Deep Nude Celebs ONLY

[profish]

vanessa marcil

 

[xairalu]

Me encantó esta app porque todos los días regala un token :-D

You must be logged-in or registered for see links

 

[sarcastor]

just started a new thread for Emily Feld

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

You must be logged-in or registered for see links

 

[StableInfusion]

Brie Larson
View attachment 2549113View attachment 2549114


I trained it, based on 584 images of Taylor Swift as well as 744 other images focused on various different aspects of nudity, all images manually captioned and reformatted into a square aspect ratio. My dataset includes wet nude women, at different zoom levels. I used RealisticVision 5.1 as a base, which is derived from 1.5. This model was actually a merge between two models trained on the same dataset. I did have to lower the weight of the word wet to 0.7 though because the wetness was a little extreme without that. (still realistic, just too wet)

I take my fully trained model and I extract an inpainting model from it by adding the difference between my full model and RealisticVision 5.1 to the RV51 inpainting model. However, I only use the inpainting model to create a low resolution image I use to guide ControlNet (realistic lineart and tile, at low weights, along with my own traced version of the details I want to preserve) and then regenerate with my full model at high resolution.

I don't have a good enough computer to really train SDXL effectively, but my datasets are high enough quality to handle that whenever I end up upgrading my GPU (although I don't expect that to be any time soon as I spent a fortune on my 3080Ti when I bought it)

Well, looks like I won't be getting realistic wet skin any time soon. That's waaay above my pay grade

 

[StableInfusion]

If you're just pasting a real head onto other nude bodies (I don't know if those are AI or not) then it's not a deepnude

A deepnude is attempting to simulate what it would look like if you removed the clothes from a real photo, using AI inpainting.

To me it looks like he's attempted to outpaint using AI

 

[fudefrak]

Eliza Dushku