304 North Cardinal St.
Dorchester Center, MA 02124
304 North Cardinal St.
Dorchester Center, MA 02124
Following up on Meta’s Cambria (Part 2): Is It Half Dome 3? has led to a bit of intrigue. One of my readers pointed out that as stated in the Facebook/Meta’s Doug Lanman’s video EI 2020 Plenary: Quality Screen Time: Leveraging Computational Displays for Spatial Computing HD3 was using Pancharatnam-Berry Phase Lens (PBPL). PBPL is also known as a Berry-phase lens or generically as a Geometric Phase Lens (GPL). This blog primarily follows AR, and PBPL has not turned up AR products . . . yet.
I had not come across PBPL/GPL varifocal lenses before and mistakenly thought that LC-Fresnel variable lens shown in Meta patent application 2020/0348528 (‘528) would be considered a PBPL (it is not). The ‘528 patent application had a figure similar to that shown at Meta Connect 2021 (see right) with the lead inventor Afsoon Jamali, Lanman mentioned in the EI 2000 video. I missed that Jamali was also adding to Meta’s prior work on PBPL lenses, including patent application 2021/0231952, Freeform Varifocal Optical Assembly.
It seems I can’t get stories out fast enough. This week, I was working on this story about PBPL varifocal technology and discovered PBPL maker Imagine Optix’s website disappeared. Based on my digging around, I was able to verify that Meta is the company that bought Imagine Optix.
The Image Optix presentation from SPIE AR/VR/MR 2021 didn’t quite come out and say they were in Meta’s HD3, but they strongly hinted at it. In retrospect, this one slide looks like a “buy me before someone else does” sign.
[News Flash: SadlyItsBradley will be doing a livestream Q&A on December 18, 2021 at 3 PM Eastern Time and will be covering Meta’s Purchase of Imagine Optix — Video link on the right]
In addition to finding patent applications by Jamali, I found a 2019 co-authored paper by Jamali, Limits of Pancharatnam Phase lens for 3D/VR/AR Applications (written while at Kent State). That paper, in turn, mentioned Imagine Optix as a maker of PBPL lenses. I’ve known about Image Optix for about 15 years for their polarization control films and had by coincidence been on their website less than two weeks ago. But this time (and for the last several days) Image Optix website domain was gone along with their Facebook page. I contacted someone I knew who was working at Imagine Optix, and all they would tell me was that they had recently left. I have reached out to several other people at Image Optix and have received no reply [I now have my answer as to why].
Trying to figure out what happened to Imagine Optix led me to SadlyItsBradley YouTube Channel with host Bradley Lynch. Lynch primarily covers Valve VR and has studied Valve’s technology and researched their patents and other activities.
Lynch and I spent about two and a half hours in a video conference comparing notes. It looks like Valve was at least part of a 2018 $9M investment in Imagine Optix, plus there was a loan for $1M for a “Master Supply Agreement.” It turns out that Valve had recently settled a lawsuit with Imagine Optix, and Lynch has been following the lawsuit. Quoting Lynch’s notes from the related video:
it looks like the initial 2018 equity purchase from Valve was for $9m on top of the loan for $1m, so it wasn’t small. Both the loan and the equity purchase have the same exact date and are referred to in the lawsuit proceedings (though the proceedings never mention the amount of the equity purchase). The SEC filing for the $9m doesn’t specify Valve’s name but it states that it all comes from a single investor. So Valve’s initial investment was actually the largest out of everyone (and larger than the $5m in 2020–whoever it was). That $9m was likely used to build out their factory and get them off of the ground to actually produce this technology at scale.
The $5 million that Image Optix raised in 2020 that apparently led to the lawsuit was from a different large, but unnamed, company that was a “Globally Recognized Brand” [Looks like it was Meta]. Lynch thinks the company might be Apple [a fair guess but I was thinking it was Meta due to the HD3]. Lynch has speculated based on the timing of the investment and the lawsuit that Valve took exception to the large company that invested the $5M.
Lynch, who closely follows Valve, filled me in that Valve has been working on PBPL using Image Optix and pancake optics (ex. 2020/0053350), the same elements as Meta’s HD3.
Lynch also said that Apple was working with PBPL/GPL and just so happened to have filed for its first GPL patent (20210048674) in the same time frame as the Imagine Optix investment. It is widely speculated that Apple’s first VR/MR will use pancake optics (ex. Apple Application US20180039052). Lynch pointed me to a 2019 partnership between Valve and Apple (and discussed in Lynch’s video). Lynch thinks that the relationship between Valve and Apple has soured since 2019.
Both Pancake Optics and PBPL/GPL variable lenses require polarized light making it reasonable to pair them Both Pancake Optics and PBPL/GPL variable lenses require polarized light, making it reasonable to pair them together. Everyone in VR and AR wants to address Vergence-Accommodation Conflict (VAC), which requires varifocal optics such as PBPL. They want very compact optics that will also support wide FOV pancake optics. There are not many available options to provide both these capabilities. Lynch also pointed me to a valve patent application for addressing focus and astigmatism, which would be critical for full vision correction in the headset.
Google is also looking at pancake optics (ex. 20210278674), but in a quick search, I did not find any PBPL patent applications (I suspect they are looking at it too). I also quickly checked Microsoft and did not see anything on either pancake optics or PBPL in the patent literature. It seems M/S are working primarily on technology that goes with their AR waveguide optics (but once again, this was only a quick search).
The question now becomes with Meta buying Image Optix, what is everyone else going to do to address VAC. The generic PBPL concept is old, but people will stake out claims on various improvements.
PBPL used by Valve, Meta’s HD3, and Image Optix are “passive” where the lens characteristics are frozen. For completeness, there are also “active” type PBPL where the lens can be driven, but the rest of this discussion will be for the passive type.
As shown in Fig. 1 (below left) from Augmented reality near-eye display using Pancharatnam-Berry phase lenses, a PBPL will take right-hand circular polarized light in and output converging left-hand circular polarized light that is converging like a convex lens, or if the input is left-handed, it will output right hand diverging light like a convex lens. So the PBPL always flips the direction of the circular polarization and converges right-hand CP light, and diverges with left-hand CP light.
The amount the light is converged/diverged is effectively a printing process. An Image Optix presentation from SPIE AR/VR/MR 2021 included a slide (above right) that describes the process. A photo-alignment layer is patterned and exposes a liquid crystal polymer. A UV light source then cures to lock the LC alignment. The processes can be repeated in a series of layers.
A drivable LC shutter selectively changes the polarized light’s right- or left-hand direction turning the passive PBPL into a varifocal lens. The Meta (aka Facebook) patent application US2021/0231952 gives examples of how multiple sets of LC shutters with PBPL can be stacked (two examples below. If they leave the LC shutter off, the circular polarization flips from left to right or right to left. If they drive the LC shutter, the polarization does not change. In this way, they can control whether each PBPL converges or diverges the light. Then at the end, they have a shutter to make the light end up left polarized based on an odd or even number of previous shutters being on, followed by a “cleanup” polarizer.
In May, I published Exclusive: Snap Spectacles Appears to Be Using WaveOptics and [an LCOS] a DLP Display. Two hours later, it was announced that Snap had bought WaveOptics for $500M. I was asked why Snap paid so much for WaveOptics, and my response was, “because they were afraid that Facebook [Meta] would buy WaveOptics the ways FB did Plessey just a few months earlier. Snap could have spent hundreds of millions on a product only to see a key component controlled by a competitor.” See my sections from an August 2021 article on Snap Recently Bought WaveOptics – Might In Part Be Due to Facebook Buying Plessey and Facebook Buyout of Plessy MicroLEDs for AR Left WaveOptics and Vuzix Scrambling).
[12/18/2021 Slight correction: Technically, Facebook didn’t “buy” Plessey, they bought exclusivity to all of Plessey’s production of MicroLEDs for use in displays. At the time it was reported that Facebook did it this way to avoid the regulatory hurdles of a buyout. No matter what it is called, Plessey stopped all work with other companies]
Does the question now become who is next? Off the top of my head, Dispelix has the closest technology to Snap’s WaveOptics. Digilens has technology for both waveguides, and their technology is very similar to that of Imagine Optics and might be able to make PBPL. DeepOptics has been developing varifocal LC technology for over a decade. And then there is Lumus which has by far the best Waveguide-based image I have seen.
[12/18/2021 Addition: Several people including SadlyItsBradley’s video have pointed me to Himax which acquired (in 2014) Liqxtal’s LC varifocal lens technology. I have not had a chance to look at it in detail but it looks like they are using a thicker type of phase LC technology to provide more optical power in a single layer.]
One thing is for sure, the big company buying spree is not over as the big companies try to corner the future of the VR/AR/MR market. And on the technical side, I don’t think this is the last article I will be writing about PBPL.