Update:

A second reader with expertise in Waveguide believes (but, like my other reader, does not have direct knowledge) that the Snap Spectacles 5 is still using a waveguide with gratings on only one side. This second reader cites the shape of the “eye glow,” particularly the triangular shape from the entrance grating. This reader says that the gradations that are visible in the Snap Spectacles 5 passive light capture are likely due to making the pillar gratings deeper to try and give a more uniform output. The reader also cites the much simpler process of making grating on one side.

The first reader was not certain and speculated based on some insider rumors about the Snap Spectacles 5 (there has been considerable movement of personnel between companies that design waveguides). Neither reader claimed to have firsthand knowledge of the Spectacles 5.

Introduction

The general premise of my last article, Meta Orion AR (Pt. 2 Orion vs Wave Optics/Snap and Magic Leap Waveguides), is that Meta Orion is using similar waveguide technology to Snap (Wave Optics) and that Magic Leap Two (ML2) is correct. However, several comments and assumptions about the specifics of waveguides with a diffraction grating on both sides were incorrect.

As I often say, “Few people will volunteer information, but many will correct you if you are wrong.” As my direct knowledge of the waveguide structures was limited, I reached out to readers in the article with an email address (newsinfo@kgontech.com). One of my readers (who wishes to remain anonymous), who designs waveguides, responded. This person explained how Wave Optics 2D waveguide works and had information about the waveguide structures used in Meta Orion, Wave Optics (now Snap), Magic Leap Two, and Hololens 2 (HL2).

I want to be clear that I am far from being a waveguide expert (see my background in the Appendix), and my descriptions of the various waveguides will probably be considered, at best, loose by experts (as they say, “I know just enough to be dangerous”). I’m primarily concerned with contrasting the various approaches and reviewing the resultant images based on the many different approaches I have seen. For example, it is generally true that as FOVs increase, image quality degrades with diffractive waveguide. Also, going to higher index waveguide substrates enables wider FOV and should require fewer TIR bounces for a given FOV, which will improve image quality.

Based on my readers’ information and some further studying, I am making this update. I am going to summarize some information from the prior article so this one can stand on its own without needing to reference back continually.

More on diffraction grating

The article “Optimization of gratings in a diffractive waveguide using relative-direction-cosine diagrams” provides a more in-depth explanation and many good diagrams showing how light propagates via various waveguide designs, including the common planar three gratings (Fig 6), a more complex grating structure (Figs 7a and 7b), and a Wave Optics type with a 2D grating (Figs 7c and 7c).

In the diagrams above, you will see a series of gray rings with red circles. These diagrams show the allowable FOV. The width of the gray rings, and thus FOV, increases as the index of refraction of the waveguide substrate increases.

Greatly Simplifying

I’m not going to profess that I understand waveguides well. However, a key basic rule is that the light in a waveguide must be turned at 360 degrees. You will note all the triangles in the diagrams with some of the three angles, of course, summing to 360 degrees. With separate gratings (such as in Fig. 7 above), the angles don’t have to be equal. In the case of older Wave Optics and Snap Spectacles 4 waveguides, the pillars that form the gratings are in an equilateral triangle arrangement to impart two 60-degree turns of the light.

Meta Orion’s Waveguide with Diffraction Gratings on Both Sides

Meta Orion, in addition to using Silicon Carbide, has what, in my experience, was an unusual waveguide structure with overlapping diffraction gratings on both sides of the waveguide substrate. Most waveguides I have seen have three gratings, entrance, expansion, and exit, on only one side of the waveguide. I then remembered Wave Optics (bought by Snap) had a similar unusual waveguide structure. I then wondered how Magic Leap 2 got its wide FOV, which resulted in my discovering that they were doing something similar.

Meta’s Orion, which has overlapping grains on both sides of the waveguide, must also have light turned by three gratings to exit. In the Orion waveguide, the two large gratings act simultaneously as expansion gratings and exit gratings. The light is first turned by the entrance gratings (1020 for red, 1022 for green, and 1024 for blue). About half the light is then turned by one of the large overlapping gratings into the opposite overlapping grating, where the light is turned to exit. The other half of the light it turned by the large gratings in the opposite order.

Another feature shown in Fig 12A above is the “Disparity correction,” which is not in Orion but, according to Meta CTO Bosworth, would be in future versions. Interestingly, the Magic Leap Two (ML2) appears to have implemented this “disparity correction” as what they called “On-Line Display Calibration” (right). The same figure below from Magic Leaps SPIE AR/VR/MR 2022 presentation shows that the ML2 has a 2D expander.

Wave Optics (Snap) 2D Diffraction Gratings on One Side (Spectacles 5 Might Have Gratings on Both Sides)

My main error about the nature of waveguide “grating” structures was a bias toward linear gratings, with which I was much more familiar. As I wrote, I hadn’t studied Wave Optics waveguides before, and I misunderstood that Wave Optics was using a set of “pillar” gratings that act like a 2D set of linear gratings, even though that was shown in their patents (see below). It turns that pillars arranged in an equilateral triangle can behave like a linear grating in two directions (see green and red lines in FIG. 16 below).

As with the “conventional” three gratings on a single plane (and with gratings on both sides of the waveguide), the light must be “turned” by three gratings. In this case, the light is first turned by the entrance grating, then turned by the effective grating on one diagonal, and then turned by the effect grating on the other diagonal in order to exit. This process is better explained in the article “Optimization of gratings in a diffractive waveguide using relative-direction-cosine diagrams.”

I have not had the chance to evaluate Snap Spectacles 4 or 5 extensively. I do have Wave Optics’ Titan development kit waveguides, which I think are similar in design (although rotated 90 degrees relative to Spectacles 4). The color uniformity issues with the Titan waveguide are shown above right (click on the image for a high-resolution picture). It will be interesting to see how this has improved on the Spectacles 5.

Wave Optics used a single-layer “pillar” in their previous devices, including the Snap Spectacles 4. However, my source thinks (I’m not sure) that the new Snap Spectacles 5 might use linear gratings on both sides of their waveguide, as shown in FIGs. 3 and 15B below, making it closer to Meta’s Orion in design. So, in a way, the last article may have been somewhat correct after all 😊.

There is evidence that Snap Spectacles 5 waveguides are significantly different than Spectacles 4. While gen 5 and 4 exhibit a similar “eye glow” with a triangle from the side of the exit grating(s), the passive (device off) light capture is different between the two generations. Note that in the Gen 5 spectacles, the light capture has gradations, whereas there are no gradations in the light capture in the Gen 4 spectacles. The passive light capture on the Gen 4 is similar to what I see with the older Titan development kit waveguides, which, no matter how I illuminate them, don’t exhibit these gradations.

Magic Leap Two with Three Waveguides with Diffraction Gratings on both sides

My source is reasonably confident that the ML2 has three waveguides with diffraction gratings on both sides. Knowing this, it becomes easier to sort through the massive number of large (many over 200 pages) applications that show what ML2 might be doing.

As stated in my last article (Part 2), US 2018/0042276 was an “omnibus application” that combined many different concepts into a single application. The filing entity will later file “divisional” applications based on the original specification (text and figures) with different sets of claims for the different concepts. This particular application was 272 pages long, with over 150 figures. The patent discussed putting exit gratings on only one side or both sides (see Fig 34B).

Based on many patents and Magic Leap’s SPIE 2022 presentation, Magic Leap has spatially separate color inputs for each of the waveguides (see below).

I also found application 2020/0158942, which showed how to put a 2D grating on one side. The details showed pictures of SEM (scanning electron microscope), showing that Magic Leap had seriously developed this approach. This led me to believe that Magic Leap 2 might be using a complex diffraction grating on one side. However, as with many patents/applications, many things that are developed don’t make it into products (at least not the ML2).

Magic Leap has literally hundreds of patent applications that discuss diffractive waveguides. Many of their new application specifications are more or less appended to their old Magic Leap One applications (“Omnibus” Applications). Thus, concepts that appear on the ML2 are buried in specifications that show details of the Magic Leap One.

Additional Correction: Hololens 2 Has Diffraction Gratings on Both Sides of Each of Two Waveguides

My first mistake in the article was stating that the Hololens 2 gratings were on one side. In going back to the patent application US2017/0353871 that I used in Hololens 2 Display Evaluation (Part 4: LBS Optics), it states, “That is, the left and right intermediate DOEs [diffraction gratings] are located on opposite sides of the waveguide.” In the combined FIGs. 13 and 14 below, the right intermediate DOE, which overlaps with the left, is on the opposite side of the waveguide, as indicated by the dotted line.

Summary

1. Hololens 2 had a two-sided waveguide. The left and right expansion gratings are on opposite sides of the waveguide.
2. Prior Wave Optics (Snap) waveguides use a pillar-type 2-D diffraction grating on one side. There is a single waveguide for full color. The new Snap Spectacles 5 is possibly using linear diffraction gratings on both sides of a single waveguide full color, as shown in this article.
   • If this is true, then Snap and Meta Orion could be using very similar waveguide structures. However, Snap may be using a single entrance grating for all colors from a field sequential LCOS device. In contrast, Meta Orion has three spatially separate entrance gratings due to using three primary color MicroLED projectors with the waveguide acting to combine them
3. Magic Leap Two uses linear diffraction gratings on both sides of the waveguide. It does use three waveguides with spatially separate entrance gratings per primary color.

As stated in the previous article, I did not attempt to determine anything about inventiveness or who invented what first. That would be a massive effort and could only be settled in court. Over the years, there have been massive amounts of people moving between companies, plus some of this could be “form following functions,” causing independent invention.

Conclusion (and Comments from Others)

The above corrections indicate that Meta Orion, Snap Spectacles 5 (Wave Optics), and Magic Leap all have overlapping linear gratings on both sides. Meta Orion and Snap likely use a single waveguide for full color, whereas the Magic Leap 2 has separate waveguides for the three primary colors.

Without seeing Meta’s Orion with my own eyes and using good test patterns (and not “demo” content that can avoid problems), I can’t say anything definitive about the image quality. I am skeptical about the image quality, considering they use a single waveguide to combine all three primary colors with a wide FOV. But I am happy to be proven wrong if I can get a unit for evaluation. There are massive numbers of variables that could affect image quality, including defects in the printing/etching of the diffraction gratings and the alignment of the top and bottom layers.

There remain the social problems of the eye glow exhibited by Meta’s Orion and Snap Spectacles 4 & 5. I found it funny that some of the Orion reviewers talked about the ability to see the eyes but nothing about the eye glowing.

Brad Lynch of the SadleyItsBradley YouTube channel uses the Apple Vision Pro (AVP) almost every day and commented that the user’s eye in Orion, which are dimmed and with eye glow, look a lot like the much-maligned AVP Eyesight front display😀.

On the right is a picture I took through a Wave Optics “Titan” waveguide from circa 2020 (they have likely improved since then). Notice the color variations, which are a common problem with using a single waveguide for all three colors.

I would also like to add some observations made by David Bonelli (of Pulsar Solutions):

• He notes that with the high-index Silicon Carbide waveguides, a person’s eyes shift more than is “natural.” Humans are very sensitive to eye behavior.
• He noted that the projection optic’s exit lens is extremely small (below Right) for making a 70-degree FOV. He is concerned that there will be a lot of optical distortion that could cause Orion’s already low resolution to drop outside the center of view, as seen with the Apple Vision Pro (see: Apple Vision Pro’s (AVP) Image Quality Issues—First Impressions) and most other VR headsets.

Coming Soon: A Video Roundtable Discussion of Snap Spectacles 5 and Meta’s Orion

Last week, Jason McDowell (The AR Show), Jeri Ellsworth (CEO Tilt 5), David Bonelli (Pular Solutions), and Brad Lynch (Sadly Its Bradley YouTube Channel) recorded a round table discussion of Snap Spectacles 5 and Meta Orion. I plan on releasing the video in several parts over the next week or so.

Appendix: Summary of My Background and the Blog

My readers should note that I have BS and MS degrees in electrical engineering. I designed and architected CPUs, Graphics Processors, Image Processors, Video Chips, and memory devices (Video RAM and Synchronous DRAM) in my 20 years at Texas Instruments (TI) and became the youngest TI Fellow in the history of TI. For about 18 of my 20 years at TI, I worked on ICs for generating images and graphics. Along the way, I studied issues of visual human factors. Most of my 150 US patents are related to my work on ICs for computer graphics and image processing.

After leaving TI (having never directly worked on DLP, I should add), I worked on LCOS devices, which were essentially (digital) CMOS I.C.s with liquid crystal on them, at two startups from 1998 to 2011. As part of my work in LCOS and my interest in photography, I started to pick up some working knowledge of optics.

Since starting this blog in 2011, my “working knowledge of optics” has grown considerably as I tried to figure out the optics that went with various display devices. Through this blog and my reporting, I have seen many different types of headsets with various optics. I believe I have “broad knowledge,” having seen and studied so many optical systems to some degree, but in some areas, such as waveguides, I have a more shallow understanding. My understanding improves as I drill down to figure out what the various companies are doing.

My main goal is always to understand the contrast and compare various displays and optics and not to design optics. Thanks to my access to many headset systems and photography experience, I try to get good “through the optics” pictures to share that help with comparing the various designs. I use my knowledge of photography, optics, and display devices to take pictures that fairly represent the image quality of these headsets.

I am trying to write this blog for a person interested in technology but without a deep understanding of optics. I try to include background information and links for more information.