DeepFake sucks

I think that my test before-and-after speaks for itself. I used Candice Swanepoel as my subject because nude photos of her already exist online. My intention was to compare DeepFake's output to those photos to gauge its accuracy. However, given the result...well...

Attached: baddeepnude.jpg (838x512, 254K)

Other urls found in this thread:

xvideos.com/video47699609/kpop_aoa_seolhyun_getting_a_hot_cowboy_ride_during_christmas_-_deepfakesporn.com
imgur.com/ILOcR9Q
deviantart.com/mell0w-m1nded/art/Twilight-And-Raven-367888655
filesDOTcatboxDOTmoe/u8k8ue.7z
youtube.com/watch?v=kSLJriaOumA
twitter.com/NSFWRedditGif

So post the deepfake picture then, but only after i jerk it to the legititmate nudes you have provided

t.discerning masturbatory-entheusiast

Have sex

FakeApp is free and infinitely better.
xvideos.com/video47699609/kpop_aoa_seolhyun_getting_a_hot_cowboy_ride_during_christmas_-_deepfakesporn.com

Bee tee eff oh

Oh neat it's another piece of shitty technology overhyped by retarded journalists

The 20somethings who wrote the pearlclutching articles on this are the exact same kind of idiots as their parents who would spread horror stories about violent video games without realizing they were providing the best possible advertising for them. I saw a twitter post where one of them were claiming their male staffer was a "hero" for being the one to install and use it for research purposes. Zero self awareness

violent video games are harmful though

Dude are you 12? It’s not a magic software that ACTUALLY let’s you see under women’s clothes. It uses a library of nudes from which to simulate nudity.

So is drinking fluoride and refined suggars. Fuck off.

Dilate

dont use an incompatible image then retard. it can have some very good results in some cases, and no matter what will always require some post work to make it look decent

so if you run it on a picture of a guy does it turn him into a post-op tranny?

please post the why yes chad meme man cos I wanna meme my friends n google just wont give me his sacred image no matter what I search

Isn't it just some algos to draw nudes from a porn database?

Huh wtf is this, where can I download it?

>xvideos
Trash 140p quality LMAO

Just use the adultdeepfakes instead, better quality and its in mp4 format so you can just right click/save the video.

I don't comprehend why you people are so interested in this
It's just sad

or you can read the instructions
imgur.com/ILOcR9Q

You really can't even follow 3 steps?

Attached: .png (567x617, 62K)

Don't matter
I can fap to this

You can replace a persons face in a video by using FakeApp. You need many pictures of different angles of the person you want to replace the face with.

where can i try this thing?
do i need to install it? do i need to install gentoo first?
does it work with cp? is such a thing even legal?

Right, it works good, OP is literally retarded

it kinda sucks
how is that nvidia's stylegan create faces virtually indistinguishable from real ones while this shit wouldn't fool anybody? human beings are supposed to be very good at analyzing human faces

I want to make nudes of my oneitis, don't care about videos just stuff like the OP

I agree that it's trash, unfortunately. I'm waiting for something better to replace it.

I WILL KEEP DRINKING refined sugars and fluoride. Here's a pic of PrOgRaMmInG SoCkS

Attached: 1535514601262.jpg (3264x2448, 1.73M)

your trapitis ;)

Because it was made by some random guy using some random free datasets. If FB used their porn detector to make a GAN it would be better.

kek

fuhgive ya motha and come back to ya fatha

I used it on underwear pictures of my ex and it worked extremely well, it actually looked so much like her it made my heart skip a beat and I got sad lol I'm such a faggot

>dat vag

Attached: horrifying.png (504x579, 253K)

its a trap

This needs a more modern GAN implementation and a better dataset.

Where do we get a better dataset from Jow Forums?

Sites like Czech casting have very good frontal nudes

Does it feel comfy when your cock rubs against that soft underwear?

>step 1: take selfie
>step 2: use FaceApp to genderswap your face
>step 3: use Deepnude to generate tits
>step 4: fap, post results, and let everyone know how good it feels to be objectified

I'm not exactly up-to-date on the latest ML techniques, but don't you need a pic of (roughly) the desired input, and a corresponding pic of (roughly) the desired output for the NN to train on? I.e. you would need a pic of a woman in some amount of clothing in some pose in some lighting from some camera angle, AND you would need roughly the same picture with the same everything except they're naked for the NN to learn a good approximation.

Correct me if I'm wrong, this stuff develops rapidly so maybe new architectures and training procedures can alleviate this need.

Deepfake almost makes the age of political disinformation and social upheavalcoming from this technology worth it. I wonder when the first major fucked up widescale use of this technology happens.

Photoshop? We had that shit in the 90s and unlike this stupid app it worked.

How do you install teh crack?

Attached: crack.jpg (997x652, 98K)

remove two original files from prog folder, replace with crack

Where's the download tho

Pix2pix is still state-of-the-art.
Dataset you'll need to create yourself. Good luck.
Congratulations, now get 600k more frontals for good-ish results. If you want your app to be usable, you'll have to get another 800k for each angle.

where do you insert the code

Anyone got a linux version ? I've seen tons of torrents and megas featuring win$h1t versions (which really does emphasize that summerfags have indeed invaded Jow Forums) ?

I've seen decent results on some threads on /b/ and all, and terrible stuff by nignogs who can't even read 3 fucking instructions, so I'd like to test the accuracy for myself.

Yeah, it's pretty bad. Don't know why people are hyping it, just use photoshop

FakeApp was deprecated like a year ago

>How is it that nvidia, a multimillion dollar company specializing in graphics acan do something better than a horny 22 year old in using python?

Why some pictures shut the software after clicking on transform?

Attached: apu sieluni satuu.jpg (793x786, 54K)

inside the BUTTOCKS

I got a warning 2 days ago for posting pic related.
Somehow you can get away with a dismorfed body showing tits and pussy for almost 12hours.

Attached: 1558037333839.jpg (750x702, 96K)

There must be no special symbols in the filepath. Only English.

why would linux furries need it? it doesn't work with animals

anyone have a download link for this shit

The top is technology, that shit is frogposting

It works all right.

Attached: deepnude.png (512x512, 295K)

Use /b/ to generate the required images?

I have also had the idea of building a more robust solution. Is there a discord or something for us to collaborate on this?

Attached: 1555664202655.gif (395x520, 1.69M)

that pic...............

Attached: 1561580130582.jpg (887x1098, 88K)

When these programs can top a good photoshop editing job then they'll be worth something

trojan

Mine litterally crashes with every picture as soon as I hit transform. Tried zooming unzooming. Tried with successful pictures already posted on /b/ threads.. is this because i'm using w7?

Keeps happening to me too

How do I do this dammit

t. Jow Forumsbedwetting

It's even worse when you use it on a big titty girl.

>random cunny and boobs
>"AI" to identify cunny and boobi location
>past the results from the list to the selected locations
pajeet technology

Yes, it needs win10.

>nobody saved the Linux version

Attached: 1554134788057_1.png (348x233, 76K)

tried it a few times it doesn't work properly at all
shit sucks

What happened to OP's pic?

Attached: sigh.jpg (237x359, 18K)

Kill yourself JustPassingThrough.
KILL YOUR FUCKING SELF.
deviantart.com/mell0w-m1nded/art/Twilight-And-Raven-367888655

Attached: Bronies, this is your mindset.png (307x157, 28K)

I have linux verion, but i'm on the phone now.

would you share, maybe we can get a nice support group to keep it alive.

>snatched it and set up a VM to use it
>actually kinda works
>realize none of my hot friends have any good bikini pics
damn

Kill yourself JustPassingThrough.
KILL YOUR FUCKING SELF.
deviantart.com/mell0w-m1nded/art/Twilight-And-Raven-367888655

Attached: Bronies, this is your mindset.png (307x157, 28K)

Thanks bro i thought it was the brightness or something. I suffer in Germöney

kys

its ok

Someone explain this meme to me.

removed for having a vagina
only took 2 days

This is the cracked premium linux version without the libs (cm.lib, mm.lib, mn.lib) which can be found elsewhere. You need to put all three aforementioned lib files into .../main/pyqtlib/, and change the preferences.json file so that the values of "init" "library" and "transform are all yes. If you use this download, there is no watermark.

That being said, it's shit software. The app is GUI only with no real options, and still crashes randomly because it can't handle image sizes that deviate too far from square. It can only nudify one woman per image, so you need to either mask or crop out other individuals. The crackers have only decompiled files related to enabling premium and removing watermarks, so if this ever sees improvements, it'll be a while.

My dumb ass forgot the link: filesDOTcatboxDOTmoe/u8k8ue.7z

Thanks dude. Let's keep this thing alive.

Attached: 1558471835589.png (202x126, 12K)

How about making general deep learning threads, so we could hide our deepnude stuff there and progress on making it even better?

Attached: 3.jpg (1024x1024, 569K)

kys

So a bunch of people who can't figure out how to use a search engine can discuss how to correctly use cd in a bash shell? Hard pass.

>How about making general deep learning threads
I'm absolutely behind this but I think that we should make them on /sci/ since they talk about AI all the time

Nice, I'm definitely interested in contributing to a deep nude improvement project, whether that's via disassembling deep nude, or creating a clone from scratch.

out of curiosity what does the dataset look like? Is it clothed / nude comparisons used to train the network?

I think that a lot of people are as I've seen many post the same post in other threads as well, and it's about time for Jow Forums to start doing something useful because think about it, there's thousands of CS and Math majors browsing this place with zero coherence and unity between themselves, imagine what would happen if you gave them a common goal and pooled their minds towards it

Well, I made the thread. Let's migrate then. >>/sci/10769848

Fuck, why formatting so hard these days

you kinda forgot to capitalize and should have also named it /DLG/ - Deep Learning General plus put links to the latest news so people have something to discuss, but this can work as well
As for the next general that you make, you should know the usual rules of generals:

>well-formatted constant subject with an acronym like /DLG/ so it's easily searchable from the catalog
>an image that is related to it, webms are better especially if it's a webm of something cool like this youtube.com/watch?v=kSLJriaOumA
>at least two to three links about something new related to the subject, or links to guides for it

Or simply check out the generals on Jow Forums and copy their style. And you should know that generals follow a downtrend and slowly die when nothing's happening, and suddenly burst with hundreds of new posters once something does happen. As AI is a very complicated subject, the threads will usually be somewhat dead until something major happens like another pro team being beaten in some game, in which case the threads will become the most active on the board. Repeat this for enough major happenings and those generals get enough posters who stick through the dead periods, which is the point when they will actually build an inside culture and start producing things like projects made by and for the posters from those generals. Until then, again, they will be dead and kinda useless. Don't get discouraged by the deadness.

Consider it's a pilot. If this thread will live and second one will definitely be formatted much more properly by me or someone else.

its shit like this that makes people start wondering about the JQ

So when can we expect fake porn with deepfake using artificially generated actors instead of celeb shits? I want the porn of the most pleasing face possible with the most pleasing body possible in human history generated by the computer.

kys