Demoing Our Displays

Time 5 minute read Calendar 2022-03-24 Person George Singer Pricetags #update and #timeline

This week, we discuss the Simula One's display system, outlining specs in more detail and ending with a video demonstration.

1 Display Specs

SpecValueExplanation
Resolution2448(H) x 2448(V)Number of pixels (measured width by height); often misleading in VR.
1 Pixel = R+G+B dots3 dots per pixel
Screen size51.78mm x 51.78mm (2.88" diagonal)Dimensions of each display
Pixel configurationR,G,B Vertical StripesPattern in which pixels are laid out on screen (see below)
Pixel pitch21.15umDistance between pixels (see below)
Number of colors16,777,216 Colors1 byte per color dot => 28 = 256 colors per dot => 2563 = 16,777,216 colors per pixel
Weight10.7 g2 displays => 21.4 g of weight.
Contrast ratio650:1Ratio of the luminance (or brightness) of all white pixels to the luminance (or brightness) of all black pixels
Refresh rate90Hz - 120HzHow many times per second a display is able to draw a new image.
Response time (BtW)5msHow long it takes for a pixel to go from black to white.
Typical brightness150 nitsBrightness of the display under normal settings
NTSC Color Gamut83%Percentage of NTSC color gamut capable of being displayed

If you're new to displays, some of the terms mentioned above (e.g. "dots", "pixel configuration", "pixel pitch") might need clarification:

  1. Pixels vs. dots. Dots (also known as "subpixels") are single-color subregions of a pixel. Depending upon the pixel configuration of the display, dots can show up as small lines or small circles assembled into various patterns (e.g. vertically aligned lines, staggered circles, etc).

  1. Pixel configuration. Pixel configuration (or "pixel geometry") refers to how the dots in a display are patterned.

  1. Pixel pitch. Pixel pitch measures the distance between pixels (or between dots of the same color) on a display.

  1. Computing the number of colors. The number of colors displayable is determined by the bytes allocated per pixel. In our case each pixel has 3 dots (R+G+B), and each dot has 8 bits (or 1 byte). Hence each dot can encode 2^8 = 256 distinct color values. Since there are 3 dots per pixel, each pixel is then capable of displaying up to 256^3 = 16,777,216 distinct colors.

  2. Response time (BtW). Response time measures how long it takes for a pixel to change colors (in this case: from black to white). This serves as a proxy for how long it will take for a display to draw a new image after it receives a new draw. Low response times are crucial for responsive VR, while high response times can cause disorientation or motion sickness.

  3. NTSC Color Gamut. The NTSC color gamut denotes a subset of the visible color spectrum which displays can benchmark against. In our case: our displays are capable of showing colors from 83% of the NTSC gamut.

2 Integrating the displays into our headset

The process of connecting a Desktop PC to a display is pretty straightforward: simply connect them together with an HDMI or DisplayPort cable. However, panels are rarely driven directly by DisplayPort.

  1. Each of our two VR Displays expects to be connected via the MIPI-DSI Serial Interface (a pretty common standard for mobile & VR displays).
  2. The CPU in our compute pack only provides a single MIPI-DSI port, but we need two!
  3. Desktop PCs do not support MIPI outputs, so you wouldn't be able to use the displays in tethered mode

Our solution to (1)-(3) is to split up a single outgoing stream of DisplayPort data into two MIPI-DSI lanes, via a VXR7200 Dual MIPI VR Bridge. With two MIPI-DSI streams, we are then able to connect them to each of our VR displays.

Here's what the VXR7200 looks like (using an evaluation board):

The overall connection setup can be visualized roughly as follows:

Note that the D-PHY interface connecting the bridge to each of the displays is a type of a MIPI interface.

3 Testing the displays

Putting this all together: below is a video of our compute unit routing video data through the VXR7200 bridge to our displays:

Note that the images in our VR displays still need to be distorted for them to look correct when viewed through our lenses (details discussed here). This distortion will be handled through a library called monado, which is an open-source implementation of an OpenXR runtime. We will explain these issues in more detail in a future post.