Slashdot Mirror
Droid Touchscreen Less Accurate Than iPhone's
gyrogeerloose writes "A test published by MOTO labs comparing the accuracy and sensitivity of smartphone touchscreens among various makers gave the iPhone top marks ahead of HTC's Droid Eris, the Google-branded Nexus One and the Motorola Droid. The test was conducted within a drawing program using a finger to trace straight diagonal lines across the screens and then comparing the results. While it's not likely that a smart phone user is going to draw a lot of lines, the test does give some indication of which phones are most likely to properly respond to clicking on a link in a Web browser."
They are testing a first generation iPhone. This test is interesting, but not really useful. As some of the comments point out, diagonal lines really aren't the best indicator for accuracy when hitting links or whatever. As usual, the lack of consistency that comes from using a single human being comes into play. While you don't need a machine that always draws perfectly straight lines, you need a machine (or guide) that draws the same lines for each phone.
Some extra detail from the story. The iPhone has poor detection along the edges (basically flattens out diagonals into vertical or horizontal lines), the Nexus One has the best. Not that important as most UI elements aren't right at the edge anyways. The waviness in some of the tests suggests that the sensors or algorithms may be biased into vertical/horizontal motions (makes sense from a gesturing point of view).
If they really wanted to test how well the touchscreen reacts to hitting links and stuff, I don't see why they just don't go test that. Load up the same sites and keep track of how well it reacts to you hitting links. At the very least, if they wanted to do the drawing program test, it would make more sense to test what happens when you try to hit points, instead of drawing lines. So you could put some magic marker dots on the screen, and have the user hit them and look for the overlap or something.
All in all... shows off some interesting stuff. Suggests some interesting things about the behavior of the different touchscreens, but really not all too conclusive, and really points to further testing/refinement of procedure.
I have a Droid and I just tried drawing diagonals in a paint program on the phone. Yes, I did get the waviness. All that means, though, is that the Droid is not a good choice for a phone if you want to draw on it. I am still able to use the on-screen keyboard just fine and even in a web browser I never have problems tapping a link no matter how far I am zoomed out. This is definitely not a deal-breaker for me. That said, the only reason why I have a Droid is because of the physical keyboard and a pretty decent free ssh client. The kids draw on it but they couldn't care less how straight the lines are or not.
As a Droid user, I think I can answer this.
In order to unlock the screen, you can use a gesture to unlock it. About 75% of the time, it works fine but the remainder of the time the gesture is not recorded correctly. There's a few games (word search) that often have issues marking an entire word.
Only owning an iPod Touch, it's hard for me to do a side-by-side comparison since I don't do the same things with the droid as I do the touch. All that aside, I love the Droid.
No effect. First, the touchscreen resolution differs from the display resolution. Second, these tests are showing results at the image level, from a distance away. We're not talking about, say, the Droid displaying wiggles on the order of 1-2 pixels. The wiggles subsume a large number of pixels. Further, even though the iPhone has a lower-resolution screen, it makes excellent use of antialiasing. You can observe position changes that, on average, are less than the pixel pitch with such methods.
It reflects resolution, and shows 'dead spots' in the touch surface that the OS/Software must 'guess' as to approximate location. Granted it's not a very scientific test, but it does show some interesting weaknesses in the varous implementations. For instance, on an iPhone, you can click on a link that is only a few pixels in height and be relatively sure you'll get the correct link out of a list of links.
If you'll think back to the days of low resolution, when you were trying to fit a decent image into a 16x16 icon representation, you get an idea of what this may be showing. If the touch capacitance screen doesn't have a grid address for a specific spot you're trying to touch, it will have to guess between the two nearest points.
This is very similar to mouse resolution.
A more valid test would be to use a 'robotic' finger that could apply exact pressure across all phones, but it does give a decent general idea as to how they stack up.
...The Droid is not as good as the iPhone if you are buying it to use as a graphics tablet, but c'mon, who does that?...
This guy. (Left 15 images... one of them was actually a New Yorker cover.)
"There are no facts, only interpretations." --Friedrich Nietzsche.
Then you haven't used the iPhone. I click on single character links regularlly. Normally page numbers or the next arrow for images. It is another ui design by apple. It goes to the closet link within so many pixels.
The droid I got to play with one day had hard enough time registering the unlock slide let alone clicking on page numbers. I will say however it displayed ars techincas mobile sitebetter than my iPhone. The droid didn't fix the damn font height but left it scalable. Along with not having to jailbreak to install myown apps are two big pluses. But at the end of the day the UI is more important than any other app. Good apps can't fix a bad ui. However a decent UI can make limited apps more usable.
i thought once I was found, but it was only a dream.
[disclamer: i'm working in touch-screen business but not for apple/cellphone company]
This test is biased as:
- user perfomed, we use robot for this kind of qualification (but you can still get an overview if you use jigs)
- strait lines are not the best to see if some king of trajectory filtering is done by the OS: use curve lines, or corners (to see over/undershoot)
- to check if the border effect will affect the point perfomance, touch the screen at regularly spaced points (use a transparent plastic with dots printed on it)
- it would be interesting to get the raw data sent from the touch sensor to check sampling rate & multi touch tracking (and thus removing, the OS and software filtering)
That said, when you are in front of a new touch sensor, the strait lines test on the border is a 'universal' benchmark performed by everyone in the field...