Jump to content

Welcome to FutureTimeline.forum
Register now to gain access to all of our features. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, post status updates, manage your profile and so much more. If you already have an account, login here - otherwise create an account for free today!
Photo

Synthetic Media & Deepfakes News and Discussions

Artificial intelligence synthetic media deepfakes media synthesis GANs artificial imagination image synthesis natural language generation automation deep learning

  • Please log in to reply
222 replies to this topic

Poll: Synthetic Media Poll (10 member(s) have cast votes)

When will deepfakes be perfected?

  1. 2020 to 2022 (1 votes [10.00%])

    Percentage of vote: 10.00%

  2. 2022 to 2024 (3 votes [30.00%])

    Percentage of vote: 30.00%

  3. 2024 to 2026 (2 votes [20.00%])

    Percentage of vote: 20.00%

  4. After 2026 (4 votes [40.00%])

    Percentage of vote: 40.00%

Vote Guests cannot vote

#161
wjfox

wjfox

    Administrator

  • Administrators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 12,160 posts
  • LocationLondon

Was a Banner Hung from Trump Tower Saying He ‘Betrayed and Murdered Kurdish People’?

 

Published 9 October 2019

 

On Oct. 9, 2019, a video supposedly showing a large banner reading “(Trump) Betrayed and Murdered Kurdish People” hanging from top of the Trump Tower in Las Vegas (while off-screen protesters chant “Not my president”) started making its way around social media.

 

This video does not show a genuine banner hanging from Trump Tower, but rather a digital creation from @PaulLidicul made in response to President Trump’s decision not to aid American-allied Kurdish forces in Syria during a planned Turkish military campaign.

 

https://www.snopes.c...er-kurd-banner/

 

 

HqfDF2E.jpg



#162
Erowind

Erowind

    Anarchist without an adjective

  • Members
  • PipPipPipPipPipPipPip
  • 1,426 posts

/\ I've been continuously amazed at how even shitty cgi can fool people. My mum bless her heart is in her mid 50s now and she thought the Cats movie trailer had real actors in costume before I pointed it out. A trained eye can usually tell if something is cg barring the most cutting edge render but the laymen, especially older laymen, apparently can't. 



#163
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

BigGAN used to generate a painting of a church

02whs8g5y0q31.jpg


And remember my friend, future events such as these will affect you in the future.


#164
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

https://en.wikipedia...Media_synthesis

I've done a tad bit more work on this, so the skeleton of the thing is there. I don't know whether I should immediately go to GANs and Ian Goodfellow or whether I should write a paragraph about pre-GAN media synthesis first. 

Then comes the task of filling out the other sections. For those, I'll probably copy and paste a bunch from their original respective articles.


And remember my friend, future events such as these will affect you in the future.


#165
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

GPT-2: 1.5B Release | It's finally out!

As the final model release of GPT-2’s staged release, we’re releasing the largest version (1.5B parameters) of GPT-2 along with code and model weights to facilitate detection of outputs of GPT-2 models. While there have been larger language models released since August, we’ve continued with our original staged release plan in order to provide the community with a test case of a full staged release process. We hope that this test case will be useful to developers of future powerful models, and we’re actively continuing the conversation with the AI community on responsible publication.


And remember my friend, future events such as these will affect you in the future.


#166
starspawn0

starspawn0

    Member

  • Members
  • PipPipPipPipPipPipPipPip
  • 1,920 posts
GPT-2 can be hit-or-miss, but when it hits, man is it scary-impressive!:

https://mobile.twitt...773448999034880

[Prompt?] John F. Kennedy was just elected President of the United States after rising from the grave decades after his assassination. Due to miraculous developments in nanotechnology, Kennedy's brain was rebuilt from his remains and installed in the control center of a state-of-the-art humanoid robot. Below is a transcript of his acceptance speech.

[GPT-2 generation?] I've said before that humans have a very unfortunate legacy of being the most powerful, violent species in the world, but we are slowly starting to correct that. I, now a robot, am a living testament to our species' potential. And to prove it, I will be inaugurating the most revolutionary program of social justice in history.

Now, I realize there are people who think that the fact that I'm a robot makes me unfit for this office. But, as the President of the United States, I don't have to prove my humanity to anyone. I'm not going to debate about whether or not my brain can be programmed to think complex thoughts. I can only say that it's the greatest privilege of my life to have my brain wired directly into the control center of a state-of-the-art humanoid robot, giving the people of this country the ability to give us all a voice.

America is a nation built on the foundation stones of equal opportunity, so it's only right that the President's head be a physical reminder of this country's commitment to equality. It's only right that the first human president to run for office be a robot.

My head is the first presidential head on display at Mount Rushmore, and for the next four years, I plan to keep it there permanently.

I look forward to continuing my work on behalf of the American people as a robot. I hope to see America united as it was on January 20th of this year.

Thank you, and God bless the United States.


Got a little surreal towards the end. NVIDIA's MEGATRON Language Model seems to have less craziness like that.

#167
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

James Dean to be 'resurrected' for new Vietnam war drama

Rights have been acquired to digitally reanimate the actor, who died in 1955, so he can star in his fourth movie

Hollywood icon James Dean is due for a cinematic resurrection after it was announced the actor has been “cast” in a Vietnam war drama about military dogs.
According to the Hollywood Reporter, directors Anton Ernst and Tati Golykh have acquired the rights to use the image of the actor – who died in 1955 – from his family in an adaptation of Gareth Crocker’s novel, Finding Jack. The novel follows a soldier who bonds with a labrador called Jack during the conflict, but is ordered to leave the dog behind, along with thousands of others, when the US military pulls out.
The film-makers will use pre-existing film and photographs to create a “full body” CGI version of Dean, with another actor providing his voice.
 
Ernst told the Hollywood Reporter: “We searched high and low for the perfect character to portray the role … which has some extreme complex character arcs, and after months of research, we decided on James Dean.”


And remember my friend, future events such as these will affect you in the future.


#168
Zaphod

Zaphod

    Esteemed Member

  • Members
  • PipPipPipPipPipPip
  • 849 posts
  • LocationUK

Deepfake Roundtable: Cruise, Downey Jr., Lucas & More

https://www.youtube....h?v=l_6Tumd8EQI

 

These Deepfakes are getting even better and more elaborate. They are going to be indistinguishable to real life in just a handful of years. 



#169
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

https://en.wikipedia...Media_synthesis

 

This draft is actually decent! It's at least publishable.


And remember my friend, future events such as these will affect you in the future.


#170
Outlook

Outlook

    Arab Muslim

  • Members
  • PipPipPipPipPipPipPip
  • 1,384 posts
  • LocationBarbary Lands

https://en.wikipedia...Media_synthesis

This draft is actually decent! It's at least publishable.

Good luck, it was a good read.


Outlook's secret song of the ~week: https://youtu.be/Gnyr3sbdKkU

#171
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

Look at this upscaler!

2bajwbm37h241.jpg?width=384&auto=webp&s=


And remember my friend, future events such as these will affect you in the future.


#172
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

Turning anime into 60 FPS


And remember my friend, future events such as these will affect you in the future.


#173
Alislaws

Alislaws

    Democratic Socialist Materialist

  • Members
  • PipPipPipPipPipPipPipPip
  • 2,106 posts
  • LocationLondon

If they keep improving the AIs then animators can steadily do fewer and fewer frames until eventually...



#174
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

AI Dungeon is back online! It's an (almost) infinite-generating story.

https://play.aidungeon.io/


And remember my friend, future events such as these will affect you in the future.


#175
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

A Very Unlikely Chess Game [GPT-2 trained to play chess.]

Last month, I asked him if he thought GPT-2 could play chess. I wondered if he could train it on a corpus of chess games written in standard notation (where, for example, e2e4 means “move the pawn at square e2 to square e4”). There are literally millions of games written up like this. GPT-2 would learn to predict the next string of text, which would correspond to the next move in the chess game. Then you would prompt it with a chessboard up to a certain point, and it would predict how the chess masters who had produced its training data would continue the game – ie make its next move using the same heuristics they would.
Gwern handed the idea to his collaborator Shawn Presser, who had a working GPT-2 chess engine running within a week...


And remember my friend, future events such as these will affect you in the future.


#176
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

"Through Fields" Animation Created using GauGAN


And remember my friend, future events such as these will affect you in the future.


#177
Outlook

Outlook

    Arab Muslim

  • Members
  • PipPipPipPipPipPipPip
  • 1,384 posts
  • LocationBarbary Lands
That felt like a dream. It's so beautiful.
Outlook's secret song of the ~week: https://youtu.be/Gnyr3sbdKkU

#178
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

StyleGAN2: Near-Perfect Human Face Synthesis

https://github.com/NVlabs/stylegan2
https://arxiv.org/abs/1912.04958

Analyzing and Improving the Image Quality of StyleGAN
Abstract

The style-based GAN architecture (StyleGAN) yields state-of-the-art results in data-driven unconditional generative image modeling. We expose and analyze several of its characteristic artifacts, and propose changes in both model architecture and training methods to address them. In particular, we redesign generator normalization, revisit progressive growing, and regularize the generator to encourage good conditioning in the mapping from latent vectors to images. In addition to improving image quality, this path length regularizer yields the additional benefit that the generator becomes significantly easier to invert. This makes it possible to reliably detect if an image is generated by a particular network. We furthermore visualize how well the generator utilizes its output resolution, and identify a capacity problem, motivating us to train larger models for additional quality improvements. Overall, our improved model redefines the state of the art in unconditional image modeling, both in terms of existing distribution quality metrics as well as perceived image quality.


And remember my friend, future events such as these will affect you in the future.


#179
Yuli Ban

Yuli Ban

    Born Again Singularitarian

  • Moderators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 22,013 posts
  • LocationNew Orleans, LA

Just checked it out again. Apparently Write With Transformer has the 1.5B parameter GPT-2 as well as XLNet.


And remember my friend, future events such as these will affect you in the future.


#180
wjfox

wjfox

    Administrator

  • Administrators
  • PipPipPipPipPipPipPipPipPipPipPip
  • 12,160 posts
  • LocationLondon

AI in the adult industry: porn may soon feature people who don't exist

Fri 7 Feb 2020 15.11 GMT

A professional headshot of a man with his face carefully poised in a neutral expression, the kind that might be used in an acting portfolio. A teenage girl with red hair and glasses, pouting at the camera against an outdoor backdrop. These photographs are the sort that saturate our online world, ones that you might find on a Facebook profile or LinkedIn page. The only difference? These people don’t exist. They are the product of an algorithm, a network of images competing against each other to create convincing fakes – and experts believe that they could soon replace pictures of real people in everything from the profiles that we match with on dating apps to the bodies that we watch in porn.

Designed by former Uber engineer Phil Wang, thispersondoesnotexist.com makes use of a code called StyleGAN (short for generative adversarial network). Wang has utilized this code to create a seemingly endless stream of faces.

“Our sensitivity to faces, when you really think about it, is a product of evolution for successful mating,” he told me. “What the site really demonstrates is that even for a data distribution that we are so well crafted to understand as human beings, the machine can pick apart all the relevant features and recompose them in a way that’s coherent.”

Wang’s innovation is fascinating and seemingly innocuous and yet it shares the same technological basis as much more sinister creations. Over the past few years GANs have been almost endemically misused to create malicious content, such as material mapping the faces of celebrities on to existing, often pornographic, footage – known as deepfakes.

“A recent study found that 96% of all deepfake videos were pornographic and in many cases are being used to harass and terrorize women,” says Rachel Thomas founder of fast.ai and expert in applied data ethics. “In general, our legal system has been slow to catch up with addressing [this kind of] sexual imagery and the use of AI is deepening and accelerating this problem.”

Read more: https://www.theguard...-who-dont-exist







Also tagged with one or more of these keywords: Artificial intelligence, synthetic media, deepfakes, media synthesis, GANs, artificial imagination, image synthesis, natural language generation, automation, deep learning

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users