!
Thank you! Enjoy :)
very good












