Issues about defining resolution are not new. Back in the CRT days, the three electron beams in a colour monitor were completely independent of the shadow mask and each other. Each played a part in creating the overall resolution, but adjustment also mattered, for example if the beams were not aligned. I seem to remember that Philips pointed out at one SID event that you could double the resolution of the beams and get apparently better resolution, although most people took the pitch of the shadow mask as the key resolution metric. I had an idea for using the detailed contrast to quote resolution, but didn’t understand how to get this kind of idea adopted back then. My idea was that a resolution should be quoted alongside the contrast between black and white lines at that resolution. A similar idea was used in the IDMS standard for display measurement that defined the resolution . (for more on this topic, see my article from several years ago (Damn Lies, ANSI Lumens and Resolutions)
Anyway, with a switch to flat panels, the pixel structures became much more clearly defined and pixel arrangement based on RGB seemed to be more straightforward in terms of how to define the resolution. However, some displays, such as the mobile OLEDs from Samsung Display, have adopted modified pixel arrangements based on work by Clairvoyante Inc and the ideas of occasional DD contributor Candice Brown Elliott with her ‘pentile matrix’ concepts. Candice clearly set out the need for the IDMS standard to be used for resolution in her article, again some five years ago. (Resolved! No More Dot Counting!)
Despite this, dot counting is still seen by many as the way to specify resolution (I see almost no resolutions quoted, ever, based on the IDMS standard in product releases and specifications). So Samsung Display had a think about the topic and has come up with an alternate way to specify resolution. Rather than the widely used pixels per inch metric, the researchers are suggesting a ‘pixel per area’ or PPA metric. The paper acknowledges the IDMS method but points out that it may not be useful where, for example, pixels might be differently aligned on alternate display lines.
(It occurs to me that this kind of research could be particularly useful as we move to microLED display structures that may have a ‘sparse matrix’ design with very bright but very small sub-pixels, rather than subpixels that are illuminated over their whole area).
The paper looks at different pixel arrangements and how they would be specified using the new metric. It also looks at the case where pixels are distributed quite unevenly. The researchers found that when pixels are distributed non-uniformly, there is a drop in perceived resolution, so it accepts that PPA alone is not enough to characterise the display resolution and a second parameter to specify the unevenness of the pixels may be needed. The group is continuing to work on this.
The simulated images here represent the input image and the three outputs (b-d) based on the arrangements a-c above. The PPA on each is the same, but the results are clearly very different.
For evenly distributed pixels, the metric appears to work well and the group points out that an advantage of the method is that you can find the PPI measurement from the PPA by simply taking the square root. (although I wondered as I wrote this article, that if that is really the case, why not simple square the PPI to get the PPA?)
The PPA metric might be useful, but the display industry has a history of using metrics that don’t really indicate the quality and value of their products. Screen size is taken as a diagonal – so a 65″ sounds twice the size of a 32″, although it is four times larger. Set makers talk about 5K displays, compared to 4K, which sounds 25% better, but there are 77% more pixels in a 5120 x 2880 display compared to a 3840 x 2160 format! (BR)
