Introduction

On October 17th, 2024, Jason McDowell (The AR Show), Jeri Ellsworth (Tilt Five), David Bonelli (Pulsar), Bradley Lynch (SadlyItsBradley), and I recorded a 2-hour roundtable discussion about the recent announcements of the Snap Spectacles 5 and Meta Orion optical AR/MR glasses. Along the way, we discussed various related subjects, including some about the Apple Vision Pro.

I’m breaking the video into several parts to keep some discussions from being buried in a single long video. In part 2, we mostly discuss Meta Orion’s AR/MR headset.

In this article, I will try and fill in some details from our on-the-fly discussions.

Part 2: Round Table Discussion about Meta Orion Technology and Applications

Below is a link to Part 2 of the video. In the subsequent sections, I will add some details and information about each section (with links in light blue in each sub-section header).

0:00 Panelist Introduction (repeated from the first part)

The first ~2 minutes is simply a repeat of the introduction of the panelist.

2:18 Meta Orion made Clear the “Real Prize” is See-Through Mixed Reality

The consensus is that Meta featured Orion at Meta Connect to make a statement to investors (and Apple with their Vision Pro) that the real prize in mixed reality is going to be with (optical) see-through and not (camera) pass-through. Several of the more broad tech media commented on how much better it is to see the real world than using camera-passthrough to a display, no matter how good that camera and display may be (e.g., Apple Vision Pro).

2:57 Orion: High Index Waveguides May Cause Distortion

David noted that people’s eyes seem to be a bit larger and move faster than is natural. He thinks this is due to the use of Silicon Carbide (SiC) waveguides, which have a very high index of refraction. It is like looking through a thicker piece of glass, looking in and looking out.

4:02 Meta Spent ~$5B Just On Orion

Jason read that Meta Spent ~$5B just on Orion. Meta is reportedly losing about $1B/month in mixed reality. The panelists discuss some of the factors that would cause such spending.

Dividing $5B by 1,000 units (the number Meta says they made) means that Meta spent $5 million/unit on R&D and manufacturing setup. So, when Meta says they would cost $10,000 to make, that is likely what Meta thinks the incremental cost would be (I’m not sure at what volume).

4:47 Silicon Carbide (SiC) Waveguides

In this section, we discuss the economic issues with SiC waveguides in more detail. A big issue is that SiC is a grown crystal on ingots that are then sliced into round wafers like semiconductors. Glass, on the other hand, is amorphous and can be heated and poured into a mold to make a thick block of glass that is then sliced.

Currently, Meta is using 300mm (~4-inch) SiC wafers, whereas 300m (~12-inch) is a common glass wafer size (to be compatible with optical equipment. When you try to arrange devices (waveguides) on a small round wafer, there is much more unusable area (“waste”) as a percentage of the total area.

I mistakenly commented that my first I.C. was made on a 4-inch semiconductor wafer. In fact, my first design, the TMS9918 (the “Sprite Chip” used in TI Home Computer, MSX computer, and Colecovision, whose sprite functionality was copied by Nintendo and Sega), was produced in 1978 on only a 3-inch wafer.

We discussed the difficulty of making waveguides with SiC. SiC is a very tough material often used to coat drill bits and grind other materials. This means that cutting and polishing SiC will be tougher and much more expensive to work than glass which is used in optics in mass production. .

Jeri joked that they should have gone all the way to Diamond. But as it turns out, SiC’s index of refraction is about 2.7, whereas diamond’s is lower at 2.42. “Normal” glass has an index of refraction of about 1.5, and “high index glass” can go up to about 2.0. Typically, the higher-index glass materials take on a yellow tint. SiC takes on a slight blue or green tint.

David commented major silicon carbide manufacturing on a large scale would likely require major government investment.

10:04 MicroLEDs and Waveguide Color Combining

Currently, the only available MicroLED displays are single-color (monochrome); thus, three displays have to be optically combined to form full-color. I discussed the issues with getting full color in MicroLEDs with Waveguides (CES & AR/VR/MR 2023 Pt. 7). Additionally, in 2024, Jade Bird Display (JBD) sampled a stacked MicroLED, with the various color LEDs coaxially stacked on three layers. There are pros and cons to the various color-combining methods, but based on history, a monolithic approach with all the LEDs on a single device is likely to prevail in the long run.

So far, companies developing products with JBD’s monochrome LEDs have been using an X-Cube. The alternative is to combine the colors in the waveguide. Magic Leap 1 and 2, while using a single LCOS device, illuminate the display with LEDs spatially separated, causing the light from each color to enter spatially separate waveguides and thus combine colors inside the waveguide.

With Orion, they have three projectors, one for each color, project into three entrance gratings in the waveguide.

13:14 Orion Three Device System

We started discussing the high-level system architecture of using the system with the glasses, dongle/compute pack, and wrist controller.

It was good that the glasses were wirelessly connected, but even so, they are beefy and have a fairly short battery life. While Meta CTO Bosworth has said they will shrink the devices, he also said on Adam Savage’s Tested that if you did the “Steve Job’s dunk-it-in water test,” there would be no bubbles, implying that everything is already tightly packed.

Note: We diverted to discussing MicroLEDs some more but will get back to the subject of the system.

14:41 MicroLED Display and Manufacturing Issues

We then got into discussing the MicroLEDs. Several have speculated that JBD might make the MicroLEDs as they are the only company known to be shipping MicroLEDs in any volume. However, Meta’s CTO Bosworth has stated that they were “designed by” Meta, and based on public FOV and angular resolution claims by Meta, the devices would have about 720 by 540 pixels, which does not match any current JBD device. Since Meta is only making 1,000, there are lots of potent ways they could have had the MicroLEDs made.

Jason stated that Orion was the “first product we have seen of any sort of scale” to use “full-color inorganic MicroLED displays.” Jason missed (and I missed correcting him) that TCL has been shipping their Ray Neo X2 since February 2024 and is planning on shipping their second generation Ray Neo X2 lite sometime late this year. Both Ray Neo devices use JBD’s red, green, and blue MicroLED displays with a color cube combiner (rather than combining the colors in the waveguide. I got a chance to try and take some pictures through the optics of TCL’s RayNeo X2 and Ray Neo X2 Lite at CES 2024 (below right).

15:17 Multiple MicroLED Displays/Eye Alignment Issues

Aligning three separate MicroLED displays to generate a single image is a non-trivial manufacturing issue. While most people think of the problem as just aligning the displays in X and Y relative to each other, the bigger problem is that all the displays must also be aligned in all dimensions, or there will be issues with focus across the image. There are several companies (including TCL) using JBD monochrome MicroLEDs with X-cubes and have similar alignment requirements, so it is not impossible. Still, it is a manufacturing challenge with very small pixels.

15:53 A Concept Device, Not a Product

David likened Orion to a “concept car,” i.e., something that will be turned into a product in its current form. I made the point that Orion has an inconsistent set of features, such as using extremely expensive waveguides and a very low-resolution display for the size of the FOV. A point I meant to make but forgot is that Orion does not address how it will support prescription lenses.

One of the things that seems to indicate that Orion might have been a rush job to get something showed up Meta’s Bosworth comment in The WILDEST Tech I’ve Ever Tried – Meta Orion at 9:55 where he stated, “And so, this stack right here [pointing to the corner of the glasses of the clear plastic prototype] gets much thinner, actually, about half as thick. ‘Cause the protector comes in from the back at that point.” They spent a reported $5B on Orion, and they didn’t fix the direction in which light enters the waveguide.

In that same video, Bosworth said, “There’s no bubbles. Like you throw this thing in a fish tank, you’re not gonna see anything.” This implies that everything is densely packed into the glasses, so other than saving the volume of the extra optics, there may not be a major size reduction possible. (Bosworth referenced Steve Jobs Dropping an iPod prototype in an aquarium story to prove that it could be made smaller due to the air bubbles that escaped).

Another point brought up is where they go from here. If they increase resolution, for example, that will tend to make the glasses bigger with bigger displays, more data, and more power.

18:00 Low Angular Resolution with a Wide FOV – Worst of All worlds

A wide FOV with low resolution means that more of the real world has to be blocked to display a small amount of information. I repeat a point made by Thad Starner, who has been wearing AR devices continuously since 1997 (see 2019’s FOV Obsession and AWE 2024 Panel: The Current State and Future Direction of AR Glasses), on how important it is that an AR display does not block too much of the real world.

We then go into a discussion about Orion’s very low angular resolution, with different panel members commenting.

20:06 Orion Projectors and MicroLED Issues

Another thing to note is that Orion has the projectors tightly packed together with the projector optics on top of the display devices. Suppose the resolution increases and the display device gets bigger. In that case, this arrangement may not be possible/practical, and some form of more complex optical routing may be necessary.

We go on to discuss some of the ways MicroLEDs are trying to make a single device with full color. Among the approaches is stacking multiple LED layers on top of each, with each layer having different color LEDs. Beyond the process complexity, the stack approach tends to have the problem of blocking/wasting light from the lower two layers by the upper layer.

22:36 A Wide FOV with a Small Projector Optics Issues

David brought up the issue that the projection lens looks too small to support the cone angle for such a wide FOV well. Orion has to make a very small display to fill a wide FOV, and David speculates that the resolution might vary across the field of view, as it does for most VR and Passthrough MR headsets (such as the Apple Vision Pro).

24:49 Angular Resolution Issues and Blocking Content

We discuss the user interface issues with using a low angular resolution. I point out that ALL the images we have seen today that may look like they are “through the optics” are actually just “simulated” images. The actual image quality is likely much worse than that which has been shown.

38:06 End credits

Next Part – More on Orion and What is Motivating Meta

In the last part of the roundtable, we will discuss what we think is motivating Meta and other large companies working in MR. We will also discuss issues with the Meta’s electromyography (EMG) wristband and eye glow.