Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Posters here seem to be stuck with the old mentality that fingerprint is 1:50,000. That number is based on capacitive 2D fingerprint sensors with 500 DPI introduced by Apple in 2013.

Current 3D ultrasonic and optical scanners look beyond the fingerprint and into the blood capillaries to prevent spoofing. The DPI has doubled.

As I just noted, adding a third dimension adds accuracy/reduces false positives. That's true whether we're assessing faces or fingerprints.

It's not whether DPI is doubled - it's how that greater resolution is used. How many measurements are taken, not just how accurate they are. You could make 500 measurements with a 500 DPI scanner, or 250 measurements with a 1,000 DPI scanner... Arguably, 500 points of comparison would be superior to 250 points of comparison, even if those 250 points are measured with greater resolution. Of course, ongoing increases to computing power mean it's likely that a higher-resolution scan would also be accompanied by an increase in the points of comparison... but I think you get my point.

The accuracy of any biometrics system stands or falls on how many measurements it can take and how distinctive the points it measures can be. I doubt that horse race is over. My appreciation for Face ID has less to do with statistics (which can change with every new iteration of a system), and more to do with functionality. "Touch vs. glance." I don't have to touch the screen to be recognized. Like voice-control vs. mouse/keyboard control - glancing and speaking can be done while hands are otherwise employed.
 
As said in other reports, if they do bring it back then it’ll be for their low budget designs in emerging markets. So for majority in this thread, its dead. If the notch was that much of an issue aesthetically, nothing was stopping them from FPS on side or even back of phone.
 
There can be huge differences between two people or fingerprints that seem to look the same, once you reduce them to measurements (biometrics, after all, means life/body measurement). The reason a twin or other close relative may fool Face ID has as much to do with underlying bone structure (distances and angles between, say, the tip of the nose, point of chin, corners of the mouth, and bottom of ear lobes) and much less to do with some of the cosmetic things we consider when assessing similarity. Genetic similarity is more than skin-deep.

Both Touch ID and Face ID are measuring angles, point-to-point distances, etc. - mathematical relationships, angles, and ratios between selected features, rather than actual images ("this face/fingerprint looks like that one").

Perhaps the key reason Face ID is superior to Touch ID (and other, less successful facial recognition systems) is that it's measuring those relationships in three dimensions rather than two - there's simply more to work with. Length x width vs. length x width x depth. Two people who look similar in full-front 2D photos may look less similar when you compare their profiles (one of the reasons the police take mug shots from various angles). Someone with a close genetic tie is more likely to be similar from every angle, not just some.
Sure I understand that, but I don’t think that really nullifies my point. Part of what makes one person look like another person are those similar measurements, even in underlying structures. The human eye/brain naturally recognizes those things, and of course more accurately when seen in real life (3d). So even if the human eye and Face ID don’t look at all the same things, there is overlap and a person’s eye can at the very least get him/her closer to what Face ID is looking for, which again would throw off the statistic.

But again, my greater point is that this better security debate seems pretty moot. It’s seems like debating whether it’s more likely a bird can fly to Mars or Jupiter. In other words, is anyone (Touch ID or Face ID user) really concerned that someone will steal their device AND just happen to have their same biometrics (except for those with nefarious siblings)?
 
As I just noted, adding a third dimension adds accuracy/reduces false positives. That's true whether we're assessing faces or fingerprints.

It's not whether DPI is doubled - it's how that greater resolution is used. How many measurements are taken, not just how accurate they are. You could make 500 measurements with a 500 DPI scanner, or 250 measurements with a 1,000 DPI scanner... Arguably, 500 points of comparison would be superior to 250 points of comparison, even if those 250 points are measured with greater resolution. Of course, ongoing increases to computing power mean it's likely that a higher-resolution scan would also be accompanied by an increase in the points of comparison... but I think you get my point.

The accuracy of any biometrics system stands or falls on how many measurements it can take and how distinctive the points it measures can be. I doubt that horse race is over. My appreciation for Face ID has less to do with statistics (which can change with every new iteration of a system), and more to do with functionality. "Touch vs. glance." I don't have to touch the screen to be recognized. Like voice-control vs. mouse/keyboard control - glancing and speaking can be done while hands are otherwise employed.

Given that most users touch their phone to wake it, full display fingerprint technology should be almost transparent to users. I understand the preference for "glance" but I think most users interact with their phones with some form of touch. Face ID hardware works well in a usage model with absolutely zero touch but how many consumers are using voice control?

If we believe powerful software and the neural engine can eventually perform attention aware and gesture control using only the CIS camera, is there still a good reason to keep the TrueDepth sensor on the front?

As you mentioned, it's an ongoing race between fingerprint and face biometrics. In terms of cost, space, power, there are a lot of benefits for fingerprint. As a result, I think the chance of it returning in a much better form - are quite a bit higher than some believe.
 
In terms of power, ultrasonic readers are measured in hundreds of microwatts for continuous operation.

For Face ID, the VCSEL flood illuminator uses a couple hundred milliwatts. At this point, there is no way to use Face ID for continuous biometrics.



Current fingerprint scanners read blood flow because they scan the blood capillaries. As a result, they can measure heart rate and potentially blood sugar level.

Face ID offers attention aware features. Sooner or later, these features will be duplicated by using the conventional CIS camera in combination with software using the neural engine. It's the whole 3D Touch situation all over again. Why spend money on the hardware if software can provide 80% of the benefits at 20% of the cost?

There's no such thing as "continuous operation" in any of these systems. They measure only as often as they need to measure. In terms of energy economy, it's foolish to do anything else.

Sampling rate is determined by rate of change. The more rapidly something changes, the higher the sampling rate and the more "continuous" measurement will seem. Compare the sampling rate required to capture a human heart rate (a frequency of nominally 20-200 beats per minute) to the sampling rate for audio (nominally 20 to 20,000 cycles/beats per second). In the case of "continuous" blood sugar measurement, what's the necessary measurement rate to detect an unhealthy change? Once every ten minutes? Less than that?

In the case of smartphone auto-lock (detecting the absence of user interaction), the sampling rates can be pretty low. Currently, the minimum auto-lock interval on iPhone is 30 seconds - two beats per minute. And the longer the auto-lock interval, the lower the sampling rate can likely be. It's a matter of re-setting that timer whenever an interaction is detected - either touch alone, or in the case of devices with Face ID, either touch or glance. Since a meaningful glance is likely to last at least one second, sampling at a rate of 2 scans per second might be sufficient. It's conceivable to me that glance-recognition doesn't even need to be active for, say, the first 50% of the auto-lock interval. If, from a behavioral statistics standpoint, a person actively engaged with the device glances at it at least once every five seconds, you need detect only one of those glances in the period between reset and expiration of the timer.
 
Given that most users touch their phone to wake it, full display fingerprint technology should be almost transparent to users. I understand the preference for "glance" but I think most users interact with their phones with some form of touch. Face ID hardware works well in a usage model with absolutely zero touch but how many consumers are using voice control?

If we believe powerful software and the neural engine can eventually perform attention aware and gesture control using only the CIS camera, is there still a good reason to keep the TrueDepth sensor on the front?

As you mentioned, it's an ongoing race between fingerprint and face biometrics. In terms of cost, space, power, there are a lot of benefits for fingerprint. As a result, I think the chance of it returning in a much better form - are quite a bit higher than some believe.

There are times when I touch the screen to wake, but more commonly I'm lifting the device to wake. My first touch comes after the device unlocks.

In either case, an input from either touchscreen or gyro is necessary to trigger Face ID. The feature does not unlock on glance alone.

I'm not convinced that "touch anywhere" is more efficient. It may unlock the screen, but the chances that your finger will be properly positioned to make the next functional touch (tap a particular Home screen icon or a button on the currently open app) seems pretty small. Even then, a lift then touch again may be required for the system to infer your intent. Further, the touch needs to be sufficient to provide a good scan. Will a brief tap with a fingertip provide a good scan, or will resting finger pad on the screen be necessary? One way or the other, there are hand gestures necessary that are unnecessary with a Face ID unlock.
 
Ask how Apple is able to do Portrait Mode on iPhone XR. Ask how Google can sharpen photos so well and do Portrait Mode with a single camera. It's a matter of time before conventional cameras can mimic Animoji functions.

Apple will eventually put the TrueDepth sensor on the rear of the iPhone. It's not as if all research is lost if Apple removes TrueDepth from the front. All that learned research will be used on AR and mapping groups of people and surroundings instead of a single face.

Portrait mode on the XR is limited. I don't think Apple just gives up on the security that FID brings.
 
Naw, if I had another iPhone I would but not a fan of using BETAs on my primary device. - Any kind of improvement on speed unlock would make me happy.

FID is noticeably faster on 13 on my primary XS Max. Been using 13 for about 2 weeks now. I test it side by side with my 12.3 secondary XS Max, it’s faster as Apple promised.
 
  • Like
Reactions: Hi Tek
It seems unlikely that Apple would go back to finger print scanners when they already have something better in FaceID.
 
It seems unlikely that Apple would go back to finger print scanners when they already have something better in FaceID.

That’s pretty much it summed up nicely, Face ID is better than what a fingerprint scanner probably could offer. And I don’t say that because of the accuracy comparisons, I say that because of the security that face ID offers. That’s what Apple cares about, is the security protocols, which they’ve invested so much R&D in.
 
That’s pretty much it summed up nicely, Face ID is better than what a fingerprint scanner probably could offer. And I don’t say that because of the accuracy comparisons, I say that because of the security that face ID offers. That’s what Apple cares about, is the security protocols, which they’ve invested so much R&D in.

The extra security is nice but it’s the convenience where faceid really shines.
 
I doubt it. I think Face ID is the future and here to stay.

Personally, I don't miss Touch ID the slightest.

They do. It's Ultrasonic under glass. It's simply a question of component supply and pricing. Once those are within margins the switch will be made. The notch will die. I'm estimating perhaps one phone in 2020, and the entire line by 2021. Fairly good odds in 2021 we will see the front facing camera concealed as well. That technology is still maturing. I'll be using 8+'s until then to save my eyes from continuous IR bombardment from the Flood Illuminator. ;)
 
They do. It's Ultrasonic under glass. It's simply a question of component supply and pricing. Once those are within margins the switch will be made. The notch will die. I'm estimating perhaps one phone in 2020, and the entire line by 2021. Fairly good odds in 2021 we will see the front facing camera concealed as well. That technology is still maturing. I'll be using 8+'s until then to save my eyes from continuous IR bombardment from the Flood Illuminator. ;)

Better not go outside either...
 
  • Like
Reactions: Wide opeN
They do. It's Ultrasonic under glass. It's simply a question of component supply and pricing. Once those are within margins the switch will be made. The notch will die. I'm estimating perhaps one phone in 2020, and the entire line by 2021. Fairly good odds in 2021 we will see the front facing camera concealed as well. That technology is still maturing. I'll be using 8+'s until then to save my eyes from continuous IR bombardment from the Flood Illuminator. ;)

I personally think you’ll be waiting forever. FaceID is the new authentication method. We are only seeing under screen FPS because other manufacturers don’t yet have the technology to do FaceID or the underlying OS to support it.

Once other manufacturers start having the technology and the underlying OS supports it, finger print scanners will be gone entirely.
 
Watching Face ID vs Touch ID debate is like arguing wire vs wireless, and there is no end of this debate.

Face ID is great, but not great enough. Especially on iPad. With four betas, the failing rate is lower, around 15% in average. By comparison, Touch ID had a failing rate of 5% during iOS 9 and iOS 10 beta as I recall. The fact that iPhone Face ID failing rate is very low means Apple would bring improvements to make iPad landscape unlock better than what 2018 iPad Pro offers.

As for Touch ID return, I am not gonna be surprised if Touch ID makes a comeback in some form later on. Apple can no longer be predicted with outdated experience. Current Apple should be treated like Trump: both act in an unpredictable fashion.
 
I personally think you’ll be waiting forever. FaceID is the new authentication method. We are only seeing under screen FPS because other manufacturers don’t yet have the technology to do FaceID or the underlying OS to support it.

Once other manufacturers start having the technology and the underlying OS supports it, finger print scanners will be gone entirely.

The technology has been here for over a year. Huawei Mate 20 and Xiaomi Mi 8 use the same 3D depth sensing array as Apple. This includes flood illminator and VCSEL.
 
PS: I have no idea why so many people are advocating the return of Touch ID, as if Touch ID is an inferior technology compared with Face ID. It’s just another form of biometric authentication, and that 1:50000 number probably does not hold up much credit either, as the test situation and related context are unknown for me.
Also, “never say never”. ;)
 
The technology has been here for over a year. Huawei Mate 20 and Xiaomi Mi 8 use the same 3D depth sensing array as Apple. This includes flood illminator and VCSEL.

The problem they have is that android doesn’t support this method of authentication so it can’t be used as a way to pay for things via NFC etc.

Once android supports faceid authentication we will see an explosion in devices with faceid.
 
Fingerprint scanners under the display are a temporary solution. I’m not sure why others want to dismiss face ID being replaced in such a short amount of time, when that’s not even how Apple operates when they bring such a major feature like Face ID to the iPhone, iPad and eventually the Mac. I mean, there’s nothing indicating Apple will abandon Face ID anytime at all. Will they offer another alternative to face ID with the fingerprint sensor under the display? Maybe, but I doubt it. Apple isn’t primarily about offering ‘choices’ with biometric security, that’s why they invested so much R&D into face ID.
 
Fingerprint scanners under the display are a temporary solution. I’m not sure why others want to dismiss face ID being replaced in such a short amount of time, when that’s not even how Apple operates when they bring such a major feature like Face ID to the iPhone, iPad and eventually the Mac. I mean, there’s nothing indicating Apple will abandon Face ID anytime at all. Will they offer another alternative to face ID with the fingerprint sensor under the display? Maybe, but I doubt it. Apple isn’t primarily about offering ‘choices’ with biometric security, that’s why they invested so much R&D into face ID.

Exactly. Android device makers have had to react to Apple’s all screen design. Because they can’t currently incorporate FaceID without having to significantly modify the underlying android OS and/or they don’t have access to the actual hardware, they’ve had to implement under screen FPSs.

Once they have the technology and the OS supports it, FaceID will replace FPSs on android devices too.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.