Trained mostly on food scenes, at restaurants, but can do paperwork and reading too.
As far as lewds go? It can do flashing and implied masturbation pretty well.
For non lewds? a lot. booth seating, restaurant, library, etc. Anywhere with a desk.
Note: of the three versions, 0.08 is the easiest to use, v1 is the best imo, and v2 is a bit harder to use, but has added data in the training for x-ray and under table.
Description
epoch eight version, it's still training, so depending on how future versions turn out, I might upload them soon.
FAQ
Comments (11)
note! this works well for sfw scenes, but the focus is more on flashing and implied masturbation,
That's original! I love it! Is there a possibility to have kinda x-ray vision to see masturbation with feet for example?
possibly. I'll have a longer trained version of this one up in an hour or so, maybe, depending on if it's actually better or not.
But after that, I can see if adding x-ray is possible.
@emberbyte I think you understood, but just to be sure. I meant in this LoRA, xray vision to see beneath the table. Thanks for your reply!
@Gaspatcher Yep! I gotcha. I'm not sure how it'll turn out, but I just added a bunch of x-ray through table and desk images to the data set. Some are pov across table, but most aren't. Lets' see if the AI understands!
@emberbyte @Gaspatcher Why not just turn table in to glass 🧐 https://civitai.com/images/551318?period=AllTime&sort=Newest&view=categories&modelVersionId=51048&modelId=46429&postId=157731 Don't mind the fingers, was just fast test to test if it works
@potatovision842 @emberbyte If you can turn it while still being suggestive, fine by me (for example, making glass just at the action location and solid on the corners, idk if it's possible, but it indicates to the viewer what is happening below while he understands everything is still solid). I don't think it's as exciting if you do it under an actual glass table in public.
@Gaspatcher check out the v1 beta, it's not quite what you were talking about, but it kinda works. It seems best at doing a photograph of what's under the table. very gacha.
@emberbyte Nice! Yeah the two pics (#4 and #9 for me) in a classroom are really close to what I want. Do you think we can generate first prompt with v0.08 then inpaint the section we want with beta version? I kinda think we can have more control on it this way. I'm gonna try this and share if I get a good pic.
@Gaspatcher that could work! You might also want to check out the v2 I just uploaded, it's harder to control, but I actually added x-ray and under table to the data sets; I probably should have left out the under table, not sure.
Not much for sample images right now. I've got something else training, so I just threw in what I had.
@emberbyte Yup, inpainting does the trick. I generated something that gives an acceptable output by:
1/ txt2img my first prompt with v0.08 to have the overall scene
2/ img2img my second prompt with v1 beta with inpaint, focusing on the "action" beneath the table
3/ img2img my last prompt with another LoRA (footjob)
Details
Files
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.