Introduction

For a while now, I’ve realized that there is a massive gap in the continuity of smartphone display history. Until the last few months of 2011 or early 2012, proper testing of smartphone displays was few and far between. Most websites tested peak brightness and contrast, and possibly white point. While some websites did go in depth, they would often only test a few phones. Of course, now things are different. Websites are starting to scrutinize display quality from all angles from color accuracy to reflectance, but no one has ever gone back to properly test old devices.

Of course, the best time to have tested these old devices was when they were new. The next best time is now, so let’s get to it. Unfortunately in the span of 4 years it’s become rather difficult to find all the devices that I’d like, but we’ll look at some of the key representative devices of each generation. This means the Galaxy S, Desire HD, Galaxy S2, and Rezound. The first three are all WVGA, 800x480 resolution displays. The Galaxy S has a 4 inch display, and all the other devices have a 4.3” display. All of them are also easily used with one hand, which is almost a surprise these days. Unfortunately, battery life testing won't be possible as all of the devices have aged too much for the results to be representative of their actual performance. As always, in order to properly test displays we use a custom workflow on SpectraCal's CalMAN 5.

Samsung Galaxy S

For most people, the Galaxy S would be their first encounter with Samsung’s SAMOLED displays and PenTile layouts. While it may have been acceptable at some point in the past, it’s quite obvious that this display hasn’t aged well at all. While high pixel density masks the effects of PenTile quite well, they’re omnipresent in the original Galaxy S. Even at relatively far distances from the display I can make out gaps between pixels in text. At the edges, there’s a strange effect where the lines are clearly not straight. Instead, the edges of text appear to be quite rough, making a zig zag pattern instead of a clean line.

Putting aside PenTile, the brightness of this display has aged relatively well. The expected brightness seems to be about 350 nits out of the box, and the T-Mobile unit I tested seemed to have lost some of its peak brightness, but it’s still quite bright at around 320 nits. Contrast is great as always here.

Grayscale is where things get ugly. Red is effectively no longer visible in the graph by 35% white, and both green and blue are completely out of control. The result is one of the worst averages for dE2000 error I’ve ever recorded.

In the saturation sweep, we see a similar story. Just about every type of mistake in color accuracy is seen here. Extreme blue/green shift on white, saturation compression, gamut far out of sRGB, hue shifts with saturation changes, and no way to improve it.

As expected, a similar result is seen in the Gretag Macbeth ColorChecker. This is simply a logical extension of poor color control in the saturations test, so this is no surprise. Looking at the Galaxy S5 LTE-A, it’s definitely incredible to see just how far Samsung has come since the early days of AMOLED technology.

HTC Desire HD

The next device we’ll look at is the Desire HD. LCD has always been more mature than AMOLED, so it would make sense for the improvement over the past four years to be much less. However, it seems that the Desire HD’s display is a far cry from today’s smartphone displays. While not as grating as the effects of PenTile on the Galaxy S, the low resolution is still quite obvious on this display. Careful examination shows quite a bit of aliasing, but from a distance it isn’t too obvious. The other issue here is the viewing angle stability. Much like the Xperia Z1s, the display rapidly washes out towards homogenous white with increasingly extreme viewing angles. This is likely due to the same *VA display technology used in both, although that’s about where the similarities end. There’s also no display lamination so reflectance is quite high, subjectively speaking.

As seen by the brightness graph, the Desire HD’s display is incredibly dim. I suspect this was done to try and keep battery life in check as the Desire HD was rather notorious for poor battery life. Contrast is also rather dismal, but I suspect this is partially due to the lack of dynamic contrast or similar mechanisms to artificially boost contrast values. The low brightness may also be due to the sheer age of the device.

In grayscale, the Desire HD is also rather poor. Although things aren’t quite as bad as the Galaxy S, they aren’t very good either. I suspect that the excessive green/blue in these devices was a method of boosting peak brightness, although by the Sensation it seems that HTC had clamped down white point to some extent.

In the saturation sweep, the Desire HD is eerily reminiscent of the One (M8), although this is mostly due to the rather extreme blues produced. The display exceeds sRGB by a certain amount in every sweep, and while the error here is relatively low, such error tends to magnify a great deal in the ColorChecker test.

It turns out that this is exactly what happens. Relatively acceptable error does become unacceptable in the ColorChecker. In reality this display is likely to be inaccurate for any color-sensitive use case, and color reproduction is unlikely to be close to the original image. It’s definitely better than the Galaxy S for color accuracy though.

Samsung Galaxy S2

The Galaxy S2 was and still is a great phone. In fact, it’s the only Galaxy S phone that ever shipped with an RGB strip AMOLED. Of course, this was necessary in order to have acceptable resolution for a 4.3” display without increasing resolution past WVGA. Just like the Desire HD, the resolution is still clearly quite poor on close examination, but the odd artifacting from PenTile is gone. This definitely helps with perceived sharpness and resolution. Unlike the Desire HD, the viewing angles are great. The RGB stripe also seems to result in almost no color shifting with viewing angle changes.

Unfortunately, it seems that peak luminance dropped significantly from the Galaxy S to the S2. Presumably this was due to the subpixel layout change, but even with the sunlight brightness boost feature the Galaxy S2 just isn’t as good as the Galaxy S in this regard. I recorded a maximum of 258.7 nits in movie mode and 282.2 nits in standard mode. Contrast is relatively similar and extremely high.

In standard mode, the Galaxy S2 isn't much better than the original Galaxy S. Definitely better, but barely. Far too much blue and green, far too little red, there's not much else to talk about here.

Turning on movie mode definitely helps quite a bit as evidenced by the significantly lower average error score. There's still too much blue and green, especially as we approach white, but it's far better than standard mode.

In the saturation sweep, Samsung has managed to outdo themselves in standard mode by regressing from the original Galaxy S. There's too much wrong with the calibration of this display to really pick at any particular issue in this mode.

Fortunately, movie mode tightens things up significantly, although blue is noticeably oversaturated, as is red, although to a lesser extent.

The same story can be seen in the ColorChecker for standard mode, with unacceptable inaccuracy. Many of the readings are outside of the gamut triangle entirely.

Things are much better with movie mode, but the result is still not the greatest. At this point, Samsung still has a ways to go before we get to the accurate AMOLED panels of today.

HTC Rezound

The Rezound is one of the last devices that hit the market before devices regularly saw color accuracy testing and closer evaluation of grayscale performance. Subjectively, this display is still great even today. While the Rezound’s display is supposedly S-LCD (implying *VA), the viewing angles are much better than what was seen on the Desire HD. The one issue here is that the display isn’t laminated to the glass lens, so at extreme angles the display quality isn’t as good as what we see on S-LCD2 or S-LCD3 from phones like the HTC One X. At any rate, it’s very obvious that HTC has put a large emphasis on display with this device.

While luminance is dramatically improved over the Desire HD, I suspect that brightness was clamped to some extent again to alleviate battery life issues. Contrast isn’t great here either, at 733:1. As before, there’s no dynamic contrast active here to try and artificially boost contrast. Modern LCD displays have around 1000:1 typical contrast, so this is just a bit worse, all things considered.

HTC definitely didn’t do a perfect job in grayscale once again, as there’s still too much blue and green, but white point is at a reasonable level compared to the wildly unbalanced white points that we saw with the Galaxy S2. Things are also significantly improved here relative to the Desire HD.

In the saturation sweep, HTC did an incredible job calibrating the display. It’s quite clear to me that HTC made display a priority with this device. There’s a hint of saturation compression, but overall things are very close to perfect. This is definitely a leap ahead of AMOLED at the time.

Unfortunately, the Rezound falls a bit flat in the ColorChecker. Poor grayscale calibration definitely didn’t help with the average, although there is significant error elsewhere. Of course, this is from the lens of the present. At the time, it seems that HTC was making a concerted effort to do things right, even though no one was truly paying attention to display accuracy yet and reviewers seemed to be impressed by blue white points and intensely oversaturated colors.

Final Words

Looking through this new information, understanding the past is now much easier in terms of display. While I was only able to test Samsung and HTC devices, they serve as a relatively accurate barometer for trends in the rest of the industry. As the past is an excellent predictor for the future, a few trends are evident. First, AMOLED has had far more progress than LCD, although checkered due to the pressure for higher pixel density displays and relatively little pressure (at first) to improve color accuracy. Second, while LCD has been slower to advance it has also started at a far better place. One of the chief issues seems to have been controlling gamma/grayscale color balance, although this has been steadily improving throughout the industry. Color gamut and saturation sweep accuracy has generally been acceptable in LCD, although there seems to be a bit of a cyclical relationship as consumers favor wildly inaccurate colors for showroom appeal. LCD seems to have stagnated in contrast, with relatively little change in static contrast ratios over the past few years, and brightness seems to be limited to around 600 nits in practice, barring unconventional subpixel layouts.