Right now attractive females can make huge money by milking simps on Twitch, onlyfans, chaturbate, or by doing straight-up porn. This, along with welfare, makes even betabuxmaxxing difficult for us. Could technology come to the rescue ?
As I've shown in other thread, computer-generated female imagery has made huge progress lately. The state-of-the-art is currently using Generative Adversarial Networks (GAN), and is best show-cased by the site http://thispersondoesnotexist.com. I've posted a thread about it already, but here are again a few generated faces for illustration:
The competing approach is based on 3D modeling, that is when the geometry is hand-crafted and modeled with polygons and textures. The results are less visually convincing but once the model is crafted it can easily be used to make videos. It's the technique we've seen in Hollywood movies and fully CGI-productions like say those from Pixar studios.
Below was for instance the state of the art three years ago :
View: https://youtu.be/iAJgZyrTeRU
These two techniques will merge eventually. I think a first pass will be 3D modeled, and it will be fed to a GAN which will improve lighting and image details to make it look more realistic.
An other problem will be physics. To look real, a video must have realistic physics simulation, most notably for body parts, especially if we consider producing porn or more generally naked female bodies moving around. Such simulations are usually computationally prohibitive, but there is hope even in this regard, as suggested by the following recent work explained below by YouTuber "Two Minutes Paper" :
View: https://youtu.be/s8Nm_ytwO6w
I'd say we have about ten years until we see all these kinds of techniques merge and put e-thots out of easy income.
As I've shown in other thread, computer-generated female imagery has made huge progress lately. The state-of-the-art is currently using Generative Adversarial Networks (GAN), and is best show-cased by the site http://thispersondoesnotexist.com. I've posted a thread about it already, but here are again a few generated faces for illustration:
The competing approach is based on 3D modeling, that is when the geometry is hand-crafted and modeled with polygons and textures. The results are less visually convincing but once the model is crafted it can easily be used to make videos. It's the technique we've seen in Hollywood movies and fully CGI-productions like say those from Pixar studios.
Below was for instance the state of the art three years ago :
View: https://youtu.be/iAJgZyrTeRU
These two techniques will merge eventually. I think a first pass will be 3D modeled, and it will be fed to a GAN which will improve lighting and image details to make it look more realistic.
An other problem will be physics. To look real, a video must have realistic physics simulation, most notably for body parts, especially if we consider producing porn or more generally naked female bodies moving around. Such simulations are usually computationally prohibitive, but there is hope even in this regard, as suggested by the following recent work explained below by YouTuber "Two Minutes Paper" :
View: https://youtu.be/s8Nm_ytwO6w
I'd say we have about ten years until we see all these kinds of techniques merge and put e-thots out of easy income.
Last edited:
