Years ago, we had apps like FaceTune, which easily bumped a woman up one or two notches. Via moving some sliders, women gave themselves bigger eyes, fuller lips, perfect skin, and a radiant glow. Smarter women, of course, would not overdo it in order to maintain plausible deniability.
Simple image filters seem to have become completely normalized. I noticed this repeatedly when bumping into female work colleagues I only knew via online video conferences. In the more egregious cases, I had women greet me randomly, and I had no idea who they were. I am not kidding. There are indeed women out there who think that in a professional context is perfectly adequate to turn yourself into a different version of yourself. I once had a female colleague who looked more like a teenager on Zoom but was menopausal in real life.
I do not think that the video filters on Zoom or Google Meet are very sophisticated, or perhaps women just tend to overdo it so they look incredibly fake. However, there is a new generation of image filters that you can’t really detect, at least if women do not fully enter clown world territory. They are based on algorithms creating a three-dimensional mesh of your face, basically digitally recreating your face. TikTok calls it “3D Face”. The filter then modifies this recreation, and the output persists. This means that the filter is so good that it does not even trip up if you move an object in front of your face. Check this out:
The chick claims this filter uses a particular kind of neural networks (“GANs”), and so does this article, which quotes an artist, not a computer scientists. TikTok does not say that it is using GANs, and I don’t think it does either. The explanation given does not convince me:
“Simply put, GANs pit two competing neural networks against each other in a fist-fight to the death,” Hurd says. In the case of Bold Glamour, it’s a competition between the camera’s view of your face and the style TikTok wants to morph you into. “Because it uses you, it then compares aspects of your face to a dataset of images that start to match against your cheeks, eyes, eyebrows, lips, and more.” Eventually, the technology combines the two sets of images into one. “If we do this fast enough, we can achieve a video framerate,” Hurd says. “And now we have a next-level effect like Bold Glamour!”
Let me geek out a little and point out that GANs are not fast enough for this. The effect is instantaneous and works perfectly fine on phones. To quote the article again,
“This has already been possible on desktop PCs using specialist software of course (e.g. deepfakes),” Akten says. “But that requires specialist software and some technical understanding. TikTok’s filter runs on a mobile device, accessible by billions of people, with no technical knowledge required, in realtime, and often (not always) looks totally believable.”
Sorry, journos, artists, and professors, but you need a pretty beefy computer to produce real-time deep fakes. Phones have advanced tremendously over the years, but they do not pack the power of an NVidia GPU that requires a 750W power-supply unit. The alternative is that TikTok’s Chinese engineers are running circles around the diversity hires in US BigTech companies. This is also quite possible. Anyway, this discussion is irrelevant. The technology exists, it can easily be used, and it is very good.
The practical consequence for any guy is that on top of not being able to trust that any picture a woman sends you is a reflection of what she looks like, it is now also the case that any video, including real-time video, may have very little to do with her looks in real life. Currently, there are surely a lot of women out there who make a lot of money via deceiving men online, be it via OnlyFans or dating websites, but this gravy train is bound to come to an end because AI will soon be able to eliminate the middle man, er, woman, completely. Why would you need the face of some ugly chick if you only digitally manipulate her face anyway? You can probably turn a man’s face into a female one just as well, and voices can already be modified very convincingly in real-time.
In the end, whoever is able to produce the sexiest, yet still plausibly realistic-looking virtual woman will win, and it could well be that the market will be flooded and split among too many competitors. The same happened with modelling, porn, and music. We no longer have super models. We no longer have porn stars. We no longer have pop music super stars. Soon, we will probably no longer have OnlyFan sluts making millions a month either because there will be thousands of AI-powered virtual whores that all look great but in different ways. Thus, whatever some basement coomer is into, he will find his perfect online make-believe girlfriend to spend money on, and the money will be spread a lot more evenly than it is nowadays. I wonder how leftists will react to this manifestation of diversity, inclusion, and equity.