Luxand Blog
In our tech-driven world, deepfakes have emerged as both a fascinating innovation and a source of concern, since it is a technology that seamlessly blurs the lines between the real and the artificial. It crafts incredibly realistic images and videos, allowing them to imitate anyone with almost uncanny precision. While deepfakes open up new avenues for creativity and innovation, they also bring a raft of ethical and security issues. As these digital replicas become more common, it's crucial for us to learn how to spot them and understand their implications. Doing so will help us navigate the digital landscape more safely and responsibly.
Mechanics of Deepfakes
Let’s start by figuring out what deepfakes actually are. To put it simpler, deepfakes are digital images or videos crafted through a process called deep learning, a type of artificial intelligence that learns in a way similar to humans. Consider how a baby experiments with different items to figure out what can be eaten; likewise, deep learning algorithms analyze numerous real images from the web to learn how to generate new, realistic-looking images. Just as the baby improves in identifying food by sight, the algorithm becomes skilled at creating convincing fake visuals over time.
Diving deeper into the technicalities of it, deep learning, a specialized branch of machine learning, uses neural networks abundant with "hidden layers" to simulate human cognitive processes. These hidden layers conduct intricate mathematical changes to transform input into output data, enabling the creation of lifelike fake images. Producing deepfakes involves two main algorithms: one that creates imitation images and another that distinguishes those from genuine images. They challenge each other by continually enhancing their skills through repeated interactions. This back-and-forth process sharpens the generative model's ability to make highly realistic fakes, often making them indistinguishable from real images and videos to the human eye.
Creating a convincing deepfake requires a combination of two main components: generators and discriminators. Generators start the process, crafting fresh images from the input data, while discriminators assess these outputs for authenticity. During their dynamic interplay, known as Generative Adversarial Networks (GANs), both components refine their processes to produce an end result that even trained eyes might find hard to distinguish from reality.
A robust collection of video, audio, and photographic data is crucial for producing persuasive deepfakes. Bigger data reservoirs yield more lifelike imitations. Although complex software and powerful computational resources are still key ingredients in the mix, the march of technology is rapidly lowering the barriers for crafting intricate deepfakes.
Who’s Behind Deepfakes?
In the realm of film and entertainment, industry professionals often use deepfake technology to turn back the clock on actors, giving them youthful appearances, or to create voiceovers from those who have passed away. Projects have even brought inspiring historical figures to life, creating a bridge between past and present for learners worldwide.
Beyond the film set, app developers have jumped on the deepfake bandwagon, offering face-swapping functionalities for users eager to see themselves starring in favorite movie scenes. Although these applications don't match Hollywood's sophistication, the effects can still be mind-boggling.
However, there's a darker side. Some use this technology for nefarious ends, such as crafting misleading political content or creating exploitative materials for blackmail and fraud. The potential for manipulating public opinion or privacy results in serious ethical and legal concerns.
Are Deepfakes Legal?
While no overarching federal law governs deepfakes in the United States, several states have taken steps to address their proliferation. For example, the creation and distribution of deepfakes linked to electoral influence or nonconsensual pornography are outlawed in certain states. Others are contemplating similar legislation as the technology continues to evolve.
Spotting Deepfakes: What to Look For
Detecting a deepfake can be tricky but not impossible. Visual indicators include:
- unnatural eye and facial movements,
- mismatched body postures,
- awkwardly animated features,
- inconsistent color tones,
- disparities between the video and audio,
- absence of natural blinking patterns,
- zooming in makes the video more uncanny.
Listening closely can also reveal inconsistencies in audio; mismatched lip-syncing or robotic tones are red flags. Digital anomalies like weird lighting or misplaced shadows further betray these forgery attempts.
Deepfakes might fail to capture subtle traits, habitual gestures, or genuine emotional responses, resulting in an odd portrayal. Inconsistencies in the backdrop, unexpected interactions with the environment, or storyline deviations may suggest manipulation.
How to Protect Yourself?
Technological tools and strategies are emerging as allies in this fight. From examining digital signatures to employing blockchain for verifying authenticity, the hunt for deepfakes is becoming more sophisticated. As tech and media industries amplify efforts to curb malicious usages, users' vigilance remains indispensable.
But what can you actually do to protect yourself?
Forewarned is forearmed. Being aware of what deepfakes are and how they function is a key defense strategy. By gaining insight into the technology and recognizing its potential for misuse, you can evaluate digital content with a more discerning eye. Check the source. Always check the origin of information. Seek confirmation from credible and trustworthy sources before believing any media content to be authentic. Deepfake detection tools. Use available deepfake detection applications. Many companies and platforms offer tools or plugins specifically crafted to spot manipulated content. Report when suspicious. If you come across media that seems to be a deepfake, report it to the platform that is hosting it. User reports are essential in helping these platforms identify and address misleading media.
Deepfakes sit at the intersection of innovation and manipulation, bringing both exciting possibilities and challenges. How we choose to use them will influence their future role. Tools like Luxand FaceSDK play a key part in this, offering effective anti-spoofing and live detection features. By using such technology, we can explore the potential of deepfakes while keeping misuse in check and ensuring digital safety.
We are excited to announce the release of FaceSDK version 8.2, which introduces the iBeta Certified Liveness Add-on, enhances stability of samples, and provides updated documentation.
Here’s what's new in FaceSDK 8.2:
- iBeta Certified Liveness Add-on for Android, iOS, Windows/x64, Linux/x64
- Stability improvements for React Native, Flutter, and .NET wrappers and samples
- New documentation for the Python wrapper
The iBeta Certified Liveness Add-on is a powerful, single-image, passive liveness solution that has achieved iBeta ISO 30107-3 PAD compliance. What's great is that it uses the same selfie taken for facial recognition to easily and accurately detect fraud, without needing any extra steps or user involvement.
Download the new version of Luxand FaceSDK 8.2
Luxand's FaceSDK has demonstrated impressive performance in the recent evaluations by the National Institute of Standards and Technology (NIST) during the Face Recognition Vendor Test (FRVT). We achieved notable results, ranking within the top 21.8% for the Visa-Border Combination assessment. This particular test involves comparing visa photos with images captured at border crossings, which are often taken in less-than-ideal conditions.
Why It Matters
The Visa-Border Combination test is crucial as it replicates a common real-world situation where high-quality visa images must be matched with lower-quality photos taken at border crossings. This scenario is further complicated by significant head turns, with yaw angles of 45 degrees or more. Luxand FaceSDK’s strong accuracy in these challenging conditions underscores its reliability and robustness for practical use.
Grasping NIST’s Evaluations
For those familiar with facial recognition technology, NIST's FRVT is viewed as the gold standard for evaluating facial recognition systems. These comprehensive tests include various scenarios such as the 1:1 matching evaluation, which involves comparing a test image to a reference image to determine recognition accuracy.
Key performance metrics include:
- False Match Rate (FMR): This measures the frequency with which different individuals are incorrectly matched. Luxand FaceSDK's low FMR highlights its exceptional accuracy.
- False Non-Match Rate (FNMR): This indicates how often the system fails to match two images of the same individual. Luxand's low FNMR demonstrates its consistent ability to accurately recognize individuals.
We are excited to announce the launch of the iBeta certified Liveness add-on for FaceSDK, enhancing user authentication and security. Available for Android and Windows/x64 platforms, this addition marks a significant advancement in biometric verification technology, and we are planning to add support for more platforms soon.
iBeta certified Liveness add-on for FaceSDK passed Level 1 iBeta Presentation Attack Detection (PAD) conformance testing with a perfect PAD score. iBeta Quality Assurance conducted Presentation Attack Detection (PAD) testing in accordance with ISO/IEC 30107-3. iBeta is accredited by NIST/NVLAP (NVLAP Lab Code: 200962) to test and provide results to this PAD standard.
During the evaluation, iBeta researchers conducted about a thousand presentation attacks using subjects’ authentic biometric samples to create artifacts. None of these presentation attacks were successful, resulting in an Attack Presentation Classification Error Rate (APCER) of 0%.
As a uniquely accredited third-party biometrics testing lab, iBeta Quality Assurance ensures that products meet the highest standards of functionality and security. The iBeta certification guarantees the most precise liveness and anti-spoofing checks for images and videos.
FaceSDK 8.1: More Platforms, Enhanced Samples, Improved Face Detection and a 50% OFF deal
09/28/2023
We have just released FaceSDK version 8.1, which now supports additional platforms, introduces more samples, and enhances compatibility.
The new FaceSDK version 8.1 now includes:
- A WebAssembly build and sample
- A Flutter wrapper with samples for iOS and Android
- A React Native wrapper with samples for iOS and Android
- .NET Standard 2.0 wrapper
We have also made improvements to the iOS samples for Swift, Android samples for Android Studio and enhanced IDE compatibility for the Python wrapper.
In addition to the expanded platform support and additional samples, we have made improvements in face detection technology.
Download the new version of Luxand FaceSDK 8.1
In celebration of this release, enjoy a 50% OFF deal. This special offer lasts until October 31st. Don't miss out on getting FaceSDK 8.1 at half the price before the planned price increase in November 2023!
We're thrilled to announce that we'll be participating in the annual Seamless Conference in Dubai, one of the largest gatherings in the tech industry happening on May 23-24.
Seamless is the meeting place for pioneers and the savvy, for those who are eager to push boundaries in security, biometrics, fintech, ecommerce, retail and much more. With 500+ companies across different sectors, the two-day event promises an enriching experience of knowledge sharing and networking.
We are excited to welcome you to our booth K94 in the Identity section where we'll be exhibiting our FaceSDK solution. This is a great opportunity to see how face recognition works, and learn more about such features as liveness detection, thermal face detection and more.
In a digital world, nothing beats a good old face-to-face conversation. It’s a chance for us to learn more about you, your needs, and how we can serve you better. It's also an opportunity for you to get to know us, the people behind FaceSDK, and our commitment to your success.
Your presence will make the event even more meaningful to us. We can't wait to meet you in person in the vibrant city of Dubai.
Please stop by, say hello, and let's make the most of this incredible event together!
We have just released FaceSDK version 8.0.1 that adds the support of arm64-based processors for the Linux platform, thus making it easier to use the latest OS on such devices as Raspberry Pi 3b+ or Raspberry Pi 4.
Also we have added more Python samples for Linux, and improved the Python wrapper for all environments by adding inline documentation for the functions.
Download the new version of Luxand's FaceSDK v 8.0.1
We are excited to announce the official release of the new FaceSDK 8.0 version! The new version now offers both passive and active liveness detection, which will improve the security and anti-fraud measures in facial recognition based solutions.
The passive liveness check, which has been widely requested by our users, is the most sophisticated anti-spoofing technology. It does not require any special hardware, nor does it ask users to perform any actions to prove that they are alive - it just works with standard images!
Download the new version of Luxand's FaceSDK v 8.0
Effective immediately we’re offering a Buy One Get One Free deal until April 30. Buy one FaceSDK license and get an extra license of equal or lesser value free of charge!
We have just released FaceSDK version 7.2.1 that adds the support of Apple Silicon processors (arm64) in MacOS, to follow the latest Apple's announcement of Mac computer transition to new arm64-based processors.
We have also added a Mask-on Face Detection model - it's now possible to detect faces covered by masks (medicine masks, scarfs, bandanas, etc), which significantly improves face detection quality during pandemics.
In addition to a few major updates above we have also fixed problems with merging Tracker databases in trackerMemoryTool.py.
Download new version of Luxand FaceSDK 7.2.1
Luxand FaceSDK 7.2: Thermal Face Detection, Major Improvements and 2-for-1 Deal
09/09/2020
We are excited to announce the official release of the new FaceSDK 7.2 version! While FaceSDK 7.1 was all about speed and ease of use improvements, version 7.2 takes a major step forward with new features.
Thermal face detection is a major feature that makes it possible to detect faces on streams and images from thermal cameras for anti-spoofing or fever screening needs. Also, you can load face detection models from files now.
We keep improving FaceSDK, so we have increased face enrollment speed 2 times on Linux, added auto rotation of loaded JPEG photos using EXIF, as well as added support for higher InternalResizeWidth values - it's easier now to detect even smaller faces on larger images and high-resolution video streams!
In addition, FaceSDK now fully supports Python and features new samples to help you get started!
Effective immediately and until September 30, we’re offering a Buy One Get One Free deal. Buy one FaceSDK license and receive an extra license free of charge!
Luxand FaceSDK 7.1: Major Speed Improvements and Halloween Two-for-One Deal
10/15/2019
We are excited to announce that we’ve successfully completed a major round of optimizations. While FaceSDK 7.0 was all about new features and new algorithms to cope with imperfect lighting conditions, the dot-one release polishes off the rough edges. The new code takes one-sixth of the time to detect faces on most Android devices (with at least 15% performance gain on the remaining few), and brings a 30% performance boost to desktop platforms. Facial feature detection received a major 100% performance boost on iOS and a 15% speedup on other platforms. Enrolling new faces now takes up to three times less time. As usual, we squashed some bugs and added new samples to help you get started.
Start Luxand FaceSDK Free Trial
If you already own a FaceSDK license, drop us an email to check if you’re eligible for a free update. If you still don’t, today is the best time to buy. Effective immediately and until October 31, we’re offering a Halloween Buy One Get One Free deal. Buy one FaceSDK license and receive an extra license free of charge!
Luxand BabyMaker for iPhone/iPad
09/01/2019
Do you know MakeMeBabies.com? That online entertainment portal can render a picture of a baby based on two parent’s faces. We’ve brought this technology to Apple iOS, releasing an app for iPhone and iPad. Luxand BabyMaker for iPhone/iPad will analyze the two faces, create a mathematical model of their future baby, render that baby’s face, and make a photo collage. Used to create some 30 million faces, the technology was featured in Graham Norton Show by Jennifer Lopez, and showcased by Globo TV in Brazil. BabyMaker’s technology is super popular, smart, funny and family-friendly.
Download BabyMaker on the AppStore!
FaceSDK 7.0 with Revamped Detection Engine, End of Year Deal
12/18/2018
The year comes to an end, and we are releasing our last update in 2018. Meet FaceSDK 7.0 with revamped face detection engine and accelerated Face Matching!
Start Luxand FaceSDK Free Trial
The revamped face detection engine helps detect human faces in the most complex conditions. Rotated faces, blurred footage, low light and excessive noise are much less of an issue thanks to the new algorithms.
If you are using Face Matching to compare faces in your Android apps, your app will work up to twice as fast. And by the way, you can now target ARM64 platform thanks to native libs recompiled with 64-bit support.
See our Release Notes for information on what was improved.
Last but not least, you can save the cost of a second license thanks to our End of Year Buy One Get One Free deal. Just place your order before 31th of December, and your second license will be on us!
Mirror Reality… Light!
06/28/2018
Did you ever want live, animated masks for your AR app? Have you ever considered Mirror Reality SDK, but were put away by the price tag? If you nodded to at least one of these questions, then Mirror Reality SDK Light might be just the right tool for you!
What’s so exciting about the Light version of Mirror Reality SDK? It’s the sticker price. We slashed the price of Mirror Reality SDK Light so sharp it can cut. Before July 6, you can order Mirror Reality SDK Light at a shocking price of $199. If you’re late to the party, you’ll pay the sticker price of $499, which is still a huge bargain compared to everything else on the market.
Buy Mirror Reality SDK Light for $199
What’s the catch? The Light version of the MR SDK will include exactly the same number of pre-defined masks, no difference there. However, instead of letting you go wild with your own creations, the Light edition will limit you to five custom masks instead of unlimited in the regular edition.
What if you buy the Light version now and your requirements grow afterwards? We’ll be shamelessly happy to upsell! At any time, you can update your Light version license to the full, regular edition by simply paying the difference between the price you paid and the actual cost of the regular edition at the time of upgrade.
FaceSDK 6.5.1 Released
06/06/2018
We’ve just released FaceSDK 6.5.1. The update increases the speed of enrollment by 20% on Windows, macOS and Android, and by 5% on iOS and Linux. It also improves the accuracy of face recognition under very low lighting conditions. It is recommended that you update:
See our Release Notes for information on what was improved.
FaceSDK 6.5 Improves Face Recognition to 99.85%, Adds Embedded Linux Support
05/15/2018
We are excited to announce an important FaceSDK update followed by the buy-one-get-one-free deal! Version 6.5 significantly improves face recognition, bringing successful recognition rates to 99.85% according to NIST FRGC testing – up from 94% in the previous build. With less than 0.15% of false rejections, you can build highly robust video surveillance and biometric identification systems – or instantly improve existing ones by simply rebuilding with the new SDK!
In addition, you are no longer limited to PC or mobile platforms. FaceSDK 6.5 can now be used to power ARM-based Linux systems, allowing you building face recognition and identification technologies into embedded Linux platforms such as the Raspberry PI. To view the complete changelog, please visit our Release Notes.
To celebrate the new release, we are rolling out a Buy-One-Get-One-Free special offer. The deal is simple: just purchase one license for FaceSDK, and receive the second license at no extra charge.
This promotion will last until the end of May. Don’t miss your chance and place your order before May 31!
Spring Sale: Get Mirror Reality SDK at 1/2 the Price before May 1!
04/18/2018
Ready for some spring shopping? Our April sale is now live! Order Mirror Reality SDK before May 1, 2018, and save 50% of the purchase price! That’s right: this is your chance to get Mirror Reality SDK at 1/2 of the price. If you purchase both iOS and Android licenses before the deadline, you will receive them at exactly one-half the usual price.
Mirror Reality SDK allows creating Android and iOS apps with live animated masks augmented to human faces in real time. The SDK comes with lots of preinstalled masks, but there’s nothing to stop you from making your own – or ordering one from us.
FaceSDK 6.4 Adds Blazing Fast Face Detection
01/31/2018
FaceSDK is back again with a major performance update. In this update, we removed the last performance bottleneck, making a leap towards real-time recognition of even the smallest faces appearing in video streams.
How did we do that? In a word, FaceSDK 6.4 greatly improves the speed of detecting faces. The last release, and the one before it, were concentrated on optimizing the facial features detection and recognition stages. However, before you recognize a face or identify its facial features, you must find it on the picture or in the video stream. No longer does this remain a bottleneck! In this release, we removed the last performance constraint, making face detection up to 5 times faster on Windows and Mac computers and up to 2.5 times on Linux, iOS and Android.
And it’s not the end. To celebrate St. Valentine’s day later this year, we are rolling out one of our signature Buy-One-Get-One-Free deals. Effective immediately, you will get a second license free when you order FaceSDK. This is your chance to receive an extra license if you buy just one.
This promotion will last until February 14, 2018. Don’t miss your chance and place your order before Valentine’s!
age-o-meter: How Old Do I Look?
12/08/2017
How old are you, really? Or should we ask, how young? We have an app for that! Luxand’s facial recognition engine now empowers a brand new app we called "Age-o-meter".
Age-o-meter can tell the age of a person by processing a video stream captured by the phone’s camera. Sounds fun? We built a highly sophisticated technology under the hood to make the complex process of recognizing the person’s age look easy and work in a snap.
The app and its underlying technology are both powered by Luxand FaceSDK, a comprehensive developers’ library for building apps with face recognition and facial feature detection in still images and video streams. FaceSDK helps create apps that can detect a person’s gender and age, recognize the moment they smile and tell if their eyes are open. All that opens some amazing possibilities for app developers, allowing to quickly build apps powered by market's most sophisticated biometric technology.
Download age-o-meter on the AppStore!
FaceSDK 6.3 Doubles Performance, Adds Age Recognition
09/19/2017
This summer, FaceSDK turned 10. With ten years on the market, we know exactly which parts of our product see the most use. With that in mind, we spent our time to thoroughly rework, improve and optimize most frequently used functions – like face recognition or detection of facial features. Thanks to low-level optimizations, we managed to squeeze every bit of performance out of existing hardware. The result? The most commonly called functions now work with double or triple the speed.
This season age is fashion. We added age recognition to the already impressive list of FaceSDK functions. FaceSDK 6.3 can now recognize how old a person is with a single photo of a few frames of video footage. To view the complete changelog, please visit our Release Notes.
To celebrate our first anniversary, we extended our Buy-One-Get-One-Free deal. This is your chance to receive an extra license if you buy just one.
This promotion will last until October 7. Don’t miss your chance and place your order before October 7!
FaceSDK to Double Performance at Half the Price
08/30/2017
We are constantly working on improving and optimizing FaceSDK. This September, we will roll out the result of more than a year of careful tuning and low-level optimizations. Thanks to these optimizations, the new release will greatly improve the performance of the most frequently invoked functions. The update will double or triple the performance of facial feature detection on most platforms and double the performance of matching faces against a database on most popular desktops and servers.
Today is your chance to receive this brand new release at half the price. Order your FaceSDK license today and receive the second license for FaceSDK free of charge – and both licenses get a free update to the newer and better FaceSDK with double performance once it’s released!
Hurry up! This deal is valid through the end of September.
10 Years of FaceSDK
08/27/2017
Ten years ago, on August 27, 2007 we released Luxand FaceSDK 1.0. It was the first SDK on the market whose Artificial Intelligence allowed to recognize individual facial features.
Today, we are celebrating our 10th anniversary. This is a big day for us, and we wanted to share some of our heritage with you.
From the start, FaceSDK was fully capable of recognizing facial features. It enabled us to accurately match individual persons, to search photos in large face datasets, and to enhance security of access control systems. Pretty soon we added webcam support, and started to recognize faces in real time. Our users could sign in to their apps just by looking into a camera.
With FaceSDK, our AI could:
- Age faces
- Guesstimate the look of a couple’s child
- Enhance faces by removing skin defects and adjusting skin tone
- Add realistic emotions to human faces, building a 3D model and convincingly animating it
- Apply facial filters in live video streams. Back in 2011, augmented reality was hardly heard of, but our Mirror Reality app ran smoothly on desktops.
We always wanted to make your life safer, and we wanted to make our tools just for that. Back then, they split video streams from security cameras frame-by-frame and matched them against a database of pre-registered facial photos. It was inefficient, not very accurate, and you would still need to fill up that database to get started.
So why not make the machine work instead? And there we have it: by 2013, we wrapped our Artificial Intelligence into the Tracker API, outing the first self-learning AI that trains on every face it sees and keeps tracking them since first encounter. No need to enroll anyone, and developers need just a single function call to handle video frames.
Since then, we’ve been busy enhancing and optimizing FaceSDK. The 6.0 in 2015 added smooth real-time processing of human faces on mobile devices. Mirror Reality SDK in 2016 allowed building live facial filters on smartphones and tablets.
Ten years after its initial release, FaceSDK supports all popular platforms and ecosystems, effectively working stand-alone on all types of devices. It featured automatic tracking of 70 facial features, and recognizes gender and facial expressions. FaceSDK is one of the most popular and the most cited developer tool. We have over 200 academic quotations. Our algorithms are protected with patents. FaceSDK is used by customers such as Universal Pictures, Samsung, LG, Philips, Citrix, Diebold, Unilever, P&G, Ford, FIS, Badoo, U.S. Department of Defence Cyber Crime Center and Banco Azteca. It is used in science and entertainment industry, in areas as diverse as makeup and security systems.
You are the reason why FaceSDK exists. You are the leading force behind its features and development. Thanks to you, we know we are doing the right thing. Thank you for being with us.
Last but not least. We are preparing a big anniversary surprise for you. Stay tuned!
FaceSDK Turns 10, Anniversary Discount
08/15/2017
This summer, Luxand celebrates its special anniversary. Our flagship product, Luxand FaceSDK, turns 10 years old on August 27. We pioneered some amazing technologies based on recognizing and interacting with still and live images of human faces.
To celebrate our first anniversary, we prepared some huge discounts. This is your chance to get Mirror Reality SDK at 1/3 of the usual price. Yes, we slash the price of our most popular iOS/Android SDK to share our excitement and show our appreciation of our valued customers being with us all that time. With Mirror Reality SDK, you can build apps making people look younger or older, turning them into walking zombies, dressing up or trimming beards – all you need is some video footage from a camera.
Ready? Action! The sale is good till Aug 27, so hurry up and pocket the huge discount before it’s over! Fill in the quote request form and we will get back to you as soon as possible.
youmask stars: Feel Like a Movie Star!
12/23/2016
Are you fond of movies? Ever wanted to be inside one? We certainly do! That’s why we made youmask stars, an app that makes you look, feel and act like a movie character… on the display of your phone of tablet!
Youmask stars builds on the foundation of Luxand biometric technologies, applying an animated mask to your face that looks real. With youmask stars, you can become someone else. When you smile, your avatar smiles. If you blink, your avatar blinks. If you frown… you guessed it: it frowns, too.
What are you waiting for? The App Store link is right here:
Mirror Reality SDK Now Available for Windows
12/22/2016
Mirror Reality SDK, our cutting-edge toolkit for building augmented reality apps, is now available for the Windows platform. With this new SDK you’ll be able to build desktop apps running on any Windows PC, laptop or kiosk.
Before the year end, take a chance to have a look at the new and update version of Mirror Reality SDK. As a reminder, Mirror Reality SDK enables iOS, Android and Windows developers to create augmented reality apps based on real-time face transforms. This coming year, Mirror Reality SDK becomes more convenient, better documented and much better illustrated.
Just before the holiday season, we added a number of masks to make celebrating Christmas and the New Year lots and lots of fun! You have a full year ahead of you to prepare for the next Christmas, so why not download Mirror Reality SDK now and have a look at the masks?
FaceSDK 6.2 Improves Real-Time Recognition, Better Live Facial Filters on Mobiles
09/15/2016
Luxand Inc. updates FaceSDK, the company’s flagship face recognition and facial feature detection suite. FaceSDK 6.2 features numerous improvements and enhancements, enabling developers to build mobile apps implementing live facial filters.
The improved Facial Feature Detection and Tracking engine is now based on 70-point detection of facial features (up from 66 facial features in the last release), improves tracking of open mouth and significantly improves speed on major mobile platforms. The new engine reaches 30 frames per second on iPhone 6 when detecting facial features on live video streams captured by the phone’s camera in landscape orientation.
“We’re making it better with each iteration”, says Alex Konoplev, Luxand CEO. “The new build packs numerous improvements in different areas allowing developers implement smoother experience and increasingly lifelike animations never imaginable before. The massive performance improvements in the newly announced iPhone 7 are highly welcome, yet we are fully committed to supporting developers targeting other popular iOS devices.”
New Benchmark Performance with Less Jitter
Luxand is constantly working on improving the speed and fluidity of its tools. The updated FaceSDK 6.2 offers a bump in frame rates, which can be used to significantly reduce jitter while detecting and tracking facial features in video streams.
Jitter is a visual artifact that requires additional processing power to reduce. In this release, FaceSDK introduced a significant increase in performance, allowing developers either to achieve a solid 30-FPS rate or on iPhone 6 in landscape orientation, or to reduce the jitter of videos.
Detailed Facial Feature Detection with Backward Compatibility
FaceSDK 6.2 can now detect 70 facial features, up from 66 features in previous versions. The more detailed detection enables lifelike facial transforms and animations that look more natural than ever. In addition, FaceSDK 6.2 improves tracking of open mouth, providing even more lifelike experience. Full backwards compatibility with previous versions of the SDK is maintained.
New Sample Code for Android and iOS
FaceSDK is all about developers. In this release, the SDK targets mobile developers, adding Android and iOS samples for detecting the person’s emotions and gender. Samples for Android Studio and Visual Studio 2015 (targeting Windows platform) are now available.
Live Demo Apps
Since FaceSDK is a developer platform used for building apps, Luxand built a number of apps and services to demonstrate the abilities of its platform. youmask is an app for Android and iOS to show what can be done with the SDK. For building apps like this, developers can use MirrorReality SDK.
Buy One Get One Free
The new build is here. What are you waiting for? Buy one, get one free before the end of September! In order to claim your discount, just fill in the quote request form and we will get back to you as soon as possible.
Luxand Contribution to Academic Research
09/07/2016
During the recent years, Luxand made substantial contributions to academic research and development. Our tools and technologies were used and referenced in over 150 papers published in renowned scientific magazines. We feel that gathering all these publications on a single Web page will help researchers find information they need.
Meet our all-new Academic Research Web page.
On this page, you’ll find more than 150 publications. Where possible, we accompanied those with full article links.
Did we miss anything? If you have a paper referencing our tools or technologies, or if you know of any publication that we missed, please share it with us! We are happy to accept publications using or mentioning our products and technologies. You are free to submit references to research papers in any format, or just use bibtex.
Mirror Reality SDK for Android
06/17/2016
We’re excited to announce the availability of our Mirror Reality SDK for the Android platform. The new SDK enables Android developers building augmented reality apps that are based on the detection and transformation of human faces. The new SDK applies real-time video face effects, allowing developers build high-grade face swapping apps for Android smartphones and tablets.
Mirror Reality SDK is now available on both major mobile platforms.
What does it look like? Check out the youmask app we built using Mirror Reality SDK!
You’re not limited to what you see in youmask. Build apps that turn the phone into a magic mirror, with your app applying real-time animated special effects to the user’s face. Turn the user’s “reflection” into a zombie, make them younger or older, apply animal masks or use fantastic special effects. Your imagination is the only limit! Have something special in mind? Contact us, and we’ll be happy to build it for you!
Face Masks in Live Video for Your Apps
03/18/2016
We’re excited to announce that we’ve just released MirrorReality SDK for iOS developers allowing to build augmented reality apps based on human faces. The new SDK applies real-time video face effects, allowing developers build high-grade face swapping apps for iPhones and iPads.
What does it look like? Your users would use the phone as a mirror, while the app applies real-time animated special effects to their face. You can turn the user’s “reflection” into a zombie, age or rejuvenate their face, accessorize, apply animal masks and fantastic special effects. If you have something special in mind, you can easily create your very own mask or special effect, or simply contact us – and we’ll make it for you.
We’ve built a demo app to show just how powerful the technology is. Download MirrorReality and check out how our new SDK fares in the real life!
Share Your Valentine with Buy One, Get One Free Deal
01/28/2016
St. Valentine's day is coming, and we’re announcing a great Buy One, Get One Free deal! For this Valentine’s day we prepared something truly special: the second FaceCrop SDK license is free when you buy one!
What is FaceCrop SDK, and why do we feel it’s a perfect Valentine gift? FaceCrop SDK is a one-click pony, providing single-call access to detecting and cutting faces out of an image regardless of its size, format or resolution. Just process a bunch photos in various sizes and formats, and you’ll get a bunch of perfectly square (or rectangular if that’s what you’re up to) head shots.
In order to claim your discount, you’ll need to order FaceCrop SDK before February 14th:
http://www.luxand.com/facecrop/order
FaceSDK 6.1: Emotion Detection and Facial Expression Recognition
12/15/2015
The updated Luxand FaceSDK 6.1 is out with new Facial Expression Recognition. Buy one, get one free if you order before December 30, 2015!
Facial Expression Recognition
What exactly is Facial Expression Recognition? The initial implementation can just detect two expressions: whether the person is smiling and whether she has her eyes open. The technology works in both still images and video streams.
How well does it work? Our internal tests demonstrated recognition quality of 92% for smile recognition and a nearly 100% success rate for open-eye detection.
What is it good for? The possibilities are truly endless. Build a happiness detector in your app, and you’ll be able to measure whether or not the target audience is happy about what they see, automate marketing research or build highly personalized marketing campaigns with two-way feedback. When building a photo album, you can help your customers pick happy photos where the people are smiling – as opposed to ones with grim and neutral expressions.
Open-eye detection comes handy when building camera apps with blink detection, helping photographers get perfectly timed shots in full auto mode. If you are developing a DVR system, especially for the car industry, detecting whether the driver has her eyes open helps implement an extra security layer by sounding an alarm if her eyes are shut for longer than a few moments.
We added a new piece of sample code. Expression Recognition sample will help developers quickly integrate new functionality into their apps.
The possibilities are truly endless. Your imagination is the only limit!
Download FaceSDK 6.1 here:
http://www.luxand.com/facesdk/download
Limited Time Offer
We like giving out discounts, especially during the holiday season. How about saving 50% off the price of a two-pack? Buy one license and get the second one free of charge – but only if you order before December 30, 2015!
Mirror Reality 2.0: Must Be Seen To Believe
10/27/2015
One of our tools just got a major update. The new Mirror Reality 2.0 adds a bunch of visual effects, including a very special one – just before the Halloween season! Turn your girlfriend into a fatty or do just the opposite by making her slimmer or applying the rejuvenation effect. And we’ve still got the classic Aging effect! You truly must see it to believe. Just watch this YouTube clip to discover what Mirror Reality 2.0 can do:
Happy Halloween! Make your friends look like a bunch of scary zombies! Scared? You can use this technology on your very own Web site. Just connect to our Zombie API: http://luxand.com/makemezombie/, and you’re set to go.
Want to try Mirror Reality 2.0 just for fun? It’s completely free for personal use! Considering using the product for business purposes? We have a limited-time offer: a commercial license for two devices for the price of a single-device license! That’s the price of only $950 for the two licenses till the end of October. Hurry up! Halloween is coming!
Click here to try Mirror Reality 2.0 now: http://www.luxand.com/download/LuxandMirrorRealitySetup.exe
System Requirements: Intel i5 Processor or higher.
To get your Halloween discount, just place an order for Mirror Reality 2.0 before October 31, 2015.
FaceSDK 6.0.1 improves compatibility of samples with iOS 9
09/22/2015
We’ve just released FaceSDK 6.0.1, a minor update which improves compatibility of the LiveRecognition and LiveFacialFeatures samples with iOS 9 and Xcode 7. The binaries of the library are unchanged, so you may continue using FaceSDK 6.0 if you did not use these samples. As you may know, Apple has recently released iOS 9. We had to update our samples to ensure they work fine with the final release of iOS 9.
Hurry up to claim your discount and get two licenses for the new FaceSDK 6.0 for the price of one! If you are already a customer, you can upgrade your FaceSDK 5 license to the latest release and save 20% off the regular upgrade price. Act now before the discount expires! The action runs till the last day of September!
FaceSDK 6.0: 60-FPS Benchmark for Facial Feature Detection and Tracking in Video Streams
09/17/2015
Luxand, Inc. releases a major update to FaceSDK, a leading face recognition and facial feature detection suite. FaceSDK 6.0 makes use of a new, highly optimized Facial Feature Detection and Tracking engine to deliver smooth facial recognition in high-FPS video streams on both desktop and mobile platforms. In addition, the new release further improves recognition quality and includes new samples for .NET, iOS and Android developers. New customers are eligible for a time-limited two-for-one deal, while existing users can upgrade to the new release at a 20% discount. Sale ends on September 30, 2015.
Benchmark 60-FPS Facial Feature Detection and Tracking
The lower latency and improved performance of the new Facial Feature Detection engine help deliver true 60-fps performance on desktops and 15 FPS on mobile platforms such as iPhone 6. The low-latency engine now allows creating mirror-like augmented reality projects for applications such as virtual makeup, face morphing, animating or aging.
Watch stunning augmented reality effects achievable with the newest facial feature detection:
Improved Precision and Reduced Jitter
FaceSDK 6.0 improves precision and reduces jitter while detecting and tracking facial features in video streams. In addition, FaceSDK 6.0 improves face recognition in video streams, allowing to track faces located near the edge of the frame. Faces that only partially appear in the frame can now be recognized and tracked.
The lower latency of the updated engine used in FaceSDK 6.0 greatly reduces jitter on video streams, enabling smooth, jitter-free appearance of processed videos.
Reduced False Acceptance Rate
With every new iteration, FaceSDK improves recognition quality. In terms of benchmarks, the new SDK makes advances in recognition quality on motion streams, improving False Acceptance Rate by up to 20% when using Tracker API. FaceSDK 6.0 achieved the new low False Rejection Rate of 6.1% on still images (compared to 6.7% in the last release of the SDK; lower is better) as measured in FRCG test with False Acceptance Rate set to 0.1%.
Additional Samples
New sample code is made available to Java, iOS and Android developers. The new samples demonstrate how to use real-time facial feature detection and tracking on mobile platforms. A sample to search for similar pictures in Java is added, and buffer functionality (for saving and copying) for Java and .NET development platforms is now available.
Pricing and Availability
Luxand FaceSDK 6.0 is immediately available. Version 6.0 constitutes a major upgrade. Users of previous versions are recommended to update their product to the latest release. A 20% early adopter discount is available on all upgrades. This limited-time offer is available to existing licensed users of FaceSDK 5.0, and expires on September 30, 2015.
New customers are eligible for a time-limited buy-one-get-one-free offer. New customers receive their second license free of charge when ordering a full license for FaceSDK 6.0. This special offer is available to new customers only, and expires on September 30, 2015.
FaceSDK 6.0 is available at http://www.luxand.com/facesdk
Offer Your Visitors Some Fun in Your Promo Campaigns!
09/09/2015
The sizzling summer is over. It’s back to work season for us, and back to home entertainment for our customers – and for your visitors. Offer your visitors some online fun using Luxand’s face animation and transformation technologies! Universal Studios, Ford, Danone, Unilever, P&G, Johnson’s Baby are already using Luxand tech.
We developed some amazing technologies based on interactions with the human’s face. Make your clients younger or older. Show your visitors how their babies would look like if they married that celeb. Give out realistic 3D avatars made from your visitors’ faces, and do much more with Luxand API’s!
We’re offering a Web API to let you create some very unique promotional campaigns such as MakeMeBabies.com (a highly viral project, 30M+ babies created), In20Years.com, MakeMeZombie.com, 3D-avatars and smileys (luxand.com/smileys or luxand.com/cards).
New app: Luxand BabyMaker for iPhone/iPad
07/14/2015
Do you know MakeMeBabies.com? That online entertainment portal can render a picture of a baby based on two parent’s faces. We’ve brought this technology to Apple iOS, releasing an app for iPhone and iPad to do just that – in offline. Luxand BabyMaker for iPhone/iPad will analyze the two faces, create a mathematical model of their future baby, render that baby’s face, and make a photo collage – all that for only $0.99. Used to create some 30 million faces, the technology was featured in Graham Norton Show by Jennifer Lopez, and showcased by Globo TV in Brazil. BabyMaker’s technology is super popular, smart, funny and family-friendly.
Download Luxand BabyMaker on the AppStore!
FaceSDK 5.0.1 for iOS Goes 64-bit
01/28/2015
As you may already know, starting February 1st Apple will stop accepting new AppStore submissions without a 64-bit version. Luxand is dedicated to supporting its customers, and is proud to roll out a new build of FaceSDK featuring a 64-bit library for Apple iOS devices. Version 5.0.1 comes with both 32-bit and 64-bit versions for iOS, allowing building compatible apps for the Apple AppStore. As usual, the update of FaceSDK maintains full code compatibility with previous versions, offering a seamless upgrade path to all who update to the latest build.
Download the new build at http://www.luxand.com/facesdk/download
Luxand FaceSDK is a true multi-platform, high-performance library for detecting and recognizing human faces on still photos and in live video streams. Developers can use FaceSDK to build a wide range of security, biometric identification and entertainment applications for most popular desktop and mobile platforms. FaceSDK offers fast and reliable operation based on the FRGC tests.
Show Live Emotions with e-Cards API
02/05/2014
We have developed an API allowing to create 3D animated e-Cards with live emotions:
These e-Cards contain the user’s face - acting and moving in 3D and showing a wide range of emotions. To create an e-Card like that, all you need is a single photo of a human face. The system will do the rest automatically.
The new e-Cards API makes it easy to create birthday cards, season’s greetings and other types of electronic postcards. e-Cards are perfect for virtual marketing campaigns: they have a great wow-factor, and are shared by the recipients in Facebook. We have a number of ready-made e-Card templates, but feel free to create your own!
GrowMeUp Helps Children Look Grown-Up in Pictures
09/17/2013
Luxand, Inc. in collaboration with Goldbar Ventures released a new application that helps children produce photos of themselves growing up. The new tool is called GrowMeUp. It’s based on Luxand’s years of experience developing biometric identification and morphing technologies.
About GrowMeUp
Using GrowMeUp could not be made simpler (after all, the target audience is very young). The user will upload a picture containing their face, specify their gender and ethnicity, and choose among the many professions available. GrowMeUp will then automatically identify the face and its features, “grow it up” by applying Luxand’s proprietary aging technologies, and carefully embed the resulting “adult” face into a photo showing a working professional.
Kids will have a wide range of professions to choose from. GrowMeUp contains photos of the following persons: Astronaut, Chef, Doctor, Firefighter, Lawyer, Policeman, Musician, Teacher, Pilot, Soldier, and Model.
The app is available at Apple Store. The online version is also available to users without an iOS device at http://growmeup.com/
FaceSDK Brings True Motion-Based Face Recognition, Adds iOS and Android Support
08/27/2013
Luxand, Inc. announces a major update to FaceSDK, a leading face recognition and facial feature detection suite. FaceSDK 5.0 adds a multitude of new features, enabling real-time, motion-based subject identification in video streams with no prior enrolment. iOS and Android platforms are now supported, enabling developers build mobile apps implementing facial identification and recognition. In addition, Version 5.0 adds automatic gender recognition and improves face recognition rates compared to earlier releases.
Motion-Based Recognition
The newly developed FaceSDK 5.0 motion-based recognition system enables developers build video-based comprehensive surveillance and security systems, CRM desks and attendance control applications requiring no prior enrolment. Implementing true motion-based recognition and identification, the system offers significantly higher recognition rates compared to current-generation systems based on key frame extraction.
Traditionally, video recognition systems were not built truly motion-based. Instead, still key frames extracted from the video stream were used to identify human faces. This approach discards motion-based information, and requires complex training and lengthy enrolment of each identifiable person. Enrolling subjects could be a lengthy and cumbersome process involving several minutes of posing, and requiring multiple captures from various angles.
Years of research and unique, patent-pending algorithms made true motion-based detection possible in real time. FaceSDK 5.0 employs true motion-based face recognition and identification, eliminating the enrolment procedure completely. Every subject appearing in a video is automatically identified and tagged. Enrolling the subject now becomes a simple matter of linking the unique tag to a database record or just putting a name on it.
iOS and Android Support
Mobile developers can now use the many features available in the mobile version of FaceSDK. Working on Apple iOS and Android platforms, FaceSDK 5.0 can be used to develop the widest range of mobile apps for the two popular platforms. Luxand published a demo tool in Apple Store and in Google Play. Developers can download the demo by looking up Luxand Face Recognition app in Apple Store and in Google Play.
Automated Gender Recognition
FaceSDK 5.0 adds the ability to automatically identify gender of a person appearing in the picture or video. This feature is in high demanded by retail advertisers and marketing specialists. Success rate is 93% on still pictures, and 97% in motion video streams.
Improved Recognition Quality
Version 5.0 improves face recognition quality on still images once again, reducing false rejection rate to only 6.7% (down from 9.9% in version 4.0) based on FRGC tests. The false rejection rate is calculated based on 0.1% false acceptance.
Pricing and Availability
Luxand FaceSDK 5.0 is immediately available. Version 5.0 constitutes a major upgrade. Users of previous versions are recommended to update their product to the latest release. A 20% early adopter discount is available on all upgrades. This limited-time offer is available to existing licensed users of FaceSDK 4.0, and expires on September 30, 2013.
Create Personalized Promo Campaigns Starring Your Customers
05/15/2013
We invented a revolutionary new way to target your audience in animated commercials. Let your customers be the stars of your commercials! Imagine a video clip promoting a revitalizing crème, selling a beach vacation or advertising winter sports – starring your customers. Our innovative technology allows creating realistic 3D avatars from any frontal face picture. We already tested the technology on several portals with great results, and we invite you to join.
Place your customers in your commercials in just seconds by simply snapping their picture and pushing a button! In just a brief moment, a video clip will begin, starring your customer. Be it an ad promoting a new line of cosmetics, girlie things or a cool macho product, the highly personalized effect of your hi-tech ad is a sure way to impress the customer.
Check out http://luxand.com/smileys/, an app that can turn any facial photo into a set of animated smileys with a range of facial expressions. Our 3D modeling technology is not limited with just smileys. You can use it to produce animated cartoons and 3D movies with 3D actors and models that look like those on the photo.
Interested? Joining is easy. No upfront payment is required. We’ll help you make use of our technology with your video clips – free of charge. The use of our services will be offered at a low cost of just 25 cents per incident if purchasing blocks of 10,000 requests per month.
More information about the new technology is available at http://luxand.com/smileys/
Develop Your Very Own Product with Hi-End 3D Avatar Modeling Technology
05/07/2013
We invented a great technology allowing to create realistic 3D avatars from any frontal face picture. While we are using this technology in our own tools and Web sites, we’re inviting everyone to join.
What can it look like? Check out http://luxand.com/smileys/, an app that can turn any facial photo into a set of animated smileys with a range of facial expressions. Of course, our 3D modeling technology is not limited with just smileys. You can use it to produce animated cartoons and 3D movies with 3D actors and models that look like those on the photo. Models are able to express a range of emotions such as anger, laugher, stress, smile, and so on. The models can then be rendered in one of the popular graphic formats (e.g. GIF) to be used in mobile applications, animated postcards, chats, games and social networks.
Interested? Joining is easy. No upfront payment is required. We offer this technology on revenue-share basis, asking a small percentage-based commission from each sale of a product based on our technology. In addition, a nominal monthly fee is charged to cover the use of our cloud Web service. Our approach is simple and straightforward. You own 100% of the final product, paying us some 10% of revenue brought by that product.
More information about the new technology is available at http://luxand.com/smileys/
Luxand Develops 3D Avatar Modeling Technology
01/30/2013
We have invented yet another technology. This time it’s about creating 3D avatars out of facial photos. We already tested the technology in one of our own tools, PhotoSmileys. The app allows creating animated smileys with a range of facial expressions out of a single picture containing a person’s face. It works like a charm:
The new avatar technology can automatically turn any facial photograph into a convincing animated 3D model. That model will be able to express a wide range of emotions such as anger, laugher, stress, smile, and so on. The models can then be rendered in one of the popular graphic formats (e.g. GIF) to be used in mobile applications, animated postcards, chats, games and social networks.
You won’t have to install or configure anything if you decide to make use of our new technology. We are about to start providing access to the API hosted on our servers. Please feel free to contact us to discuss the details!
More information about the new technology is available at http://luxand.com/smileys/
Limited-time event: Mirror Reality: Aging free license giveaway
01/15/2013
We want to share some holiday spirit with those who need it most, and are offering a limited-time giveaway. Request your Mirror Reality: Aging license before January 31, 2013, and receive a license for absolutely free!
Who Can Apply for a Free License
We are giving away the free licenses to social agencies, non-commercial organizations and educational facilities. Are you a college, university or charity? Get in touch soon to claim your free license! Hurry up, the time is limited: we are giving away licenses until January 31, 2013!
What is Mirror Reality: Aging
Mirror Reality: Aging is a software tool that turns your webcam-equipped PC into a magic mirror, altering age of whoever looks into it. With Mirror Reality: Aging, you can make people suddenly look older by simply moving the slider:
http://www.luxand.com/mirror-reality/
Mirror Reality: Aging is a perfect anti-drug propaganda tool: you can use it to show how a drug addict will look like in just a few years. Highly sophisticated and thoroughly optimized biometric morphing and animation algorithms deliver convincing real-time performance with no perceivable lag.
Get Your Free License
To help us issue your license faster, make sure you know which PC, workstation or server will be used to host Mirror Reality: Aging, as we’ll need that computer’s Hardware ID to produce your free license.