by hyapet » Thu Oct 30, 2025 7:28 pm
Yes and no.
What you're seeing is a lot of "retouching" apps that can take a face and "young-ify" it. So, I could upload a picture of yourself, and then after like fifteen seconds or so, an image pops out where all of your wrinkles are gone, your complexion is absolutely perfect, and the light in your eyes shine in the most noticeable manner as well.
These aren't apps that "recreate" the whole picture - they are apps that "uplift" an existing picture - and it's not just Gal Gadot - but everyone out there that are using these. In fact - they've been using them for years before AI. Every Smartphone that takes a picture automatically makes anyone that's legitimately 75 years old look like they're 35. And anyone that's below 60 as if they're 18 to 25 years old.
The problem is - the AI isn't applying new and unique techniques to every phone it "uplifts" - rather, just the same technique over and over and over again. Meaning, that perfect application of light and shading on the picture you just uploaded? It's the same as every other picture that gets uploaded (essentially). What this does is it creates this kind of "sheen" where - having grown accustomed to what it is and what it looks like - everybody can now tell when a person's photo has been "uplifted by AI." And, in turn, the desirability of such photos has fallen.
There's a term for it - and it's called "AI Slop."
Now - it gets even worse.
What's happened is that most of the early models of AI that were created were made by copying and studying and diagnosing real photos. These produced very life-like results. Actually, a bit too life-like.
Many of the same actors and actresses that didn't appreciate people stripping them down or DeepFaking them were absolutely pissed that now people could put their picture into an AI square and, essentially, have a nude created of them that looked, essentially, life-like. So, what they did was put pressure on all the legal apparatuses of the United States (where most of this stuff is coming from) to ban these programs from being able to, well, essentially function.
The AI creators got scared. And so, what they did as a result, is they started to lean their models more heavily into the "cartoon" aspect of things. Now almost everything the latest models produce looks like an early 2000's CGI puppet. They've completely nuked their own models to make them less capable of creating life-like skin or poses or, especially, faces.
Even worse? AI cannot train on AI. Meaning, all of the images it made originally and flooded the Internet with (that looked super life-like), make for extremely poor training material. But train on these images it does anyways. So, between the nerfing of the programs by the developers themselves, the corrupted pool of data that the training models get their data from, and the fact that most people have developed an "uncanny-valley eye" for all things AI, the popularity and capability of image generation has largely fallen off a cliff.
Add to this - there are "safety measures" that have been installed - to make sure that anything that approaches the face of a person produces absolute garbage as a result.
I mean, the things AI image programs are pumping out now are remarkably worse than the very first images it produced when it first started. That's how bad things have gotten. You tell an AI image generator to fix somebody's face - and it cannot even render a 3D image. It makes stuff that the original 3D modelers in the 90's would absolutely wince at.
So - AI capability is still capable of doing these things - but it's all underground now. You have to actually run set-ups and mainstay programs on your machine - and then get legacy libraries from the pre June or July 2024 - and then recreate a completely backwards compatible environment to produce things that are as good as they were 18 months ago.
But that's images.
And video technology wasn't that far along yet at that point. So, finding an AI video making program that can operate on the level of mid-2024 AI image creation in a NSFW manner in this day and age? I don't want to say it's impossible - but I haven't heard of or found one yet.
Most video creation programs can only do 10 second clips anyways. And then it looks like that "AI Slop" I mentioned earlier.
Those who have private set-ups in their homes (and I'm talking top-of-the-line computers here) that run the latest available AI video producing programs - AND - have the old-school libraries of super realism installed - even they can only, really, produce a minute long clip at maximum, before elements just start deforming from one another.
Only Google and ChatGPT and probably a few Chinese firms have come up with the capacity to create long-form video content - and that stuff just isn't readily available in the NSFW space yet - and probably won't be for a long time. You have to remember - putting real people's faces on content that is NSFW and making it look 100% real - is the exact recipe for folks to start suing the companies or for governments to get involved and say, "You can't handle this yourselves, so we're going to get involved."
It will always exist underground - but to create such a thing - will take a lot of money, a lot of time (and training), and then a lot of patience to make what it is you desire. And then - and even then - when you've accomplished all of this - and you've got the most realistic thing that AI can possibly produce - your brain is still going to rebel against it with the "uncanny-valley eye" - in that, you'll just be able to feel that something is off. And really, you won't be able to do anything about that.
So AI video generation is and will probably remain for some time ... completely unsuitable for your needs. That's your particular needs. Of wanting a (somewhat realistically) aging porn-star to be grafted onto what legitimately looks like a real body.
Which brings us to the DeepFakes.
Which is that whole "node" system I mentioned. That transfer the face and keeps the "AI Slop" element out of it entirely. So, find a Premium Bukkake video that looks like Lucy Lee's body, or the body you would like Lucy Lee to have, and then have someone transfer the face over.
And for that - it wouldn't be that difficult to accomplish this. Why?
Because the face doesn't do too many twists or turns during the shoot? The camera isn't going crazy. The scenes aren't changing everywhere all the time. It's essentially just the girl looking straight at the camera for fifteen to thirty minutes straight while receiving up to a hundred shots of glorious goo all over their faces. That straight-forward not-moving effect would make creating the in-video model a lot less time-consuming, I would guess.
So, in that sense, it's probably doable.
What needs to really happen is for AI to be able to complete the "node masks" quickly on models. Once that happens - DeepFakes will really be able to take off. Once they teach AI to essentially create a node-mask of a girl's face by just looking at it - they can probably go and create a swap of any two actresses in the timeframe of an afternoon. I don't know - maybe it's already there - but I haven't seen anything to suggest that it is.
DeepFake would be the way to go.
AI, for the most part, has turned into complete slop.
We're probably going to have to wait for a Chinese company to come around with the more authentically realistic face-model system yet - to just have the entire North American market go over and use that stuff instead - at which point, you might see things start moving the other way again. Until then, however, relying on AI is ... well, it's passable - it's great even - but, you have to understand, things do change with time.
The face that looks great today - might not look so great in a week from now - and might look like a clay model abomination in half a year's time. Once your mind begins to pick up the tricks needed to discern that these photos aren't actually real - they in turn become a lot worse. It depends on how much time you've spent looking at them.
But, regardless of how much time you've spent looking at AI photos, with every passing minute that you look at it, in a culminative manner, it's just going to get worse.
Who knows - maybe the sheppards of online AI image generation will improve their models to the point where there will be a breakthrough in quality - and the "uncanny-valley" feeling will reset itself. Either way, though.
DeepFakes are most likely the way to go.