Deep Nostalgia, MyHeritage, & D-ID: The New Talks Of The Town6 min read
I am a sucker for memes. Big time!
Half of the day I scroll pages on social media just to find them.
Recently, when FC Barcelona made a comeback against Sevilla in the 2nd leg of the Copa Del Rey semi-final, I saw numerous memes.
And the most common one (the one that caught my attention) was an animated face of Sevilla defender Jules Koundé.
This time, more than the meme, I was curious about how this animation was created. I thought I could give it a try as well.
After some research (a very little research, to be honest.), I found out that it was created using Deep Nostalgia.
(Deep Nostalgia was already a buzzword. Maybe I was living under the rocks, that’s why I missed out on this).
So, I thought of writing something on this terrific technology.
What Is Deep Nostalgia?
A generic definition or an explanation won’t be fun when we are talking about technologies that involve artificial intelligence (AI).
On top of that, it’s not an independent tech; it’s the brainchild of a company called D-ID.
D-DI (which is also called de-identification) is a Tel Aviv-based AI company.
What do they do? They are the experts in deep learning, computer vision, and image processing and offers a pretty straight-forward service — protect identities from face recognition.
Simply put, D-ID employs deep-learning methodologies to remove sensitive biometric data from images to prevent them from being cross-referenced.
According to the company, “D-ID protected photos looks no different to the human eye. Yet, they cannot be decrypted or reverse-engineered.”
What Is MyHeritage?
Now, let’s talk about MyHeritage.
(Yes! I am coming to Deep Nostalgia. But just hold on for a bit as MyHeritage is also something interesting.)
MyHeritage is the global discovery platform by D-ID. The basic idea behind this platform is to determine your unique ethnic background and match you with newfound relatives. Its DNA test reveals the percentage of DNA you share with the matches and tell you how closely related you are.
Today, over 60 million users from around the world use this platform. You can imagine how much of an impact this has made over the years.
Now, let’s come to the main topic — Deep Nostalgia
In RootsTech Connect 2021, D-ID unveiled an amazing feature that lets you animate pictures uploaded via the MyHeritage Apple or Google app or their website.
“Over 10 million faces have been animated so far. And a few thousand photos are being animated every minute,” according to a report.
And they are right!!! Twitterati is going all out about this new tech.
But while it’s a complete joy ride for you, seeing your grandmother and grandfather blinking, smiling and shaking their heads, have you ever wondered how this tech works?
Let me explain that to you.
How Deep Nostalgia Works?
Deep Nostalgia is a masterpiece of several deep learning algorithms.
According to a source, it uses several VFX drivers of MyHeritage and each driver has got a fixed sequence of movements and gestures.
Now, these drivers guide the animation in an image and produce realistic video footage that mimics human facial movements.
However, there’s a limitation to Deep Nostalgia i.e. it can only animate one face. Meaning, if there are multiple faces in an image, the drivers can only be applied to one.
Should We Consider It A Threat?
Well, to be honest, I feel Deep Nostalgia is like a double-edged sword.
Why am I saying this? Because while many people are amused by this nifty piece of technology, many are a little freaked out.
And I feel the worry is legit!
Invented in 2014 by Ian Goodfellow, a PhD student who now works at Apple, Deepfakes have produced some of the mind-bogglingly convincing videos.
Here’s a video by Bloomberg Quicktake that shows how fake videos and audio keep getting better. And it’s potential is increasing for harm if put in the wrong hands.
Further, just like Deep Nostalgia, Deepfakes also use deep learning technology that applies neural net simulation to data sets.
And with time, it has witnessed massive advances with generative adversarial networks (GANS) playing a crucial role in synthesizing the engine to make better forgeries.
This tech has LITERALLY shown that its capabilities are beyond what we can imagine. And if threat actors master the art of abusing it, they create realistic doctored videos for negative purposes.
When it comes to Deep Nostalgia, the company has stated that it is intended for nostalgic use and doesn’t even include speech in order to prevent abuse.
It is almost certain that advancements are on the way for this technology. And what if tomorrow someone merges deepfake and deep nostalgia? The results can be controversial.
We are not new to forgery media. We all have seen how misinformation and fake media have skyrocketed the damage in society. People have suffered physical, mental and psychological harm in the past.
Now with more and more advanced technologies invading, we have to be more conscious.
Even the Tesla boss has warned the world about the dangers of AI.
“I am really quite close, I am very close, to the cutting edge in AI and it scares the hell out of me. It is capable of more than almost anyone knows and the rate of improvement is exponential.”Elon Musk
It is the need of the hour to decide whether we should put in every ounce in creating digital superintelligence. And even if we decide to do so, we have to be VERY careful.