What is a Deepfake?
Deepfake is a video or image manipulated with artificial intelligence to trick you into believing something that isn’t real. It is most commonly used as a meme, but there are bad actors who use it to make misinformation go viral.
Some examples of the use of deepfakes are to make people who don’t exist and show real people doing or saying things they didn’t really do. Deepfakes can be used to create highly deceptive content, which is why they can be dangerous.
Here are the top 10 deepfake apps you can try for fun and understand the technology
The acceleration of digital transformation and technology adoption have benefited many industries. It has given rise to many innovative technologies and deepfakes are one of them. We all saw how Barack Obama called Donald Trump a ‘complete dipshit’. This is an example of deepfake videos. Deepfake technology uses AI, Deep Learning, and a Generative Adversarial Network or GAN to build videos or images that seem real but are actually fake. Here are the top 10 deepfake apps and websites to experiment with for fun and to further understand the technology.
It is an AI-powered app that allows users to swap faces in videos and GIFs. Reface was formerly known as Doublicat, which had gone viral soon after its launch. With Reface, you can swap faces with celebrities, memes, and create funny videos. The app intelligently uses face embeddings to perform the swaps. The technology is called Reface AI and relies on a Generative Adversarial Network.
|High ratings on Apple and Android app stores||Miss out on key features with free version|
|Easy to use||Lots of ads|
The latest addition is a new feature by Reface that enables users to upload their own content other than selfies. The new feature is called Swap Animation and it lets users add content other than selfies like photos of any humanoid entity, animate it, and do face swap.
My Heritage is a genealogy website that has an app with a deepfake feature. The startup uses a technology called Deep Nostalgia, which lets the users animate old photos. MyHeritage nostalgia feature took the internet by storm and social media was flooded with different experimental photos. This deepfake technology animates the photos uploaded by making the eyes, face, and mouth displaying slight movements.
Zao, a Chinese deepfake technology app, rose to popularity and went viral in the country. Zao’s deep fake technology allows users to swap their faces onto movie characters, it lets the users upload any piece of video and in minutes you get a deepfake generated. The app is only released in China and it efficiently creates amazingly real-looking videos in just minutes. The app enables users to choose from a wide library of videos and images. Zao’s algorithm is mostly trained on Chinese faces and hence, might look a bit unnatural on others.
This editing application recently went viral due to its unique features that enable users to apply aging effects. Social media was flooded with people trying different filters from FaceApp in recent times. This is a free app, and this makes it even more viral among the audience. FaceApp leverages artificial intelligence, advanced machine learning, deep learning technology, along with an image recognition system.
|Many photo-editing features are available||Limited features with free version|
|Easy to use||Lots of ads|
5. Deepfakes Web
It is an online deepfake software that works in the cloud. Deepfakes Web allows the users to create deepfake videos on the web and unlike the other apps, it takes almost 5 hours to curate a deep fake video. It learns and trains from the videos and images uploaded, using its deepfake AI-based algorithm and deep learning technology. This platform is a good choice if you want to know the technology behind deepfakes better and understand the nuances of computer vision. It allows the users to reuse the trained models so that they can further improve on the video and create deepfakes without using a trained model. The platform is priced at USD3 per hour and promises complete privacy by not sharing the data with a 3rd party.
6. Deep Art Effects
As the name suggests, it is not a deepfake video app, but DeepArt creates deepfake images by turning them into artistic. The app uses a Neural Style Transfer algorithm and AI to convert the uploaded photos into famous fine arts paintings, and recreate artistic images. DeepArt is a free app and has more than 50 art styles and filters. The app offers standard, HD, and Ultra HD features, in which the latter two are priced versions. The app allows its users to download and share the images created.
Wombo is an AI-powered lip-sync app, wherein users can transform any face into a singing face. There is a list of songs to choose from and users can select one and make the chosen character in an image to sing it. The app creates singing videos that have a Photoshop quality to them and hence, it seems animated and not realistic. Wombo uses AI technology to enable the deepfake scenario.
8. DeepFace Lab – Best Deepfake Software in General
It is a windows program that lets users create deepfake videos. Rather than taking deepfake technology as a fun element, this software program allows its users to learn and understand the technology better. It uses deep learning, machine learning, and human image synthesis. Primarily built for researchers in the field of deep learning and computer vision, DeepFace Lab is not a user-friendly platform. The user needs to learn the documentation and also needs a powerful PC with a high-end GPU to use the program.
9. Face Swap Live
Face Swap Live is a mobile application that lets users swap faces with another person in real-time. The app also allows its users to create videos and apply different filters to them and directly share them on social media. Unlike most of the other deepfake apps, Face Swap Live does not use static images and instead enables to perform of live face swaps with the phone camera. Face Swap Live is not a fully deepfake app, but if you are looking to use deepfakes for fun, this should be the right one. The app effectively uses computer vision and machine learning.
AvengeThem is a website that lets users select a GIF and swap their images onto the faces of the characters from the Avengers movie series. Although it is not a completely deepfake website as it uses a 3D model to replace the faces and animate them. The website has about 18 GIFs available and it does not take more than 30 seconds to create this effect, which does not look very realistic.
Are There Any Benefits of Deepfakes?
There are a lot of applications for DeepFake technology, and it can really have some hugely positive effects. For example, it could be used in films where the actors couldn’t be there for any legitimate reason.
Deepfakes are so persuasive that they show characters at a young age or replace those who have passed away. They haven’t proven to replace CGI in the film industry just yet, but it’s still too early to tell.
The fashion industry could also be a potential customer of this technology and it is looking for ways to fulfill its clients. Deepfakes would allow for customers to see what a particular piece of clothing will look like on them before committing to the purchase.
Is deepfake AI?
Yes, deepfake apps and websites use AI, ML, and machine vision to create deepfakes.
What Risks Do Deepfake Apps & Websites Pose?
Deepfakes have positive uses, but they are often used for bad purposes and manipulation. In the film industry, they can help to create better content while in the fashion industry they can provide a level of authenticity to the clothes being sold. The problem is that deepfakes are often used for nefarious purposes, such as disinformation attacks and fake celebrity videos.
Deepfakes can be used in social engineering scams and financial fraud, as well. In 2019, a voice deepfake was used to commit CEO fraud after stealing $243,000 from an unnamed UK company.
Deepfakes could lead to serious consequences for society. They might make cybersecurity measures pointless, undermine political stability and affect the finances of corporations or individuals.
Service from MyHeritage uses deep learning technique to automatically animate faces
An AI-powered service called Deep Nostalgia that animates still photos has become the main character on Twitter this fine Sunday, as people try to create the creepiest fake “video” possible, apparently.
The Deep Nostalgia service, offered by online genealogy company MyHeritage, uses AI licensed from D-ID to create the effect that a still photo is moving. It’s kinda like the iOS Live Photos feature, which adds a few seconds of video to help smartphone photographers find the best shot.
But Deep Nostalgia can take photos from any camera and bring them to “life.” The program uses pre-recorded driver videos of facial movements and applies the one that works best for the still photo in question. Its intended purpose is to allow you to upload photos of deceased loved ones and see them in “action,” which seems like a lovely idea.
Like most “deepfakes” like “deepnude” – the name for the popular use of this technology to map one person’s face on to footage of another – the service is exceptionally good at smoothly animating features and expressions. But it can also struggle to generate data to fill in the “gaps” in what it can see from the source photos, causing a sense of the uncanny.
Users have to sign up for a free account on MyHeritage and then upload a photo. From there the process is automated; the site enhances the image before animating it and creating a gif. The site’s FAQ says it does not provide the photos to any third parties, and on its main page a message reads “photos uploaded without completing signup are automatically deleted to protect your privacy.”
Naturally, the program has become something of a meme-generator on Twitter, with users trying to push the AI to its limit. An archaeologist used photos of ancient statues, and yes they included some with the blank eyes. Sorry in advance for the nightmare fuel (but hiiii there Alexander the Great):
“Some people love the Deep Nostalgia feature and consider it magical, while others find it creepy and dislike it,” MyHeritage says about its technology. “Indeed, the results can be controversial and it’s hard to stay indifferent to this technology. This feature is intended for nostalgic use, that is, to bring beloved ancestors back to life. Our driver videos don’t include speech in order to prevent abuse of this, such as the creation of ‘deep fake’ videos of living people.”
Not every video created with the service is elegantly animated, or even good enough to be unsettling, of course. An animated version of the infamous bust of Ronaldo, for instance, is exactly as distressing as the static version:
And while the automatically produced videos of Deep Nostalgia are not likely to fool anyone into thinking they are real footage, more careful application of the same technology can be very hard to distinguish from reality.
Deep Nostalgia can only handle single headshots and can only animate faces, so you’re not going to be able to reanimate mummies to make it look like they’re walking (hey I wondered, OK?). You can upload five photos for free to the MyHeritage website for Deep Nostalgia treatment, after that you have to register for a paid account.
Last month, a new TikTok account named deeptomcruise racked up millions of views with a series of videos that are, it claims, deepfake versions of the actor talking to camera. The Cruise fakes are so accurate that many programmes designed to recognise manipulated media are unable to spot them.
- How does the Deep Nostalgia software illustrate the growing problem of deep fakes?
- What are the biggest trends in deep fakes at the moment?
- How can users protect themselves against being fooled by deep fakes?
In 2019, the Ukrainian IT-company Neocortext (current RefaceAI) released the Doublicat mobile app (now Reface), with which the user can replace the face on the gif with his own. Six months later, the application was already changing faces to video, and by August the number of its installations exceeded 20 million.
Initial Release Date: Dec 23, 2019
Content Rating:Rated 12+
According to the analytics service App Annie as of August 15, Reface is among the ten most popular apps on iOS in 15 countries, and on Android – in 19.
How the user interacts with the service?
To insert their face into a GIF or video, the user takes a photo in the application and selects a template, for example, a fragment from a movie. After that, the algorithm changes its appearance in a few seconds. The result can be downloaded immediately or shared on social networks.
How does the technology work and where is it used?
Usually, when creating a deepfake, it takes a lot of time to train a neural network. At the same time, a separate network must be trained for each person.
However, RefaceAI has created a universal Artificial neural network to replace any human face, thanks to which a deepfake is obtained in seconds. The developers have trained the network on millions of images from open libraries (the company does not disclose the name), so it can change faces in both photos and videos.
Having received the user’s photo, the network “translates” it into face embeddings – an anonymized set of numbers. According to it, the machine determines the facial features and transfers them to the template.
Deepfake turns out to be more realistic thanks to machine learning, including a GAN-type neural network – its peculiarity is that it includes two networks that train each other. In the case of Reface, for example, they “adjust” the color of the user’s face to the lighting of the original video or picture.
Startup Success Story
RefaceAI, the company behind Reface, was founded in 2011 by:
- Roman Mogilny – CEO.
- Oles Petriv – Technical Director.
- Yaroslav Boyko – Chief Operating Officer.
Before the face-swapping app, entrepreneurs had been involved in various projects for seven years: developing websites, collaborating with post-production studios for Hollywood films, where machine learning technologies were needed. For example, they converted a video from 2D to 3D format.
In 2018, the company came up with the idea to create an app that would replace faces in photos. At that time, RefaceAI employed six people.
RefaceAI has raised 1 round. This was a Pre-Seed round raised on Dec 5, 2019. Adventures Lab has invested in the startup between $300,000 and $500,000.crunchbase.com & ain.ua
In March 2019, Elon Musk posted photos on Twitter with his face instead of Dwayne Johnson’s. The image featured the Reflect watermark. Due to this publication, application traffic has grown tenfold, entrepreneurs noticed.
By September, the co-founders realized that simply changing faces in photos was not enough. At that time, product manager Ivan Altsibeev joined the team. He will suggest switching to gifs. The idea turned into a Doublicat app. It was presented at Product Hunt in January 2020.
Six months later, the company added face-to-video to the app and renamed Doublicat to Reface. With the new feature, the service has grown in popularity, with Britney Spears, Snoop Dogg and other celebrities sharing their videos.
Reface currently has 20 million installs and continues to grow. How quickly, the company does not specify. Her spokesperson added that 65% of users share content created in the app.
The basic version of Reface is free. The company receives income from advertising and paid subscriptions, where you can turn off the watermark: 199 rubles per week, 299 rubles per month and 1990 rubles per year. The company does not disclose the total revenue from the service.
To replace faces in photos, developers use images with open licenses, and for gifs they partner with sites like Tenor.
In the case of the video, the company adheres to the advice of lawyers:
- Content falls under US copyright fair use and therefore does not require licensing.
- Limits the length of the videos, their quality and the rest of the content.
If the copyright holder wants to exclude their materials from the application, Reface App will remove them.
In an interview Mogilny, Petriv and Boyko explained that the popularity of such applications is usually short-lived, so they use mechanics to retain users.
According to entrepreneurs, Reface will move forward not only the appearance of new content, but also its localization – so that the user can insert his face into a clip with a popular star in the country.
Since 2018, RefaceAI has grown to 40 employees. She is currently conducting closed beta testing of the Reface Studio web platform. With its help, creators of entertainment content will be able to insert faces into any video. In the future, the company plans to replace bodies as well.
As conceived by the founders, the new service will work in the b2b segment as well: it will be useful for creative agencies, filmmakers and computer game developers.
One of the problems that Reface Studio can face is using the service to create fake news and replace the faces of famous people. To prevent possible harm to the public, developers will apply two approaches:
- You cannot use the service anonymously.
- The video created in the “studio” will have an invisible mark that the project was created using Reface Studio.
Top In-App Purchases from AppStore
- Weekly $2.99
- Annual Plus $27.99
- Monthly $4.99
- Annual $27.99
Original post https://vc.ru/ml/149769-dostupnyy-dipfeyk-chto-interesnogo-v-servise-dlya-zameny-lic-reface-iz-ukrainy-kotoryy-vzletel-v-reytingah-prilozheniy