Chasing pixels: Why pixel density isn’t as important as you might think

This opinion feature was written by Bob Myers, an engineering and tech veteran with years of experience working for companies including Qualcomm, HP, and other tech industry leaders. 

A long time ago, when I was working in the LCD monitor and TV industry, I attended a talk by a couple of engineers from a leading display-controller makers. They accused the whole room, those of us responsible for spec’ing the performance of these displays and designing them, of practicing “noogie” – or N.O.G.E., which means “nose on glass engineering.” According to them, we were focusing on improvements only noticeable when pressing your face right up against the screen. We were pushing specs that really didn’t make a difference in normal use. They were absolutely right.

Today, the mobile device industry is doing the same thing. If you look at what’s being pushed in terms of smartphone and tablet display specs, you’ll quickly see that it’s pretty much just the pixel format or “resolution,” and then maybe the specific display technology (IPS, OLED, and so forth). But are these really the only things we should be worried about, or even the most important?

Think back just seven years, to the introduction of the iPhone 4 – with what Apple called the “Retina” display. At 326 PPI, it was supposedly called that because it matched the resolution of human vision; you couldn’t possibly need anything better, because you couldn’t see the difference. While this claim was debated by some (notably Dr. Ray Soneira, the head of display-test specialists DisplayMate Technologies), even the critics agreed that this level of resolution was pretty close to all you could practically use. 300 DPI is about what you get in glossy magazine photos, and no one was complaining that those needed to be a whole lot better.

Flash forward to today. The highest resolution available in the smartphone market is the whopping 806 PPI of the Sony Xperia Z5 Premium, which packs a full 4K image (2160 x 3840 pixels) onto a 5.5” screen. There are several phones with 1440 x 2960 pixels or thereabouts on 5.5-6.0 inch screens, for upwards of 550 PPI. Even Apple, who first told us that 326 PPI should be more than enough, will be upping that to 458 PPI with the “Super Retina” display of the iPhone X.

The technical term for all of this, my friends, is “insanity.”

To be sure, you can detect some subtle differences even up to the 500 PPI level, if you’ve got excellent eyesight and you’re holding your phone less than a foot from your eyes. But just because something is possible doesn’t mean that it should be done, or that these products represent the best overall display performance.

It takes more power (both battery and graphics processing) to drive all those pixels at the needed rate. The more pixels you make in a given backplane process, the less room there is for the actual “open area” – the part where the light comes through – in each pixel. So you’re giving up brightness, backlight power consumption, or both.

The technical term for all of this, my friends, is 'insanity.'

What should we be looking for in terms of specs for real improvements in image quality, if not just chasing higher pixel counts?

Modern display technologies are immune to problems like geometric distortion, linearity, which we haven’t seen since the demise of the CRT display over a decade ago. So don’t we have basically “perfect” screens already? The answer, of course, is no.  I can list at least three things that should be much higher priority over packing in even more pixels.

Beyond PPI

The first of these is better sunlight readability, which generally means higher luminance (brightness) and real, as-delivered-to-the-viewer contrast. An emissive display – one that creates it own light – needs to make whites roughly as bright as its surroundings in order to be comfortably viewable.

Beyond just brightness (which costs power), the display needs to deliver adequate contrast in those high-ambient-light viewing conditions. OLED displays routinely claim contrast specs of 100,000:1 or even 1,000,000:1, but that’s a lot of nonsense again. Those are the numbers you’d get in a completely dark environment, comparing just the display’s own white and black levels. In actual use, contrast is pretty much always limited by the reflected ambient light, and that’s where current displays come up short. It’s a rare screen that can do much better than 50:1 or so even in a typical indoor environment, and nowhere near that under brighter lighting. We’d love to see a full-color, full-video-rate reflective display technology, but so far nothing has come to market.

The next thing we should be looking for is better color accuracy, not bigger “color gamut” numbers. These numbers measure how much of the visible color space a display can cover and have been touted by OLED and now QLED displays for some time, but they don’t denote more accurate color. Wider gamuts would be great, if there was much source material that could actually use them. The typical “wide gamut” display just makes things look unnaturally brightly colored and cartoonish.

What we should be looking for, instead, are screens that deliver the color that was intended by the content creator, within its intended color space ( sRGB or Rec. 709 at present). Color accuracy is best measured via the “delta E star” (ΔE*) spec, which shows the error of given reference; a delta-E* of 1.0 is a just-noticeable difference. Show me a display spec that guarantees a low delta-E* over some reasonable number of test colors, and we’ll have something.

Part and parcel of delivering accurate color and good overall image quality, though, is providing the right tone response— commonly known as the right“gamma.” The details of how “gamma” works are a bit too much to go into here – maybe we’ll cover that in the future. Getting it right is critical to good quality. Most of the color error in current LCDs and OLEDs is attributable to an incorrect response, or not matching the response properly across the three primaries.

Wrap-up

Let’s stop chasing pixel counts, and instead start demanding that our screens actually deliver in those areas that can really contribute to better visual performance. There’s a lot more to making a good looking screen than just seeing who can get the most pixels on a piece of glass.