[a / b / c / d / e / f / g / gif / h / hr / k / m / o / p / s / t / u / v / vg / vm / vmg / vr / vrpg / vst / w / wg] [i / ic] [r9k / s4s / vip] [cm / hm / lgbt / y] [3 / aco / adv / an / bant / biz / cgl / ck / co / diy / fa / fit / gd / hc / his / int / jp / lit / mlp / mu / n / news / out / po / pol / pw / qst / sci / soc / sp / tg / toy / trv / tv / vp / vt / wsg / wsr / x / xs] [Settings] [Search] [Mobile] [Home]
Board
Settings Mobile Home
/g/ - Technology

Name
Options
Comment
Verification
4chan Pass users can bypass this verification. [Learn More] [Login]
File
  • Please read the Rules and FAQ before posting.
  • You may highlight syntax and preserve whitespace by using [code] tags.

08/21/20New boards added: /vrpg/, /vmg/, /vst/ and /vm/
05/04/17New trial board added: /bant/ - International/Random
10/04/16New board for 4chan Pass users: /vip/ - Very Important Posts
[Hide] [Show All]


Janitor acceptance emails will be sent out over the coming weeks. Make sure to check your spam folder!


[Advertise on 4chan]


This was some time around 2020 (or earlier?). This was when AI generated images looked like bad dreams. You know, that time when you'd get hazy images of half formed buildings and people that looked strange or even monstrous.

Some account on the reddit nude subreddits was posting nudes of a blonde girl in her bedroom that had picrel quality. I hadn't seen this quality AI images before but they did look weirdly "perfect" to me. One person in the comments did point out that these images are not real and no one listened to him. That account didn't get too popular. The top post probably had around a hundred upvotes. But it's so weird remembering seeing this "aesthetic" back when it wasn't this normalized.

Wtf was that?
>>
unless you can actually link to something concrete its probably a false memory
>>
>>109397304
similarly, wtf was up with all the drama surrounding 'deepfakes'? even now it's not trivial to do that
>>
>>109397304
Did you save it? Can we see?
>>
File: 1653267362316.jpg (192 KB, 1637x1772)
192 KB JPG
This girl?
>>
The difference is the technology back then was all in GANs. The technology since then is a form of a diffusion model. Diffusion models can produce all sorts of novel images based on known elements. That's why you always see "koalas dunking a basketball" as the demo prompts. Because they can combine known images of koalas and basketball dunking.
GANs are more like finding the average of a set of pictures. Anything that breaks from that average is extremely blurry. There are plenty of photos online of straight-on portrait shots, so it's actually really good at that sort of generation. Because all example photos follow the same shot composition.
>>
>>109397364
whatever happened to gans? you never hear about them anymore
>>
>>109397373
The other anon literally explained it to you in clear text you bufon. I'll add :
They aren't easy to make multimodal
They are super hard to train
>>
>>109397373
Diffusion made them redundant by being better in every way.
>>109397304
Before AI we had this thing called photoshop where you could edit the fuck out of images and make things that look unreal. Every attractive woman used them.
>>
>>109397415
he only explained the difference between diffusion and gan, which doesn't explain why gan went away, and there's not as clear a boundary as you think, you could use gan while training a diffusion model
>>
>>109397304
that isn't from 2020 or earlier
>>
>>109397443
>yea, but why did people stop using the horse AFTER the car was invented????
>>
>>109397443
>you could use gan to train a diffusion model
Yes you absolute retard and you could use it to train a transformer, but it's a shit method that tends to diverge easily so getting your gradients from another NN is only used in stuff where we can't have gradients like RL
>>
File: Guava.jpg (689 KB, 1074x1676)
689 KB JPG
>>109397321
>>109397328
I didn't save it unfortunately

>>109397339
>>109397442
I'm pretty sure that's the girl! I remember the freckles. If this is edited, how the fuck did the exits capture the aesthetic of AI images that didn't exist yet? Her bedroom nudes looked like those "Samus playing Metroid" AI images. Airbrushed photos etc existed before but they didn't look like AI, they had a different kind of fakeness.

I've found people talking about it:

https://www.babepedia.com/babe/Riley_Guava

https://www.reddit.com/r/PrettyGirls/comments/r0p7vx/riley_guava/

Like everything, the girl, her background, any objects all looked like AI does today. How?
>>
File: bor5m9nisfte1.png (2.78 MB, 1024x1536)
2.78 MB PNG
>>109397544
The nudes in her bedroom had this aesthetic.
>>
>>109397544
They were probably made with Photoshop and one of those DeepFake face swapping programs.
>>
>>109397304
I remember this exact image and it was 2022, not 2020.
In 2019 and 2020 we had Dungeon AI and other stuff. We also had 15.AI which nobody remembers anymore which Eleven Labs stole from.

Wasn't Deepfakes in 2022 ? All I remember back then is that you could take somebody's face and put it IN A VIDEO.
Because otherwise taking a face and putting it in an image could already be done with any Photoshopper. Was the heal tool available in 2020?
>>
AI face swapping was popular many years before diffusion models. remember faceapp?
>>
>>109397722
Yeah I remember seeing a girl similar to this on reddit years ago, always gave me a distinct uncanny valley feeling. Pretty sure it was an AI face swap or something and nobody in the comments ever pointed it out.
>>
File: 1634607245223.jpg (174 KB, 1125x1500)
174 KB JPG
>>109397738
This girl?
>>
>>109397452
>>109397670
The pic in the OP isn't what I saw. That's a recognizable AI image from like the past couple of years. I just put ot there to show how photorealistic the ones I saw were.

>>109397592
>>109397722
>>109397738
Okay, but a deepfake would only have the face looking weird. The rest of the body would actually be real footage/photo. This girls whole body had the AI look and her background too. However, I acknowledge that it could have been an edit that captured the AI look before it even existed.

>>109397339
Dude, I'm pretty sure it was this "Riley Guava" creation. Hiw did you know I was talking about her? Are you the creator?
>>
>>109397304
naka-dashi naka-dashi naka-dashi naka-dashi
>>109397339
naka-dashi
>>109397555
naka-dashi
>>109397852
naka-dashi
>>
>>109397322
>even now it's not trivial to do that
nigga what
>>
>>109397322
It's a losing battle, one day when AR glasses will just look like regular glasses, so any random stranger staring at a girl on a train journey or whatever will be able to turn the footage into an AI replication of them at home, to do whatever they want with, like LoRA training on steroids.

On that day, all the shut-in incels will start going outside, and it'll be to catch girls they see out and about as if they were pokemon.
>>
>>109398209
you can already do this in real time with imagination
>>
>>109398686
That's the problem.
They've lost the ability to use their imagination or think.
They are less then animals now.
>>
>>109397852
Was she AI all along?
>>
You have discovered a secret of ai image gen, which is that they use porn.

the "aesthetic" is "very retouched" and "mistakes retouching"

at least SOME body horror is actually just retouching mistakes that the model learned.
>>
File: arthurs_mom.jpg (62 KB, 600x900)
62 KB JPG
>>109397555
damn she looks way hotter than Arthur's mom
>>
>>109397544
did you forget humans can get pretty good with photoshop and computer graphics
>>
>>109399083
ai was trained on porn, and this is becoming more and more obvious, ok?

krea 2 clearly was trained on budget porn.
>>
>>109397304
Fuck off.
>>
>>109397544
Looks like a face from StyleGAN (thispersondoesnotexist) + some editing work, not fully AI generated.
>>
>>109399216
No



[Advertise on 4chan]

Delete Post: [File Only] Style:
[Disable Mobile View / Use Desktop Site]

[Enable Mobile View / Use Mobile Site]

All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.