Why are there no AI's that can generate images of porn tailored to a user, and how can we get this done?
Things that could be used:
youtube.com
Why are there no AI's that can generate images of porn tailored to a user, and how can we get this done?
Alexander Brooks
Other urls found in this thread:
Wyatt Williams
The highest form of AI we currently have is NN's. Please explain how you would train a NN to generate a realistic human, then after this is completed they NN would also need to understand how intercourse is performed.
After you've given me a pseudo-logical explanation of that I will attempt to show you how much computational power you would need to render said film in any decent timeframe.
Jace Cooper
I said images not video, progressive GANs could do this.
Caleb Garcia
bump
Joseph Nguyen
bump, I'm sure you guys know how profitable this could be if this gets developed further.
Eli Gonzalez
The first step is to compile a reliable dataset of actresses. Any online open stashes classified by body positions, activity, etc?
Juan Perez
couldn't you just chuck in a random dataset of porn in a gan then let users rate the output progressively
Nolan Barnes
>still no app to remove bikinis from girls
Why, God, do thee let us suffer?
Carter Ortiz
How much would you pay for this, I'm capable of making and it sounds like a good idea
Xavier Hall
OP here, I'd pay $15 for an android app capable of this but i'd prefer to watch ads for a free version.
Matthew Sanchez
>porn is about visuals and nothing else
How can we get ai to make people that think this into fertilizer?
Ryder Sullivan
>porn is about visuals and nothing else
where did I say that and why does this relate.
Anthony Walker
Just toss in a nn that matches porn audio either from database or generated, that's a secondary problem
Joseph Bailey
I made a discord for everyone
discord.gg
Dominic Martin
Actually xhamster is already doing this
Ian Mitchell
link or bullshit
Eli Walker
I've thought about this. You'd need to train it for a long time, just like everything else. It'd basically be just like the google AI. That means lots of input.
Get google to do porn captchas and you are one step further.
>select all BBC
Joseph Sanders
Kazakhstani scientists did this a long time ago.
They called it VAGINE learning.
Jeremiah Flores
>he watches 3dpd p*rn
>he can't fap to visuals alone
James Peterson
>3d
?????
Mason Lee
It's really not secondary. I need long, specific storylines to get off
Nolan Watson
Okay that's worse then
Evan Howard
>train it to be able to develop stories for existing fictional characters which could realistically play out in their respective work of fiction
Ian Peterson
As of now the most advanced GAN architectures are still very limited in recreating generic outputs. Porn as this site demonstrates, can go in many many directions and has to many free variables to just throw data at and hope for the best. What can be done with current systems would be 'porn analogies'. Essentially start with a well defined base expectation and generate modest permutations from it. Like a POV cowgirl shot, or a blow job scene from a specific angle. Once we have a good archive established we can train the network to distinguish feature such as the race of the girl, her breast size, hair color, foreskin on the penis, etc which we can then arbitrarily permute.
Leo Morgan
Rumours are that this was the original intent behind GIMP, specifically a rather chiffonous swimsuit on a certain 90's supermodel.