DIY Touchscreen Analysis
The success of the iPhone has triggered the adoption of touchscreen systems in a wide range of mobile devices, and a bevy of new gadgets equipped with capacitive sensing technology have now hit the market. MOTO has years of experience developing products that use capacitive touch, and we’ve had the opportunity to test many of the latest devices. Our conclusion: All touchscreens are not created equal.
It takes finesse to create a touchscreen system that’s pleasant to use, because touchscreens require seamless integration between hardware components, software algorithms, and user-interface design. If a manufacturer cuts corners or flubs any of the critical elements, the user’s experience with a touchscreen product is likely to suffer.
Simple and True
Although we usually use sophisticated tools to test touch screen accuracy, MOTO has also developed a simple technique anyone can use to evaluate the resolution and accuracy of a touchscreen device. All you need is a basic drawing program (download one if necessary), a steady hand, and a few straight lines drawn very slowly on the screen.
This video shows what happened when we recently took several touchscreen systems out for a test drive:
The Virtue of Slow
Why do you need to draw slowly? On a good touchscreen, users can draw clean straight lines, even while going very slowly, so the graphics that appear on screen accurately represent what was physically drawn.
On inferior touchscreens, it’s basically impossible to draw straight lines. Instead, the lines look jagged or zig-zag, no matter how slowly you go, because the sensor size is too big, the touch-sampling rate is too low, and/or the algorithms that convert gestures into images are too non-linear to faithfully represent user inputs.
Pressure Matters
Also, even on a single device, the amount of pressure and the part of the finger you use on the screen has an impact on how well it senses. A good touchscreen device will produce linear output regardless of whether you’re using the full pad of your finger, or just the dry corner of your cuticle. When comparing devices, make sure to use even pressure across all of them.
If you want to show the most extreme case, draw very lightly with the corner of your finger. The artifacts will increase significantly, showing which device is really the best with a weak signal. This is important because quick keyboard use and light flicks on the screen really push the limits of the touch panel’s ability to sense.
Here you can see the results of our test:
Edge Performance
Take careful note of the performance at the edges of the screen. The performance at the edge is challenging to tune, and separate from the basic “waviness” test. The iPhone tracks all curve very strongly as you approach the edge of the screen, despite a straight finger trajectory. This is especially obvious at the bottom, where the iPhone has a sensitivity problem.
The Droid Eris [Nexus One] is actually the clear winner for edge performance — the signal tracks right off the edge of the screen very consistently.
[edit] As of time of first writing, we hadn’t tested the Nexus One. It does slightly better than the Eris. In fact, they both use the same touch controller IC.
A Quest for High Signal-to-Noise Ratio
To create a superior touchscreen experience, it’s essential to develop a touchscreen sensor that has the highest possible signal-to-noise ratio, or SNR. When a manufacturer gets it right, the device tracks touch inputs almost as if they were connected to physical objects in the real world. Get it wrong and consumers end up with inferior touchscreen systems that are inaccurate, insensitive, and absolutely infuriating to use for typing.
Key drivers of SNR include:
- Conductive sensor material
- Substrate material
- Substrate thickness
- Distance from display (the biggest noise source)
- Sensing waveform
- Sensor pattern
- Sensor pitch
- Analog sensing circuitry
- Sample rate
Touchscreens are a catalyst for innovation and a powerful way for device manufacturers to differentiate their products in an intensely competitive marketplace. But as our demonstration shows, there’s a right way and a wrong way to deploy the technology. MOTO has worked with capacitive touch interfaces for more than 15 years, and here are some essential dos and don’ts for anyone entering the field:
- Don’t skimp on materials. With touchscreen hardware, manufacturers get what they pay for — and consumers will notice the difference.
- Allow ample time to develop your algorithms. Don’t treat touchscreen algorithms as an element of component sourcing; for best results, create a distinct touch development track under your own roof to make sure your products are both responsive and accurate.
- Closely integrate touchscreen hardware, software, and user interaction development, and do so as early as possible in the product development process. Never treat them as separate tasks.
January 8th, 2010 at 10:55 pm
Interesting … I’m glad I don’t do much drawing on my Droid.
January 9th, 2010 at 5:18 am
The iPhone appears to be superior. However at the end of your video you highlight the importance of materials to the signal to noise ratio and it seems you are presenting a 1st Gen iPhone here which used glass where the 3G and 3GS use a plastic. Would it not be more fair to compare the other newer devices with an iPhone which is currently available on the market?
January 9th, 2010 at 7:10 am
interesting topic to compare, but it looks like the finger strokes aren’t applied equally across the phones. i.e. on the third downward-right line for the droid, there’s noticeable finger sway. maybe the examiner was more careful on the first phones and got lazy? this is also evident because the results in the video are far better than the final presented picture.
this could mean that the smaller the variance detected by the phone, the *greater* the sensitivity and/or accuracy, which is the exact opposite of the conclusion. or it might be a reflection of the interpolation/smoothing algorithm and not the hardware itself. still, it’s probably better to have a phone that knows what you’re trying to do rather than one that thinks it’s right
I know this isn’t meant to be a scientific test, but perhaps double-blind testing or multiple-person/different fingers would give more accurate results.
January 9th, 2010 at 12:42 pm
Clear winner? That must clearly be the Google Nexus. To me it is clear that only with the Nexus the lines don’t bend near the edges. Not so with the Droid. Strange conclusion to set the Nexus as the winner in this regard.
January 9th, 2010 at 2:13 pm
What information is there that suggests it matters whether or not the screen can draw a perfectly straight line at low speeds? I can’t think of a single type of interaction that would require slow-speed drawing of straight lines. What interactions would fail on the Droid, vs. the iPhone?
January 9th, 2010 at 2:26 pm
>I can’t think of a single type of interaction that would require slow-speed drawing of straight lines.
How about games using touch screen control?
January 9th, 2010 at 2:59 pm
Does screen resolution come into play? It would seem to me on a device with a high density of pixels per inch, the software would have a harder time to determine just where the line should be drawn. If true, I wonder about the pixel densities of these devices… Does the iPhone have the highest? Or is it the lowest? Does that mean the Apple engineers had the easier job?
January 9th, 2010 at 3:43 pm
Re BHowell: I agree—they shouldn’t have tested an iPhone two models old—they should have tested the current version.
But ALL iPhones (including 3G and 3GS) and ALL iPod Touch models have always used glass screens. Never plastic. The newer models are oil-resistant, but still glass.
January 9th, 2010 at 3:45 pm
Re Dave:
What actions would fail that require drawing a straight line at low speeds?
1. Drawing carefully.
2. Some games.
Besides, it’s just a way to test the screens. Kind of like when they copy 1000 files over and over to test hard disk speeds. Nobody needs to do that particular task, it’s just designed to be a useful test. I’d they came up with a useful test here. Inaccuracies drawing a slow line mean it’s ALSO going to be inaccurate for other actions.
January 9th, 2010 at 4:12 pm
If I might ask, what apps are you using for this comparison? I know there are several decent drawing apps for the iphone, and not any that I know of for android. Thanks!
January 9th, 2010 at 5:10 pm
Did you not hear the explanation that this inaccuracy can lead to faulty touches when typing on a software keyboard? It would also matter on any game where the user creates paths. i.e. Any drawing app, games with paths (e.g. Flight Control) and perhaps even in gestures I suppose.
January 9th, 2010 at 5:22 pm
Joost – I don’t reach the same conclusion – on the light pressure test, the results on the Nexus look worse that the HTC Droid, although on the medium pressure test the Nexus looks a lot better that the iPhone. I’m not sure what conclusions you can draw from that – the amount of pressure expected is subjective, in the same way as trackpad behaviour. (I wonder if any of these devices adjust themselves based on user behaviour?)
January 9th, 2010 at 5:34 pm
First, all generations of iPhones, and iPod touches for that matter, have glass screens… Not plastic. Second this accuracy does matter as anyone who has used a droid vs an iPhone can tell you. It is much easier to accurately touch small HTML links on the iPhone than the droid without zooming in, as one exapmle. One can definitely feel the difference in everyday use in the greater accuracy the iPhone has over other touchscreen phones.
January 9th, 2010 at 6:09 pm
“the third downward-right line for the droid, there’s noticeable finger sway.”
Are you kidding? Look at the other tests. The tracking on the Droid is pretty bad and you can’t write that all off as the tester’s finger swaying a bit.
“What interactions would fail on the Droid, vs. the iPhone?”
I’ve played lots of games and used lots of apps on the iPhone already that require slow and/or accurate finger tracking at some point or another, but I understand that having a wide variety of games and apps to use is not a problem for owners of Android phones.
January 9th, 2010 at 6:14 pm
“Interesting … I’m glad I don’t do much drawing on my Droid.”
And neither will professional artists. They can do it on an iPhone however.
http://news.cnet.com/8301-1023_3-10250989-93.html
January 9th, 2010 at 6:35 pm
I have to agree with Dave: there’s no evidence that your test has anything to do with the actual usability of a touch screen display.
Also, I wonder what you mean with “sensing waveform” or “sensor pitch”. These words look a bit like random keywords copied from wiipedia…
January 9th, 2010 at 6:43 pm
@Dave
It illustrates how accurately the device interprets touch
January 9th, 2010 at 7:05 pm
The iPhone is a clear winner, for the simple fact that, as stated in the video, the straighter the lines are, the less difficult it’ll be to make mistakes when typing on a virtual keyboard. And since using a virtual keyboard (or any small-sized keyboarad) is a pain in the ass by itself, the iPhone stands out victorious.
January 9th, 2010 at 7:15 pm
Very interesting comparison, but one has to wonder whether the wavyness on the other phones is a result of a poor algorithm or perhaps an intentional algorithm that is biased towards horizontal or vertical interactions.
I am not familiar with the algorithms at all, but personally if I’d be designing these systems I’d definitely introduce an option to use an error correcting algorithm for use in standard interfaces (button, sliders, etc.) where the correction assumes the intention of mostly horizontal or vertical interactions. Of course, this won’t be applicable in a free-form interaction such as a drawing application, where it’d need to be turned off by the developer of the application.
Does anyone know if such algorithms are at play here? Or at all in touchscreen algorithms? If it is, that could explain the wavyness — it is a bit too constant and not as irregular as one would expect if it is solely caused by inaccuracy.
Just throwing ideas out there
January 9th, 2010 at 7:35 pm
@ BHowell: I’m pretty sure both the 3G and original iPhone have glass screens.
January 9th, 2010 at 7:59 pm
For Mk.II, repeat the pictures with vertical & horizontal lines.
The reason I say this is that it would be interesting to know whether the error is in one or both of these. The interface uses the two differently, for example aren’t more buttons rectangular than square? But horizontal scrolling is more rare than vertical.
January 9th, 2010 at 8:02 pm
While “stairstep” artifacts clearly indicate inaccurate tracking, very smooth lines can be a sign of a low sample rate for tracking.
January 9th, 2010 at 8:31 pm
Dave: as the article explains, inaccurate drawing of lines means the screen is unable to accurately determine where you are pressing, and that will drastically increase errors when typing (errors which will appear to be user-error, but are in fact a hardware issue).
January 9th, 2010 at 8:51 pm
Dave,
It’s not that it is important to be able to draw straight lines, this test is to show the accuracy and sensitivity of the touchscreen ie how accurately it will interpret your touches and movements. So for example if you were surfing a webpage and there were a list of links it is less likely that you will touch the wrong one.
January 9th, 2010 at 9:13 pm
The statement ‘noticeable loss of sensitivity at edges’ for the iPhone is probably an incorrect analysis. If you move the fingertip off the edge of the screen at an angle of 45 degrees the middle of the fingertip, which is tracked while the fingertip is on the screen, will have left the screen when one side of the fingertip is still on the screen and is being tracked quite accurately.
Also the statement in the movie that the iPhone has a sensitivity problem, especially at the bottom of the screen is probably inaccurate. For proper verification one should perform tests starting to draw lines at both ends of the screen. Also, it would help if the test gives hard information about the area of the finger that is touching the screen. More high-tech testing…
January 9th, 2010 at 9:30 pm
@ BHowell
1. All generations of iPhones have glass screens. The 3GS has an Oleophobic Glass Screen. That’s the only difference. Its a coating I believe.
2. He was using at least a 3G or 3GS. That is not a 1st Gen iPhone. Its a white 3G or 3GS.
January 9th, 2010 at 10:22 pm
@BHowell
All versions of iPhone use glass screens.
@Joost van Tongeren
Right, even if it’s true, it would be useless because almost no UI on successful touchscreen phones is on the very edge of the screen. People don’t use the edge for various usability reasons.
January 9th, 2010 at 11:03 pm
@Orenge: Thanks for the clarification on screen type. I own a model from each generation and the latest iPod Touch, only the first seems like glass hence my comment. But with your info and some further investigation on http://www.isuppli.com it is clear you are correct. That oil-resistant glass on the newer models obviously makes the difference as you pointed out.
@icruise: Thanks also.
January 9th, 2010 at 11:42 pm
Can’t see the video on my iPhone.
January 10th, 2010 at 12:15 am
Cool test, I love it!
You showed that the iPhone has some issues near the bottom, but I believe this is really just a bias from the drawing direction. Would you see the same thing at the top if you reversed the drawing direction? In the end, it must come down to the (i,j) order in which the matrix of 2D touch zones are processed, and how they handle the reduction of information as you get near the screen edges (a numerical boundary condition, so to speak). I think it would be interesting to take one device, maybe iPhone since it shows a prominent edge/bias effect, and look at it with lines drawn in different directions to see if there is an actual hardware/software skew, or just a skew based on the direction of drawing.
January 10th, 2010 at 12:17 am
This test is not well designed. To do this test properly, you would need a mechanically driven “finger” which is known to be absolutely straight. Otherwise, even small deviations from what is believed to be a straight line by the human finger are likely to cause variations. Rather than conclude that the Driod is inferior to the iPhone in this test, I would conclude that the Droid is more sensitive to small motions than the iPhone, and shows those small motions better. Of course, these small deviations might be “filtered out” or “averaged” to make the line appear more straight than the actual data would cause it to be in the iPhone. In some cases, that could be good, in other cases that would be bad. Overall, the test is useless and a waste of time. When your experimental technique improves, do the test again.
January 10th, 2010 at 1:16 am
More than 50% of the time I try to hit a link on a web page on my Droid, I get the wrong one, when it’s OBVIOUS that I’m pressing the RIGHT one. Now I finally know why.
January 10th, 2010 at 4:50 am
I certainly hope no one uses this “study” to guide them when determining which phone to buy. There are more holes in this than a piece of Swiss cheese.
I’m not saying that the screens don’t track like the lines suggest, because they probably do. But these lines are just a surrogate for what we really care about – accuracy in typing and swiping. Those could be measured in real world simulations much better than by just displaying the lines that are traced.
I agree with Fra, but there are much more valid ways to measure this, such as measuring the actual accuracy of clicking links on a cluttered webpage on various platforms.
If differences exist, this could provide an explanation why. However a phone with the best line tracings isn’t *necessarily* any better than the worst one.
January 10th, 2010 at 5:17 am
@BHowell and @Orenge
I have never noticed any difference in screen performance between my first gen iPhone and my 3G and my 3GS and that is backed up by trying the same drawing tests on my iPhone 3GS.
I agree that using a current iPhone would make more sense but given the results, there was really no benefit to using the new model over the old.
January 10th, 2010 at 5:55 am
The iPhone 3Gs screen is glass with a bonded oleophobic polymer coating.
http://gizmodo.com/5302097/giz-bill-nye-explains-the-iphone-3gss-oleophobic-screen
January 10th, 2010 at 6:50 am
I am a Storm user. How does this test run on a Blackberry Storm and a Storm 2. Can you compare this as well to the iPhone or iTouch.
Thanks.
January 10th, 2010 at 5:13 pm
Hi All,
The Nexus went up after the initial writing, so I originally described the Eris as the best edge performer.
The Nexus is better in that regard. Edited for truthiness.
The iPhone used in the photo tests is a 3G — the video was my older 1st-gen.
They perform very similarly.
Cheers
morgan
January 10th, 2010 at 6:22 pm
Wow, suddenly lots of electronic touch sensor design experts in the comments. Read the introduction – it says “Although we usually use sophisticated tools to test touch screen accuracy, MOTO has also developed a simple technique anyone can use to evaluate the resolution and accuracy of a touchscreen device.”
This is a simplified test to allow normal users a glimpse into the design compromises that are made during touch sensor development. The conclusions drawn seem fair and balanced to me – the iPhone tracks more accurately but has a significant roll at the edges. The Droid has significant waviness that clearly comes from the sensor, not from small movements of the tester’s finger. This seems to fit with what I know from experience and have read about these phones. The important thing here is not whether we can draw diagonal lines on our phones – it’s to illustrate the trade offs that are made in the design phases. People should just jump to pro-Apple or pro-Droid camps – there’s reasons for the choices that each company made, but they may not suit the way you work with the phone.
@biggsjm – It’s clearly a first generation iPhone. The edge shape is wrong for a 3G / 3GS and looking at the bottom you can clearly see the contrast between the black antenna area and the back of the case. A white 3G phone is white all the way down because the back of the case is made of plastic. In the first gen iPhone the case back was metal, so they needed a plastic cover for the antenna.
January 10th, 2010 at 11:39 pm
Wha application can I Dow load free to test my iPhone other devices?
Respectfully,
Tony Magnifco
January 11th, 2010 at 1:13 am
hello,
why not add to the comparative HTC HD2. I think the multi-touch this device is very effective
January 11th, 2010 at 6:31 am
The results on the Nexus look worse that the HTC Droid, although on the medium pressure test the Nexus looks a lot better that the iPhone.
January 11th, 2010 at 6:32 am
CRAP TEST!
I dl’d the first free draw appp I could find on droid. I got PERFECTLY STRAIGHT lines.
I’m not the Fanboy type. I own a Droid. I like my Droid. I would love an iPhone. But this test is just pathetic.
January 11th, 2010 at 7:30 am
I wish they had included the Blackberry Storm for their testing. I’d like to know how my phone compares to these others, especially the iphone.
January 11th, 2010 at 7:35 am
Odd that this comparison fails to mention that the screen resolution on the Motorola and Nexus is significantly higher than the iPhone, has it occured to you that the iPhone doesn’t stairstep because it lacks the resolution to do so. Naturally if you are zoomed in at 200% or more you are going to draw more accurately than if you are zoomed out.
duh!
January 11th, 2010 at 7:40 am
I am using the program “Draw!” on my droid and am able to get much straighter lines across my screen.
January 11th, 2010 at 7:56 am
That’s pretty interesting. I’ve personally never used a capacitive touch screen, so I’m left wondering how would it compare to resistive one? Such as Nokia N900.
January 11th, 2010 at 8:37 am
I would like to have seen the N900 included in this competition.
January 11th, 2010 at 8:52 am
This is a ridiculous unscientific test which ridiculously fuels anti/pro iPhone sentiments. Now if the test used the same cross-platform app with an identical input method (instead of clearly different drawing apps with a tester “feeling” his way through trying to have identical swipes) then I’d trust the results.
January 11th, 2010 at 9:48 am
Here’s a way to test tap-pointing accuracy:
1) Open up your drawing program, set the pen on it’s smallest size (e.g. 1 pixel)
2) Touch once towards the top of the screen, and touch again towards the bottom. You should have two dots.
3) Now, tap once in between the two dots.
4) Now tap once in between the three dots, and keep going
You should theoretically end up with a perfect line of evenly spaced dots to the point where they fill in to form a line. How far you vary from that shows the variance of the touch screen and the user as a combined system (which is what matters). On an iPod Touch, I can get very straight, evenly spaced results. On a Droid, the variance was significantly worse. (Haven’t seen a nexus yet)
January 11th, 2010 at 10:45 am
Dude i think these guys are funded by apple which is trying to prove android is not a match for stinking apple iphone. I’ll go ahead and fund some bullshit guys to prove that droid is 10x better than apple iphone.
January 11th, 2010 at 11:10 am
Can You test HTC HD2? It has capacitive screen too and I think that it is fairly good.
ChAr
January 11th, 2010 at 11:34 am
I get much straighter lines when I do this test on my droid. However, when I first did this test I got a screen full of wavy diamond boxes…I cleaned up a couple problems with my technique, used a consistent pressure along with a physical finger guide, and made a series of lines that resembles the iphones for straightness.
I’m guessing that the droid is sensitive to where the exact center of the screen contact is, which shifts when the finger pressure changes.
January 11th, 2010 at 12:02 pm
I own the Droid and it looks perfectly straight, except when I curl my finger while moving along the edge of an envelope as a straight edge, even without it, I can still make straighter lines than what’s seen on the iPhones test.
January 11th, 2010 at 12:12 pm
This is why I don’t draw with my mobile device.
http://cl.pixelspotlight.com/
January 11th, 2010 at 12:15 pm
Also, the true test would be, sure decrease the line thickness but also have the phones “clamped” to a conveyor belt and a pen tip point precision of constant pressure above it clamped from moving: but due to the laws of physics when two objects rub it causes vibrations anyway.
The only issue I had with decreased touch sensitivity is with those privacy screens, which was a waste once I found the key scratch test.
January 11th, 2010 at 12:18 pm
This is utterly ridiculous. I instantly pulled out my droid, used DRAW! (free) and was easily able to draw a grid of straight lines. After 5 attempts I still never got those insane squiggly lines you have shown here. This is ignorance at its best.
January 11th, 2010 at 1:22 pm
This doesn’t at all match with my results on a DROID: http://twitgoo.com/beqee
January 11th, 2010 at 1:32 pm
I want to first say that I respect the professional opinions that MOTO stake their reputations on. And for my own personal ddaclosure, I got a Droid Eris on Xmas eve and I am typing this on said phone while at work. I do not use or like iPhones, but I do have to support them as part of my job. This note was only made to qualify my knowledge or he device.
After reading this article I was compelled to try this test on my Eris. I loaded the market, searched for “drawing”, picked the first app, which is called “DrawNoteK” to perform this test, myself. The results were surprising.
Like the testing application they used here, my entire acreen was white and I performed the same tests based on description that they did. I also drew lines the opposite direction to test edge sensitivity.
My Eris drew very accurate and consistent lines both under ‘edge of finger’ light, slow, and what I would consider ‘normal’ pressure circumstances. I was able to produce images such as is shown above, and it was due to my own hand shaking in an effort to draw a straight line. However, I was more consistently able to produced results that appear almost identical to the the “medium, full finger” image for the Eris. I also carefully witnessed the movement of the line under my finger as it moved. There waa next to zero deviation.
I was only able to reproduce the jaggy, wavy lines of the first image if I made a conscious effort or allowed my hand to shake. It is definitely NOT the rule.
With this being as un-scientific a test as it admittedly is, the results above should be considered only as a case by case, user by user demonstration. Not as a real world test or the technology. Very subjective.
unfortunately I can believe how bad the Droid is. When testing a demo unit, I found the sensor was so sensitive hat it responded to touches outside of th normal screen area. I had positive touch feedback when touching the outside bezel of the screen. It was that fact that was one or my reasons against deciding on q Droid.
These results are entertaining, and should be taken with a big spoon of salt. By no means should you use this to decide your next device.
Just my rather elaborate 2c.
January 11th, 2010 at 1:40 pm
Doesn’t the fact that the CEO of Moto is a former Apple employee constitute a conflict of interest?
January 11th, 2010 at 2:48 pm
The conclusion regarding iPhone non-linearities at the edge of the screen is flawed, I think. I tried the test on my iPod Touch and was able to duplicate the results shown here if I used the tip of my finger. When I retried the test using the tip of my fingernail (a much smaller diameter contact patch), I got straight lines all the way to the edge.
This makes sense if you consider that as your finger tip moves off screen, the center of the remaining contact patch moves further along the edge, while it stays at the edge of the screen — the touch screen hardware does NOT track the center of your fingertip off screen. If anything, this indicates that the iPhone/iPod Touch is more sensitive and can work with a lighter touch.
I say re-test the iPhone using a more pointed stylus or a fingernail in order to see the true edge linearity.
January 11th, 2010 at 4:43 pm
Hmmm. Perhaps the tester needs a larger sample size. I get dead straight lines on my Droid. I wonder if he had the charger plugged in while doing the test? The touch screen on some Droids can get a little squirrelly when the AC charger is plugged in.
January 11th, 2010 at 5:50 pm
The change in the artefacts when you touch more lightly isn’t (mostly) because of “a weak signal”, it’s because with a light touch the region of contact between your finger and the screen is small, which means your finger couples to fewer of the sensing elements, which means that whatever position-calculating algorithm the touch sensor uses has less data to work with. (Extreme case: suppose the contact area is so small that the finger is “seen” by only *one* of the sensing elements; then all the system can possibly tell is “your finger is somewhere around here”. In practice, that won’t happen unless you touch so lightly that your finger is considered not touching. Perhaps not even then.)
Different users may get different results because the size of your finger matters. (How much it matters depends on the size and shape of the sensing elements, and exactly what algorithms are used to calculate its position.)
Looking at those pictures, I’d say that the three Android devices are all using Synaptics-style interlocking diamonds (I don’t know whether they’re actually Synaptics sensors) with roughly the same pitch (= distance between sensing elements), and that the Droid maybe has thinner glass (or plastic or whatever) in front of the sensor than the other two. The nasty jagged bits on the Droid make me suspect a different vendor from the other two, too. The difference at the edges between the Eris and the Nexus 1 is interesting; I suspect that the Nexus 1’s sensor has a bit more space around the edges of the screen. I’m pretty sure the iPhone uses an entirely different design of sensor — Apple bought a capacitive sensing company back before there even was an iPhone and I think they’ve done their sensors in-house since then. It’s hard to tell much about the sensor when the evidence basically says that aside from the edge effects it Just Works.
[(Dis)claimer: I used to work for Synaptics. I know quite a lot about capacitive sensors. I have absolutely no inside information about any of these devices.]
January 11th, 2010 at 6:56 pm
@ Petri – “That’s pretty interesting. I’ve personally never used a capacitive touch screen, so I’m left wondering how would it compare to resistive one? Such as Nokia N900.”
A modern resistive touchscreen will in most cases be accurate. The only problem is you can’t do the light finger test because a resistive requires some pressure unlike capacitive touchscreens which can sense light pressure. No point in comparing two different technologies. If this test is correct then the iPhone’s result will be the closest to a resistive screen performance in say an N900, if it’s same accurate as the N97.
January 11th, 2010 at 7:49 pm
Hi all.
I made some testing with my motorola milestone and the results are much more better than your results with the iphone. I use an application called Simple Draw.
To the readers of this page. I beg you to try to do it with your phone because maybe the purpose of this news is just not show the truth.
Take care.
T.
January 12th, 2010 at 8:05 am
What I think accounts for the curvature at the edge of the screen is the fact that the person testing is useing a wide area of his finger tip. The phone detects the touch and uses the center of the touch, which would be the center of the finger tip. However, when part of his finger goes off the edge of the screen, it makes the center of the touch move inward, resulting in the curvature at the edge of the screen. Is suspect that the curvature at the edge of the screen would be approx. the width of the area of the fingertip used for the touch ( the curvature being measured diagnally.). There would be considerable less curving if a very small area of the finger tip was used.
January 12th, 2010 at 8:32 am
I am trying to be unbiased here (droid owner), but I think maybe the test is flawed…or perhaps the conclusions are flawed. I do get similar results on my Droid….I don’t debate that. However, the results may indicate a number of things….the accuracy of the touch sensors, the spacing/placement of the sensors, or as mentioned they may indicate the algorithms used and how the data is interpreted/displayed (error correction in play?).
I think a series of more tests would help clarify.
1. Do the same tests using a straight edge and something like a stylus that doesn’t vary its point of contact like a finger does. See if the results are the same.
2. Do the same tests with a stylus and intentionally make slightly squiggly lines. If the lines are crooked that would indicate an accurate reflection of the touch sensors. If the lines are straight it would indicate whether there is any sort of error correction in play.
3. Design some sort of targeting test. Put a series of very small circles on the screen and using a stylus, touch the center of the target and see if the screen registers the placement of your mark accurately.
What do you all think?
January 12th, 2010 at 9:22 am
I just did this on my HTC Eris and it doesn’t look anything like his test. It looks straight and smoother than the iphone in this test in fact. i think he used less pressure or something. If you’re hardly touching it, you get the waviness. In fact, i can only get it to look as wavy as his ‘medium pressure – eris’ if I’m hardly touching it at all. Having said that, I can’t draw a straight line with a pencil on paper if I’m not using any pressure.
January 12th, 2010 at 9:42 am
Funny – I get perfectly straight lines with no stairstepping at all on my HTC Pure
Thought it might be because it’s resistive but I also tested it on my multitouch tablet that has a capacitive screen and all lines also come out perfectly straight
Maybe it has to do with the sampling speed of the phones in question
January 12th, 2010 at 10:23 am
Why doesn’t the author just POST the apps used to test this? I am VERY sure the rest of the community wants to test this on their own, especially, since so many sites are running this article without independent confirmation.
January 12th, 2010 at 11:10 am
This is just an iPhone fanboy test. I just did this on my HTC touch pro 2 (Windows mobile with resititive screen, using my finger, not the stylus) and got a perfect set of lines, and that’s WITH a screen protector. I work in construction and use it all the time for diagrams etc.. I use Phatpad but the results using the built in Notes app weren’t much worse.
AWFUL TEST !
January 12th, 2010 at 2:22 pm
This would be an awesome and thorough test, except for the fact that the premise is completely and utterly flawed and bogus. This in NO way tests the capabilities of the hardware touchscreen, all this is testing is the sensitivity and precision of whatever cobbled together drawing program you happen to be using on whatever platform.
Im not saying the Droid isnt the worst, i am saying that this is not a good way to test this.
Try again.
January 12th, 2010 at 7:23 pm
Why no Palm WebOS devices in this comparison ?
I know their hardware looks weak but this is not a true comparison without’em
January 12th, 2010 at 11:51 pm
I have to say this is a very nice and original way of testing a touch device!
However, why not compare the iPhone to Palm Pre and to HTC Hero, which are definitely of the best smart-phones around?
From this article it may seem like the iPhone has the most accurate touch, which may actually be NOT true since the comparison is lacking its primary competition.
January 13th, 2010 at 7:43 am
@Andre Richards & Darrell
check out http://www.littlesunsoftware.com/ doodledroid, it’s the software I use to draw on my droid and its pretty capable of producing good results, the site has some examples of work people have produced with it.
January 13th, 2010 at 3:21 pm
To be fair Moto develop a platform based on Android so I don’t see why they would deliberately want to make the nexus look bad.
Though I think the test should have used a ruler as a finger guide to eliminate the posibility of movement left and right which could potentially cause the less than straight lines. Also could that indicated that actually the capture resolution is higher, and the iphone is disregarding this extra information and making the line straigther?
Wondering why my previous comment seems to have been deleted, maybe because I included a link to an android app?
January 13th, 2010 at 11:41 pm
As a scientist who regularly conducts experiments I can tell you that this one would be laughed out of the lab. If in fact you moto.com guys are actual lab testers this would be obvious to you which leads me to wonder if there is some ulterior motive.
January 14th, 2010 at 8:25 am
I just downloaded Draw! on my Droid, and it passed all the straightline tests, too, not a wriggle among them.
My conclusion? It’s the drawing apps, not the screen.
I will confirm some difficulty w/ hitting small hyperlinks, but perhaps that’s the Droid’s browser too. Will have to try another like Dolphin before jumping to another incorrect conclusion.
January 14th, 2010 at 9:10 am
Wavyness isn’t a problem in case of Motorola Droid as the guy claims. It has a hardware keyboard so I doubt someone would use screen keyboard at all
January 14th, 2010 at 5:59 pm
It would also be interesting to see how Palm Pre’s touchscreen compares.
January 14th, 2010 at 11:55 pm
Even though I think my Motorola Droid is superior to my iPhone (in doing what I want it to do), it is not even disputable that the iPhone has a better touch screen interface. In particular, it seems to make up for the fact that your finger actually touches the screen just below what you actually wanted to point at. On the Android phones, you have to touch *slightly* above links in the browser so that the pad of your finger touches the correct link. That being said, my Droid is very customizable, has a USB connector, and has a removable SD card…all features which are extremely important to me.
January 17th, 2010 at 11:01 am
Hi Everyone,
Morgan here from MOTO. Yep, I’m the guy in the video. It’s amazing to see such an active discussion around this simple set of images. Thanks for blowing this up!
An especially big thanks to those of you who tried it and posted results here!
A lot of folks have raised a lot of interesting points here which merit further discussion. I want to address a few key things:
First of all, we don’t intend this to be a rating of the phones as products. We are looking at touchscreen performance here. I could lecture for hours about the nerdly specifics, but the truth is, we’re just trying to share a DIY approach which consumers can use to get a quick understanding of the performance of devices.
It’s not meant to be journal-quality scientific authorship.
Second, just for the record, I am not an Apple Fanboy. I am greatful to Apple for marketing this stuff so heavily, because it drives competition focused on integrated user experience rather than raw features. As an alum of the glory days of IDEO, I’m into that kind of thing. I use Ubuntu at home, XP at work, have an iPhone, many Zunes, many iPods, Sansas, iRivers, HTC phones, Samsung phones, MIDs, tablets, etc. You name it, I’ve used it, torn it apart, criticized it in frustration, and done my best to understand WHY the designers and engineers who built it made the decisions they did.
I have a Nexus One at the moment, and I’m stoked about it — there are a lot of rough spots, but overall the flexbility and power of Android is really starting to shine for me.
I would like nothing more than to see really high quality, ultra-linear touch solutions from every vendor. In my day-to-day work as a product development consultant, that’s good for everyone.
Here are some good points many of you have made, with my comments to help clarify:
1) “You drew crappy lines!!1!”
Yes, yes I did. If you do it by hand, it doesn’t look pretty. But I promise you, I don’t have a tremor which leads to a perfect 10mm periodicity in the waviness of the motion of my fingers.
The clearly-periodic waviness you see is from the sensors, not my finger.
Maybe next time we’ll show you how we do it for *real*, with the robot sitting behind me in the video, but the idea in this post was to show something you can do on the floor at Best Buy if you’re so inclined.
Try it yourself!
2) “You drew faster on the iPhone!”
In the video, I did. Sorry. But in the images shown in the photos, I didn’t, I swear. I drew so damn slowly in those it was aggravating.
Even if I drew 30% faster in the video, the iPhone sensor samples at a rate about 30% higher than the other devices tested, so it’s still not a horrible error. The linearity demonstration is legit.
Try it yourself!
3) “Software can fix that bad touch data!”
Not necessarily. Depends on why/how it’s bad.
If you have some grounding in remote sensing or signal processing, you know that there are certain limits to how much information you have to work with from a transducer of any sort.
Good algorithms can compensate for weird non-linearities (like, say, wavy lines!) in some situations, but you CANNOT create signal where there is none.
One reason Apple’s touch sensor is so sensitive to light touch is that they use 12V drive versus 3-5V for most solutions on the market right now. It costs more power, but you end up with considerably higher voltages to sense, which means higher fundamental SNR coming out of the sensor. This is a *huge* advantage, and it’s something that any touch vendor COULD do if they wanted to.
4) “Medium touch on Android works as well as light touch on iPhone” (@recharged95)
Yes. This is true. However, really light touch is useful for making flicking around menus and maps feel more fluid — if you have to push against the screen, you have to deal with more friction.
You can certainly design your UI around the flaws in your sensing apparatus, like, say, bad edge performance, or bad linearity.
You increase the target sizes, put stuff in the center of the screen, etc.
But you’re paying for that by sacrificing creative freedom.
Every touchscreen should try to be perfectly linear everywhere, so people can build whatever UI they want.
The experience of using a smartphone is the fusion of a massive number of people’s design and engineering expertise — typically a team of 50-200 people or more, when you start counting the design teams at the big ODMs like FoxConn, Quanta, Flextronics, etc.
These products are supposed to be integrated experiences that balance every aspect, from the lightness of touch you can use, to the smoothness of onscreen animation, to bells and whistles like Google Maps and compasses and gyros and accelerometers, too.
Don’t sell yourself short! Demand the best performance from the gadgets you buy in every aspect!
5) “Keyboard performance is hurt more by edge issues than overall linearity issues” (@recharged95) Not enough information here. Honestly, I’ve sat in the user test labs and run these tests. I’ve seen the error rates which real people have on different touch keyboards, and if the data were not proprietary to my clients, I could share examples demonstrating that overall linearity problems with light touch CAN and DO significantly impact typing performance.
A few mm of positional error is a big deal for your key error rate when you’re typing 1000’s of characters.
6) “The lines were fat on the Droid in the video”
Yes they were. Sorry about that. For whatever reason, I couldn’t get the Droid to download the simple drawing app I was using (“DrawNoteK”) until I got to CES and we’d already shot the video with another app (“Draw! Free”).
On the iPhone we used “SimpleDraw”.
The hand-drawing I did on the photos was much more consistent than in the video. It’s still hand-drawing, of course, so verify for yourself if you’re skeptical.
I just got “Simply Draw” on the Nexus One, which I like even better, as it has thinner lines. We’ll probably use that in the future, as it supports multitouch input.
7) “You used different software!”
Yeah, we didn’t have time to fix that in the video. But all the *photos* were done with DrawNoteK on the android devices.
Honestly, I have a pretty refined eye for this stuff, as I’ve written all kinds of SW for data visualization, and I don’t think DrawNoteK or SimpleDraw is doing any major smoothing.
CERTAINLY not to an extent which would hide the waviness we observed.
That waviness is characteristic of the mechanical construction of the touch sensors, which have diamond patterns at a 5mm pitch – you’re seeing the signal waving between diamond rows due to low SNR and imperfect positioning algorithms.
I wanted to include it — there’s one by my desk, and I know the team that designed it. But I couldn’t for the life of me find the login credentials for that particular device.
I’ll probably test it next week just for kicks. I might post the result on our labs.moto.com website.
9) “My Pre/Hero/Whatever performed differently and I can prove it!”
Of COURSE it did. You used your finger, your hand, your muscles, your electromagnetic environment, and your device.
The range of finger contact patch size for normal people ranges from about 5mm to about 15mm diameter for those with really big finger pads. It’s wildly organic. Some people drag lightly, some people stab at buttons, some people scrub firmly across the surface. The amount of oil on your fingers varies widely, too — this means different lubrication properties while sliding.
That’s why this is a DIY test!
You really may get great results with a device that I get mediocre ones on!
Try it yourself!
10) “Analog resistive touch screens kick capacitive screen’s ass for linearity”
Hell yes they sure do. And they work great with stylus. The only problems are:
* they don’t do multifinger (do you really care? maybe not…)
* they require force to actuate, so they don’t work as well for light touch flicking-type interfaces
* they have worse optical properties than capacitive screens, because they generally include air gaps
Stantum, TouchCo and others have some interesting stuff to show on the multitouch resistive front, but they don’t fix the no-pressure-required problem. Still, I’m excited to see scanned digital and hybrid systems come to market. It’s gonna be super cool, and will offer some cool differences versus capacitive, especially in the glove/stylus space.
11) “No way! You’re just trying to make the iPhone look better!”
Really, honestly, no. I’m a Freedom and competition guy.
I have played with touch dev kits from *every* other manufacturer, including plenty of stuff that’s not on the market yet, and these wavy artifacts are always a challenge. Apple has been spending a lot of money paying a lot of smart people working on this for a long time.
There’s good news, though: most of the silicon driving these touchscreens is a generation out of date already. The HTC phones only do “1.5 touch” right now — fine for pan/zoom, but not for good rotation or “Ocarina”-style true multifinger input.
Just about everyone will have stuff shipping with true multitouch and much more refined algorithms and sensor designs Real Soon Now, so it’s gonna get better fast.
It’s going to be a great year.
Thanks for listening, and for keeping the pressure on the manufacturers in the blogosphere!
February 1st, 2010 at 1:54 pm
I just tested this out on my Palm Pre with the new free drawing app that just came out (Paintr Lite) and assuming I did things the way they did in the test the results were very good. Comparable to the iphone: straight lines but some minor bending near the edges
February 11th, 2010 at 3:41 pm
I just tested the Palm Pre Plus using the Paintr Lite program as well (using different paintbrush styles) and I thought it performed just as well as the iPhone, if not better!
February 15th, 2010 at 2:41 pm
Morgan,
Palm pre please?
March 5th, 2010 at 12:15 am
this test is useless. They could have used a guy with parkinsons to draw the lines on the non apple devices for all we know.
March 12th, 2010 at 6:47 am
It would also be interesting to see how Palm Pre's rouchscrsen comparws.;
March 26th, 2010 at 2:44 am
Wait Samsung Galaxy S, the new screen SUPER AMOLED.. It will beat iPhone..!
March 27th, 2010 at 3:30 am
That’s completely bogus!! I just did the same drawing test on my Motorola Droid and i have straight lines!! I used Draw(er), free from the market. Completely not convinced at all!
March 28th, 2010 at 6:17 pm
Any tools for Symbian to test it?