Up close and personal

Posted by Jon Peddie on March 10th 2017 | Discuss

I’m seeing more than I want to, or maybe not enough

Our eyes are just too damn good. Too good for the VR and AR displays we can afford and are able to build today. 

VR as you probably know has a taxonomy that can be reduced to a quadrant: 

Samsung’s gear is upper left, HTC and Oculus are lower left, and the stuff that started the VR field, and where the money is made is the lower right. 

The lower left is highly over hyped by the participants, investors, market researchers (who should know better), the press and pundits. They are all practicng the Emperor’s New Clothes, seeing things they want to see that aren’t there—so they tell each other how great it all is, and how wonderful it will be. They will get tired of the charade and turn their attention to the next shiny new thing, and likely turn on each other in the process. 

Consumer VR, especially the interactive, PC/PS4-based type has two major problems, Latency and something called the screen-door effect. 

The screen-door effect (SDE) or fixed-pattern noise (FPN) is a visual artifact of displays, where the fine lines separating pixels (or subpixels) become visible in the displayed image. It is sometimes incorrectly referred to as pixel-pitch. In the simplest of explanations, pixel pitch is the distance from the center of an LED (LCD, or OLED) cluster (or pixel) to the center of the next LED cluster/pixel measured in millimeters. “Screen door effect” is a phrase used to describe a display that has a visible gap between individual pixels. However, there is no technical specification (that I know of) for this gap. 

The screen door effect is the result of several components. PPI (pixel per inch) is a huge contributor. Another element that is a big contributor to perceived resolution, is the viewing distance to the actual display. The math is quite straight forward: the diameter of the pixel, times the viewing distance and (in HMD’s) the power of the optic system, this provides the “system” resolution, which is what you are really looking at in HMD’s. 

The HTC Vive, has two 91.9-mm (dia) 1080 × 1200 OLED displays giving 446 ppi per eye, while the and Oculus Rift is slightly smaller (90-mm) screen with the same resolution giving 456 ppi per eye. The PS4 has a 220-mm screen with 960 × 1080 resolution providing 254 PPI across both eyes. And a Samsung Galaxy S7 used in Gear is a 217-mm screen with 2560 × 1440 resolution giving a 534 PPI across both eyes. 

The consumer VR HMDs place the screen between 55- and 65-mm from your eye. At that distance the visual acuity is about 1-mm, that means if you had alternating black and white lines that were all 1-mm-wide, it would appear to most people as a mass of solid gray, like the upper right corner of the diagram. 

But because our eyes are so good we are able to discern one arc minute of resolution. At that distance, we see the lines, and so we see the “screen-door” effect of the display. The trick is to either use a super high resolution display, or a very small panel display with tiny pixels and move it away from the eye so the display exceeds the acuity of the eye so we can’t discern the lines between the pixels. 

Entrepreneur and technology pioneer, Karl Guttag calculates that if you want to support a 120 degree FOV at near the eye resolving power you need about 7,200 pixels horizontally that are about 9-micron pitch. 

You can, and must, do that in an AR system, but none of the consumer VR systems do it. That’s because of cost, the super high-res screens, using microOLEDs or microLEDs can be expensive (depending upon volume), and the (high-quality) optics needed to move the display from the eye are expensive and tricky to build and mount. Consumer VR HMDs are already considered too expensive—they are above that magic threshold of $300. Adding any costs just aggravates the problem and thwarts the price-elasticity opportunities. 

Closing comments 
So consumer VR will be thwarted by latency, and acuity. Acuity is fixable, at a cost. Latency needs a new paradigm, like local dedicated image processors and ultra-tightly-coupled and super-fast positional sensors. Cost, cost, cost. That’s why VR has been stuck in the military and scientific realms for the past 30 years—they can afford it, and because of the critical nature of their missions can/will tolerate a bit of discomfort. 
Consumer VR, with the possible exception of passive 360-videos, isn’t going to happen fast. Our eyes and brains are just too damn good for it. 

Further reading. Karl Guttag has an in-depth discussion on this at his blog page.

Discuss this editorial