More Deep Fakes and Other Things to Look For
Wednesday, November 20, 2019
Technology and video editing is getting so good that videos that are real and ones that are fake are starting to blur the lines of reality and nearly indistinguishable.
A few months ago we noticed a few articles by the Verge that showed how Jordan Peele uses artificial intelligent or AI to make the previous U.S. president forty-four look like he was saying movie opinions and other phony fake things that he wouldn’t normally say. The article goes on to say that software can make the edges of a mouth turn up and down and also open eyes similarly to the person that’s actually voicing them much like puppetry and ventrilloquy.
These apps are open for downloading and you can probably find in the future open source software versions of many fake-making apps in the future that simply use the power of CUDA enabled GPU cards or ATI hardware acceleration. There used to be software a long time ago that could create fakes but current technology is getting faster and better at interpolation of pixels and stitching together the individual images into realistic videos. There have been various celebrities from Gal Gadot to Nicholas Cage that have had their likenesses deepfaked and dropped into movies they never intended to be in. These technologies are part Adobe video software and this “FakeApp” face switching tool. A few months ago there was even an aging app called “FaceApp” with a “C” instead of a “K” that could be used to make realistic pictures of yourself, sort of like an aged filter for Snapchat but using a different app program. That came under intense scrutiny since it was foreign made and people were worried about privacy concerns and what the app would do with your face data.
Our current world is speeding forward where digital fakes could be everywhere and some companies are trying to figure out algorithms to determine the real videos from altered and fake videos much like some of the “doctored footage” in certain conspiracy and hoax videos.
A few weeks ago we also noticed some really interesting videos by Corridor Digital doing a parody of Boston Dynamics robots that start to become sentient and rebel. It is only a matter of time before robots will be able to exceed many human tasks in every way but that is still off into the future but getting closer every day. Some of the smartest minds in AI, tech, blockchain and robotics are getting into revolutionizing machines to such an extent that you will see the next exponential wave of development hit many industries when this becomes in large part usable with applications across the board. Dr. Ben Goertzel of SingularityNet and David Hanson of Hanson Robotics and Marc Raibert of Boston Dynamics (which was later acquired by Google X) are at the forefront of all this technology.
Another noteworthy video is the Deepfake Roundtable with George Lucas, Tom Cruise, Robert Downey Jr by Collider Videos entitled “Deepfake Roundtable: Cruise, Downey Jr., Lucas & More – The Streaming Wars” which can be found on YouTube and with hashtags of “AboveTheLine” and “DisneyPlus” and “Deepfake”. You can find this and other interesting videos such as “Tom Cruise Replaces Robert Downey Jr. As Iron Man In Deepfake Video & It’s Terrifying” on Heroic Hollywood which also has a link from Collider Video to a Twitter link.
How to tell a fake?
- The lips are slightly out of sync sometimes.
- The eyes don’t wrinkle quite in the right way.
- The eyes don’t blink or look dead.
- The mouth might not close all the way.
- There seems to be a blur.
- The head or face isn’t quite in the right position.
- A body part is not in the correct proportion.
- There seems to be flickering.
- The lighting is not matching.
- A portion of the body moves or shifts in the wrong way that it never would in real life.
- A robot is also monotone often and there is not enough enunciation or stress on the right words.
- If it’s a 3D made object or person or animal it looks too smooth or perfect and there’s no textural variations or improper shading and gradients.
- Shadows are not present or in the wrong place as well as bright lights or colors bleeding or not in the correct positions.
- The shape of an object is wrong.
- An object looks like it’s just floating or covering something else.
- An object is not moving naturally like a person swinging hands when walking or with movement of legs the way an object naturally is supposed to.
- There are duplicates or several replications of the same thing that appear too closely similar such as many boxes with the same color or design.
- Colors are incorrect or it looks too dark.
- The picture is too blocky or pixellated.
- There seems to be a lot of smoke or residue in the picture.
- An object appears out of nowhere that wasn’t in a frame before.
- An object becomes stationary or motionless. An example is a Geico dance competition commercial where a dancing partner gets tossed up into the air and comes back down a few minutes later in the exact position after some gravity.
As in the movie Simone, things will get harder and harder to discern real from reality as technology gets better and our televisions and media become digital. Some kids in the future may have unrealistic expectations also of how the world works from seeing too many action movies and their resulting physics although many 3D motion software is designed to account for physics of moving objects, colliding, particles being sprayed or falling such as in water or snow. We are headed to a Ready Player One type world where millennials are already stay at home bodies due to the comfort and convenience and safety and inexpensiveness of a stay at home and Netflix night. (A later article we will cover residency and financial stability for millennials.)
Remember that every day we’re getting closer to a full fledged revolution where we’re getting more digital. Information is getting harder to weed through. And data is being captured by big companies and harvested for their own purposes. Make sure you safeguard your information also.
Speaking of fakes. There are also scams out there that use people’s pictures from Facebook and dating profiles. So if you’re into security, consider marking your photos or overlaying your images if the platform you’re uploading to allows it that says it’s only to be used for a specific purpose. An example is let’s say a company you order from needs to validate your ID. You can over lay your photo with the words “For identification at computer discount store 11/15/2019” as an example and replace it with the relevant information. That way the person that’s checking your ID doesn’t try to use it to create a fake ID which is rare but can happen on sites. A full on front picture of your headshot can be made into a badge or ID without you even knowing it. We’ve had it happen before one time for demonstration purposes and didn’t even know it at the time!
Deep fakes can be entertaining and useful in the film industry but also used in adult entertainment as well as for “revenge” which can also be bad. And deep fakes may eventually be used to commit fraud or scams.
It all goes back to being vigilant with your information and what is put out there as videos and pictures shared on Instagram, Snapchat, Facebook, and combed through histories can be one day made into an entire fake profile or picture of you complete with maybe even biometric data and fingerprints. In fact a recent article talked also about Fitbit being acquired by Google much to the chagrin of many of its loyal users. There were some people that after learning of the acquisition decided to leave their Fitbits at work according to one article. People are starting to become wary about privacy after big social media giants like Facebook and others were revealed to be data collecting agents.
After revelations of PRISM, GCHQ and Five Eyes people are starting to realize that privacy and encryption wars are being waged with one camp touting transparency to fight crime but at the slippery slope of invasion of personal data and overcollection of personal information. One only has to look at covert surveillance programs of the 50’s to 70’s for instance to see that given time programs can be abused.
So much information is out there that disinformation can also end up being spread to confuse, divide or create fear, uncertainty and doubt. This has also happened in many times through history as the person that controls the media controls the story and narrative.
Be careful where you’re getting your information. Be careful what Twitter account you follow or believe in. And be careful what link you click on or put your information in. As the Internet meme goes, it’s a trap.
Information warfare and cyber-warfare is the new fight. (See Stuxnet also.) Be vigilant. We aren’t encouraging paranoia, just to be smarter about what you consume and why.
Bonus link (11/24/2019) Check out “Interrogating Zuckerberg” by “A Bad Lip Reading”. Even though it’s not a deep fake it is still an interesting fake video that uses some dubbing and matching of similarity of lip movements and the mind to make a somewhat comical attempt at testimony by the CEO of FB social media.