VAC By Oculus and Microsoft . . . Everywhere and Nowhere

Technically Interesting New Papers At Siggraph 2017

Both Oculus (Facebook) and Microsoft’s are presenting interesting technical research  papers at Siggraph 2017 (July 30th to August 3rd) that deal with Vergence/Accommodation (VAC).  Both have web pages (Oculus link and Microsoft link) with links to relatively easy to follow videos and the papers. But readers should take to heed the words on the Microsoft Page (which I think is applicable to both): “Note that this Microsoft Research publication is not necessarily indicative of any Microsoft product roadmap, but relates to basic research around holographic displays.” I can’t hope to try and get into all the technical details here, but they both have a lot well explained information with figures and for those that are interested, you can still learn a lot from them even if you have to skip over some of the heavy duty math. One other interesting thing is that both Oculus and Microsoft used phase controlled LCOS microdisplays at the heart of their technologies.

Briefly, VAC is the problem with stereoscopic 3-D where the apparent focus of objects does not agree with were they seem to appear with binocular vision. This problem can cause visual discomfort and headaches. This year I have been talking a lot about VAC thanks first to Magic Leap (ML article) and more recently Avegant (Avegant VAC article ) making big deals about and both raising a lot of money (Magic Leap over $1B) as a result. But least you think Magic Leap and Avegant are the only ones, there are dozens of research groups over the last decade working on VAC. Included in that number is Nvidia with a light field approach that they presented a paper in 2013 also at Siggraph (The 2013 Nvidia Paper with links embedded at the bottom of the Abstract to more information and a video)

The Oculus paper has a wealth of background/education information about VAC and figures that help explain the concepts. In many ways it is a great tutorial. They also have a very lengthy set of references that among other things confirm how many different groups have worked on VAC and this is only a partial list. I also recommend papers and videos on VAC by Gordon Wetzstein of Stanford. There is so much activity that I put “Everywhere” in the title.

I particularly liked Oculus’s Fig. 2 which is copied at the top of this article (they have several other very good figures as well as their video). They show the major classes of VAC, from a) do nothing, b) change focus (perhaps based on eye tracking), to c) Multifocal which is what I think Magic Leap and Avegant are doing, to d)&e) Oculus’s “focal surfaces(s), to f) light fields (ex. Nvidia’s 2013 paper). But light fields are in a way a short cut compared to real/true holograms which is what Microsoft’s 2017 paper is addressing (not shown in the table above but discussed in Oculus’s paper and video).

I put the “real” in front of the work “hologram” because confusingly Microsoft, for what appears to be marketing purposes, has chosen to call stereoscopic merged reality objects “holograms” which scientifically they are not. Thanks to Microsoft’s marketing clout and others choosing “if you can’t beat them joint them” in using the term, we now have the problem of what to call real/true holograms as discussed in Microsoft’s 2017 Siggraph paper.

High Level Conceptually:
  • Light Fields are a way to realize many of the effects of holograms such such as VAC and being able to see around objects. But light fields have piece-wise discontinuities. They can only reduce the discontinuities by massively trading off resolution; thus they need massive amounts of processing and native display resolution for a given visual resolution. Most of the processing and display resolution never makes it do the eye as based on where the eye is looking and focused, all but a small part of the generated image information is never seen. The redundancy with light fields tends to grow with a square law (X and Y).
  • Focus planes in effect try and cut down the Light Field square law redundancy problem by having the image redundancy grow linearly. They need multiple planes and then rely on your eye to do the blending between planes. Still the individual planes on “flat” and with a large continuous surface there would be discontinuities at the point where it would have to change planes (imagine a road going off in the distance).
  • Oculus Surfaces are in essence and improvement on focus planes where the surfaces try to conform more to the depth in the image and reduce the discontinuities. One could then argue whether it would be better to have more simple focus planes or fewer Focus Surfaces.
  • Holograms have at least an “n-cube” problem as they conceptually capture/display the image in X, Y, and Z. As the resolution increases the complexity grows extremely fast. Light fields have sometimes been described as “Quantized Holograms” at they put a finite limit on the computational and image content growth.
Oculus’s Focus Surface Approach

In a nutshell, Oculus is using an eMagin OLED to generate the image and a Jasper Display Phase Shift LCOS device to generate a “focus surface”. The focus changes focus continuously-gradually, and not on a per-pixel basis, which is why they call is a “surface”.  The figure on the right (taken from their video) shows the basic concept of a “focus surface” and how the surface roughly tracks the image depth. The paper (and video) go on to  discuss how having more than one surface and how the distance approximation “error” would compare with multi-focus planes (such as Magic Leap and Avegant).

While the hardware diagram above would suggest something that would fit in a headset, it is still at the optical breadboard stage. Even using microdisplays, it is a lot to put on a person’s head. Not to mention the cost of having in effect two displays (the LCOS one controlling the focus surface) plus all the additional optics. Below is a picture of the optical breadboard.

Microsoft (True This Time) Holograms

While Oculus’s hardware looks like something that could fit in a headset someday, Microsoft is much more of a research concept, although they did show a compact AR Prototype “glasses” (shown at right) that had a small subset of the capability of the larger optical breadboard.

Microsoft’s optical breadboard setup could support either Wide FOV or Multi-Focal (VAC) but not both at the same time (see picture below). Like other real time hologram approaches (and used by Oculus in their focal surface approach), Microsoft uses a Phase LCOS device.The Microsoft paper goes into some of the interesting things that can be done with holograms including correcting for aberrations in the optics and/or a person’s vision.

In many ways Holograms ultimate end game in display technology where comparatively everything else in with VAC is a hack/shortcut/simplification to avoid the massive computations and hardware complexities/difficulties of implementing real time holograms.

Resolution/Image Quality – Not So Much

The image quality in the Oculus Surface paper is by their admission very low both in terms of resolution and contrast. As they freely admit, it is a research prototype and not meant to be a product.

Some of these limitations are the nature of making a one-off experiment as the article points out but some of the issues may be more fundamental physics. One thing that concerns me (and pointed out in the article) in the Oculus design is that they have to pass all three colors through the same LC material and the LC’s behavior varies with wavelength. These problems would become more significant as resolution increases. I will give the Oculus paper props for both for is level of information and candor about many of the issues; it really is a very well done paper if you are interested in this subject.

It is harder to get at the resolution and image quality aspects of the the Microsoft Hologram paper as they show little images from different configurations. They can sort of move the problems around with Holograms; they can tune them and even the physical configuration for image quality, pupil size, or depth accommodation, but not all at the same time. Digital/real-time holograms can do some rather amazing things as as the Microsoft paper demonstrates but but they are still inordinately expensive both to compute and display and the image quality is inferior to more conventional methods. Solving for image quality (resolution/contrast), pupil/eyebox size, and VAC/image depth simultaneously makes the problems/cost tend to take off exponentially.

Don’t Expect to See These In Stores for Decades, If Ever

One has to realize that these are research projects going for some kind of bragging rights in showing the technical prowess, which both Oculus and Microsoft do impressively in their own ways. Note the Nvidia Light Field paper was presented at Siggraph 2013 years ago and supporting decent resolution with Light Fields is still a very far off dream. If their companies thought these concepts were even remotely practical and only a few years away, the companies would have kept them deep dark secrets. These are likely seen by their companies as so out in the future that there is no threat to letting their competition see what they are doing.

The Oculus Surface approach is conceptually better on a “per plane” than the “focus planes” VAC approaches, but then you have to ask are more simple planes better overall and/or less expensive? At a practical level I think the Oculus Surface would be more expensive and I would expect the image quality to be considerably worse. At best, the Oculus Surface would be a stop-gap improvement.

Real time high resolution holograms that will compete on image quality would seem to be even further out in time. This is why there are so many companies/researchers looking at short cuts to VAC with things like focus planes.

VAC in Context – Nowhere

VAC has been a known issue for a long time with companies and researchers working in head mounted displays. Magic Leap’s $1+B funding and their talk about VAC made it a cause célèbre in AR/VR and appears to have caused a number of projects to come out from behind closed doors (for V.C. funding or just bragging rights).

Yes, VAC is a real issue/problem particularly/only when 3-D stereoscopic objects appear to be closer than about 2 meters (6 feet) away. It causes not only perceptual problems, but can cause headaches and make people sick. Thus you have companies and researchers looking for solutions.

The problem IMO is that VAC is would be say about 20th (to pick a number) on my list of serious problems facing AR/VR. Much higher on the list are based image quality, ergonomic (weight distribution), power, and computing problems. Every VAC solution comes at some expense in terms of image quality (resolution/contrast/chromatic-abberations/etc).

Fundamentally, if you eye can pick what it focuses on, then there has to be a lot of redundant information presented to the eye that it will discard (not notice) as it focuses on what it does see. This translates into image information that must be displayed (but not seen), processing computations that are thrown away, and electrical power being consumed for image content that is not used.

I’m Conflicted

So I am conflicted. As a technologist, I find the work in VAC and beyond (Holograms address much more than VAC) fascinating. Both the Oculus and Microsoft articles are interesting and can be largely understood by someone without a PhD in the subject.

But in the end I am much more interested in technology that can reach a sizable market and on that score I don’t understand all the fuss about VAC.  I guess we will have to wait and see if Magic Leap changes the world or is another Segway or worse Theranos; you might be able to tell which way I am leaning based on what I understand.

Today, the image quality of headsets is pretty poor when compared to say direct view TVs and Monitors, the angular resolution (particularly of VR) is poor, the ergonomics are for the most part abysmal, and if you are going to wireless, the batteries are both too heavy and have too short a life. Anything that is done to address VAC makes these more basic problems not just a little worse, but much worse.

Near Eye Displays (NEDs): Gaps In Pixel Sizes

I get a lot of questions to the effect of “what is the best technology for a near eye display (NED).” There really is no “best” as every technology has its strengths and weaknesses. I plan to right a few articles on this subject as it is way too big for a single article.

Update 2017-06-09I added the Sony Z5 Premium 4K Cell Phone size LCD to the table. Their “pixel” is about 71% the linear dimension of the Samsung S8 or about half the area but still much larger than any of the microdisplay pixels. But one thing I should add is that most cell phone makers are “cheating” on what they call a pixel. The Sony Z5 Premium’s “pixel” really only has 2/3rds of an R, G, and B per pixel it counts. It also has them in a strange 4 pixel zigzag that causes beat frequency artifacts when displaying full resolution 4K content (GSMARENA’s Close Up Pixtures show of the Z5 Premium fails the show the full resolution in both directions). Note similarly Samsung goes with RGBG type patterns that only have 2/3rd the full pixels in the way they count resolution as well. These “tricks in counting are OK when viewed with the naked eye at beyond 300 “pixels” per inch, but become more problematical/dubious when used with optics to support VR. 

Today I want to start with the issue of pixel size as shown in the table at the top (you may want to pop the table out into a separate window as you follow this article). To give some context, I have also included a few major direct view categories of displays as well. I have grouped the technologies into the colored bands in the table. I have given the pixel pitch (distance between pixel centers) as well as the pixel area (the square of the pixel pitch assuming square pixels. Then to give some context for comparison I have compared the pitch and area relative to a 4.27-micron (µm) pixel pitch which is about the smallest being made in large volume. Finally there are columns showing how big the pixel would be in arcminutes when view from 25cm (250mm =~9.84inches) which is the commonly accepted near focus point. Finally there is a column showing how much the pixel would have to be magnified to equal 1-arcminute at 25cm which gives some idea about the optics required.

In the table, I tried to use smallest available pixel in a given technology that was being produced with the exception of “micro-iLED” for which I could not get solid information (thus the “?”). In the case of LCOS, the smallest field sequential color (FSC) pixel I know of is the 4.27µm one by my old company Syndiant used in their new 1080p device. For the OLED, I used the eMagin 9.3 pixel and for the DLP, their 5.4 micron pico pixel. I used the LCOS/smallest pixel as the baseline to give some relative comparisons.

One thing that jumps out in the table are the fairly large gaps in pixel sizes between the microdisplays versus the other technologies. For example you can fit over 100 4.27µm LCOS pixels in the area of a single Samsung S8 OLED pixel or 170 LCOS pixels in the area of a the pixel used in the Oculus CV1. Or to be more extreme you can fit over 5,500 LCOS pixels in one pixel of a 55-inch TV pixel.

Big Gap In Near Eye Displays (NEDs)

The main point of comparison for today are the microdisplay pixels which range from about 4.27µm to about 9.6µm in pitch to the direct view OLED and LCD displays in 40µm to 60µm that have been adapted with optics to be used in VR headsets (NEDs). Roughly we are looking at one order of magnitude in pixel pitch and two orders of magnitude in area. Perhaps the most direct comparison is the microdisplay OLED pixel at 9.3 microns versus the Samsung S8 at 4.8X linear and a 23x area difference.

So why is there this huge gap? It comes down to making the active matrix array circuitry to drive the technology. Microdisplays are made on semiconductor integrated circuits while direct view displays are made on glass and plastic substrates using comparatively huge and not very good transistor. The table below based on one in an article from 2006 by Mingxia Gu while at Kent State University (it is a little out of date, but gives lists the various transistors used in display devices).

The difference in transistors largely explains the gap. With the microdisplays using transistors made in I.C. fabs whereas direct view displays fabricate their larger and less conductive transistors on top of glass or plastic substrates at much lower temperatures.

Microdisplays

Within the world of I.C.’s, microdisplays used very old/large transistors often using nearly obsolete semiconductor processes. This is both an effort to keep the cost down and the fact that most display technologies need higher voltages than would be supported by smaller transistor sizes.

There are both display physics and optical diffraction reasons which limit making microdisplay pixels much smaller than 4µm. Additionally, as the pixel size gets below about 6 microns, the optical cost of enlarging the pixel to be seen by the human start to escalate so headset optics makers want 6+ micron pixels which are much more expensive to make. To a first order, microdisplay costs in volume are a function of area of the display so smaller pixels means less expensive devices for the same resolution.

The problem for microdisplays is even using old I.C. fabs, the cost per square millimeter is extremely high compared to TFT on glass/plastic, and yields drop as the size of the device grows so doubling the pixel pitch could result in an 8X or more increase in cost. While is sounds good to be using old/depreciated I.C. fabs, it may also mean they may not have the best/newest/highest yielding equipment or worse yet, they close down the facilities as being obsolete.

The net result is that microdisplays are no where near cost competitive with “re-purposed” cell phone technology for VR if you don’t care about size and weight. They are the only way to do a small lightweight headsets and really the only way to do AR/see through displays (save the huge Meta 2 bug-eye bubble).

I hope to pick up this subject more in some future articles (as each display type could be a long article in and of itself. But for now, I want to get onto the VR systems with larger flat panels.

Direct View Displays Adapted for VR

Direct View VR (ex. Oculus, HTC Vive, and Google Cardboard) have leveraged direct view display technologies developed for cell phones. They then put simple optics in front of the display so that people can focus the image when the display is put so near the eye.

The accepted standard for human “near vision” is 25cm/250mm/9.84-inches. This is about as close as a person can focus and is used for comparing effective magnification. With simple (single/few lens) optics you are not so much making the image bigger per say, but rather moving the display closer to the eye and then using the optics to enable the eye to focus. A typical headset uses a roughly 40mm focal length lens and then put the display at the focal lens or less (e.g. 40mm or less) from the lens.  Putting the display at the focal length of the lens makes the image focus at infinity/far away.

Without getting into all the math (which can be found on the web) the result is that with a 40mm focal length nets an angular magnification (relative to viewing at 25cm) of about 6X. So for example looking back at the table at the top, the Oculus pixel (similar in size to the HTC Vive) which would be about 0.77 arcminutes at 25cm end up appearing to cover about 4.7 arcminutes (which are VERY large/chunky pixels) and about a 95 degree FOV (depends on how close the eye gets to the lens — for a great explanation of this subject and other optical issues with the Oculus CV1 and HTC Vive see this Doc-Ok.org article).

Improving VR Resolution  – Series of Roadblocks

For reference, 1 arcminute per pixel is consider near the limit of human vision and most “good resolution” devices try to be under 2 arcminutes per pixel and preferably under 1.5. So let’s say we want to keep the ~95 FOV but improve the angular resolution by 3x linearly to about 1.5 arcminutes, we have several (bad) options:

  1. Get someone to make a pixel that is 3X smaller linearly or 9X smaller in area. But nobody makes a pixel this size that can support about 3,000 pixels on a side. A microdisplay (I.C. based) will cost a fortune (like over $10,000/eye if it could be made at all) and nobody makes transistors that a cheap and compatible with displays that are small enough. But let’s for a second assume someone figures out a cost effective display, then you have the problem that you need optics that can support this resolution and not the cheap low resolution optics with terrible chroma aberrations, god rays, and astigmatism that you can get away with 4.7 arcminute pixels
  2. Use say the Samsung S8 pixel size (a little smaller) and make two 3K by 3K displays (one for each eye). Each display will be about 134mm or about 5.26 inches on a side and the width of the two displays plus the gap between them will end up at about 12 inches wide. So thing in terms of strapping an large iPad Pro in front of your face only, it now has to be about 100mm (~4 inches) in front of the optics (or about 2.5X as far away at on the current headsets). Hopefully you are starting to get the picture, this thing is going to huge and unwieldy and you will probably need shoulder bracing in addition to head straps. Not to mention that the displays will cost a small fortune along with the optics to go with them.
  3. Some combination of 1 and 2 above.
The Future Does Not Follow a Straight Path

I’m trying to outline above the top level issue (there are many more). Even if/when you solve the display cost/resolution problem, lurking behind that is a massive optical problem to sustain that resolution. These are the problems “straight line futurists” just don’t get; they assume everything will just keep improving at the same rate it has in the past not realizing they are starting to bump up against some very non-linear problems.

When I hear about “Moore’s Law” being applied to displays I just roll my eyes and say that they obviously don’t understand Moore’s Laws and the issued behind it (and why it kept slowing down over time). Back in November 2016 Oculus Chief Scientist Michael Abrash made some “bold predictions” that by 2021 we would have 4K (by 4K) per eye and 140 degree FOV with 2 arcminutes per pixel. He upped my example above by 1.33x more pixels and upped the FOV by almost 1.5X which introduces some serious optical challenges.

At times like this I like to point out the Super Sonic Transport or SST of the 1960’s. The SST seemed inevitable for passenger trave, after all in less than 50 years passenger aircraft when from nothing to the jet age; yet today, over 50 years later, passenger aircraft still fly at about the same speed. Oh by the way, in the 1960’s they were predicting that we would be vacationing on the moon by now and having regular fights to Mars (heck, we made it to the moon in less than 10 years). We certainly could have 4K by 4K displays per eye and 140 degree FOV by 2021 in a head mounted display (it could be done today if you don’t care how big it is), but expect it to be more like the cost of flying supersonic and not a consumer product.

It is easy to play arm chair futurist and assume “things will just happened because I want them to happen. The vastly harder part is to figure out how it can happen. I lived through I.C. development in the late 1970’s through the mid 1990’s so I “get” learning curves and rates of progress.

One More Thing – Micro-iLED

I included in the table at the top Micro Inorganic LEDs, also known as just Micro-LEDs (I’m using iLED to make it clear these are not OLEDs). They are getting a lot of attention lately, particularly after Apple bought LuxVue and Oculus bought InfiniLED. These essentially use very small “normal/conventional” LEDs that are mounted (essentially printed) on a substrate. The fundamental issue is that red requires a very different crystal from blue and green (and even they have different levels of impurities). So they have to make individual LEDs and then combine them (or maybe someday grow the dissimilar crystals on the common substrate).

The allure is that iLEDs have some optics properties that are superior to OLEDs. They have tighter color spectrum, more power efficient, can be driven much brighter, less issues with burn in, and in some cases have less diffuse (better collimated) light.

These Micro-iLEDs are being used in two ways, one to make very large displays by companies such as Sony, Samsung, and NanoLumens or supposedly very small displays (LuxVue and InfiniLED). I understand how the big display approach works, there is lots of room for the LED and these displays are very expensive per pixel.

With the small display approach, they seem to have to double issue of being able to cut very small LEDs and effectively “print” the LEDs on a TFT substrate similar to say OLEDs. What I don’t understand is how these are supposed to be smaller than say OLEDs which would seem to be at least as easy to make on similar TFT or similar transistor substrates. They don’t seem to “fit” in near eye, but maybe there is something I am missing at this point in time.

Avegant “Light Field” Display – Magic Leap at 1/100th the Investment?

Surprised at CES 2017 – Avegant Focus Planes (“Light Field”)

While at CES 2017 I was invited to Avegant’s Suite and was expecting to see a new and improved and/or a lower cost version of the Avegant Glyph. The Glyph  was a hardly revolutionary; it is a DLP display based, non-see-through near eye display built into a set of headphones with reasonably good image quality. Based on what I was expecting, it seemed like a bit much to be signing an NDA just to see what they were doing next.

But what Avegant showed was essentially what Magic Leap (ML) has been claiming to do in terms of focus planes/”light-fields” with vergence & accommodation.  But Avegant had accomplished this with likely less than 1/100th the amount of money ML is reported to have raised (ML has raised to date about $1.4 billion). In one stroke they made ML more believable and at the same time raises the question why ML needed so much money.

What I saw – Technology Demonstrator

I was shown was a headset with two HDMI cables for video and USB cable for power and sensor data going to an external desktop computer all bundle together. A big plus for me was that there enough eye relief that I could wear my own glasses (I have severe astigmatism so just diopter adjustments don’t work for me). The picture at left is the same or similar prototype I wore. The headset was a bit bulkier than say Hololens, plus the bundle of cables coming out of it. Avegant made it clear that this was an engineering prototype and nowhere near a finished product.

The mixed reality/see-through headset merges the virtual world with the see-through real world. I was shown three (3) mixed reality (MR) demos, a moving Solar System complete with asteroids, a Fish Tank complete with fish swimming around objects in the room and a robot/avatar woman.

Avegant makes the point that the content was easily ported from Unity into their system with fish tank video model coming from the Monterrey Bay Aquarium and the woman and solar system being downloaded from the Unity community open source library.  The 3-D images were locked to the “real world” taking this from simple AR into be MR. The tracking was not all perfect, nor did I care, the point of the demo was the focal planes, lots of companies are working on tracking.

It is easy to believe that by “turning the crank” they can eliminate the bulky cables and  the tracking and locking to between the virtual and real world will improve. It was a technology capability demonstrator and on that basis it has succeeded.

What Made It Special – Multiple Focal Planes / “Light Fields”

What ups the game from say Hololens and takes it into the realm of Magic Leap is that it supported simultaneous focal planes, what Avegant call’s “Light Fields” (a bit different than true “light fields” to as I see it). The user could change what they were focusing in the depth of the image and bring things that were close or far into focus. In other words, they simultaneously present to the eye multiple focuses. You could also by shifting your eyes see behind objects a bit. This clearly is something optically well beyond Hololens which does simple stereoscopic 3-D and in no way presents multiple focus points to the eye at the same time.

In short, what I was seeing in terms of vergence and accommodation was everything Magic Leap has been claiming to do. But Avegant has clearly spent only very small fraction of the development cost and it was at least portable enough they had it set up in a hotel room and with optics that look to be economical to make.

Now it was not perfect nor was Avegant claiming it to be at this stage. I could see some artifacts, in particularly lots of what looked like faint diagonal lines. I’m not sure if these were a result of the multiple focal planes or some other issue such as a bug.

Unfortunately the only available “through the lens” video currently available is at about 1:01 in Avegant’s Introducing Avegant Light Field” Vimeo video. There are only a few seconds and it really does not demonstrate the focusing effects well.

Why Show Me?

So why were they more they were showing it to me, an engineer and known to be skeptical of demos? They knew of my blog and why I was invited to see the demo. Avegant was in some ways surprising open about what they were doing and answered most, but not all, of my technical questions. They appeared to be making an effort to make sure people understand it really works. It seems clear they wanted someone who would understand what they had done and could verify it it some something different.

What They Are Doing With the Display

While Avegant calls their technology “Light Fields” it is implemented with (directly quoting them) “a number of fixed digital focal planes, and then interpolate the planes in-between them.” Multiple focus planes have many of the same characteristics at classical light fields, but require much less image data be simultaneously presented to the eye and thus saving power on generating and displaying as much image data, much of which the eye will not “see”/use.

They are currently using a 720p DLP per eye for the display engine but they said they thought they could support other display technologies in the future. As per my discussion on Magic Leap from November 2016, DLP has a high enough field rate that they could support displaying multiple images with the focus changing between images if you can change the focus fast enough. If you are willing to play with (reduce) color depth, DLP could support a number of focus planes. Avegant would not confirm if they use time sequential focus planes, but I think it likely.

They are using “birdbath optics” per my prior article with a beam splitter and spherical semi-mirror /combiner (see picture at left). With a DLP illuminated by LEDs, they can afford the higher light losses of the birdbath design and support having a reasonable amount of transparency to the the real world. Note, waveguides also tend to lose/wast a large amount of light as well. Avegant said that the current system was 50% transparent to the real world but that the could make it more (by wasting more light).

Very importantly, a birdbath optical design can be very cheap (on the order of only a few dollars) whereas the waveguides can cost many tens of dollars (reportedly Hololen’s waveguides cost over $100 each). The birdbath optics also can support a very wide field of view (FOV), something generally very difficult/expensive to support with waveguides. The optical quality of a birdbath is generally much better than the best waveguides. The downside of the birdbath compared to waveguides that it is bulkier and does not look as much like ordinary glasses.

What they would not say – Exactly How It Works

The one key thing they would not say is how they are supporting the change in focus between focal planes. The obvious way to do it would with some kind of electromechanical device such as moving focus or a liquid filled lens (the obvious suspects). In a recent interview, they repeatedly said that there were no moving parts and that is was “economical to make.”

What They are NOT Doing (exactly) – Mechanical Focus and Eye/Pupil Tracking

After meeting with Avegant at CES I decided to check out their recent patent activity and found US 2016/0295202 (‘202). It show a birdbath optics system (but with a non-see through curved mirror). This configuration with a semi-mirror curved element would seem to do what I saw. In fact, it is very similar to what Magic Leap showed in their US application 2015/0346495.

Avegant’s ‘202 application uses a combination of a “tuning assembly 700” (some form of electro-mechanical focus).

It also uses eye tracking 500 to know where the pupil is aimed. Knowing where the pupil is aimed would, at least in theory, allow them to generate a focus plane for the where the eye is looking and then an out of focus plane for everything else. At least in theory that is how it would work, but this might be problematical (no fear, this is not what they are doing, remember).

I specifically asked Avegant about the ‘202 application and they said categorically that they were not using it and that the applications related to what they were using has not yet been published (I suspect it will be published soon, perhaps part of the reason they are announcing now). They categorically stated that there were “no moving parts” and that the “did not eye track” for the focal planes. They stated that the focusing effect would even work with say a camera (rather than an eye) and was in no way dependent on pupil tracking.

A lesson here is that even small companies file patents on concepts that they don’t use. But still this application gives insight into what Avegant was interested in doing and some clues has to how the might be doing it. Eliminate the eye tracking and substitute a non-mechanical focus mechanism that is rapid enough to support 3 to 6 focus planes and it might be close to what they are doing (my guess).

A Caution About “Demoware”

A big word of warning here about demoware. When seeing a demo, remember that you are being shown what makes the product look best and examples that might make it look not so good are not shown.

I was shown three short demos that they picked, I had no choice. I could not pick my own test cases.I also don’t know exactly the mechanism by which it works, which makes it hard to predict the failure mode, as in what type of content might cause artifacts. For example, everything I was shown was very slow moving. If they are using sequential focus planes, I would expect to see problems/artifacts with fast motion.

Avegant’s Plan for Further Development

Avegant is in the process of migrating away from requiring a big PC and onto mobile platforms such as smartphones. Part of this is continuing to address the computing requirement.

Clearly they are going to continue refining the mechanical design of the headset and will either get rid of or slim down the cables and have them go to a mobile computer.  They say that all the components are easily manufactureable and this I would tend to believe. I do wonder how much image data they have to send, but it appears they are able to do with just two HDMI cables (one per eye). It would seem they will be wire tethered to a (mobile) computing system. I’m more concerned about how the image quality might degrade with say fast moving content.

They say they are going to be looking at other (than the birdbath) combiner technology; one would assume a waveguide of some sort to make the optics thinner and lighter. But going to waveguides could hurt image quality and cost and may more limit the FOV.

Avegant is leveraging the openness of Unity to support getting a lot of content generation for their platform. They plan on a Unity SDK to support this migration.

They said they will be looking into alternatives for the DLP display, I would expect LCOS and OLED to be considered. They said that they had also thought about laser beam scanning but their engineers objected to trying for eye safety reasons; engineers are usually the first Guinea pigs for their own designs and a bug could be catastrophic. If they are using time sequential focal planes which is likely, then other technologies such as OLED, LCOS or Laser Beam Scanning cannot generate sequential planes fast enough to support that more than a few (1 to 3) focal planes per 1/60th of a second on a single device at maximum resolution.

How Important is Vergence/Accomodation (V/A)?

The simple answer is that it appears that Magic Leap raised $1.4B by demoing it. But as they say, “all that glitters is not gold.” The V/A conflict issue is real, but it mostly affects content that virtually appears “close”, say inside about 2 meters/6 feet.

Its not clear that for “everyday use” there might be simpler, less expensive and/or using less power ways to deal with V/A conflict such as pupil tracking. Maybe (don’t know) it would be enough to simply change the focus point when the user is doing close up work rather than have multiple focal planes presented to the eye simultaneously .

The business question is whether solving V/A alone will make AR/MR take off? I think the answer to this is clearly no, this is not the last puzzle piece to be solved before AR/MR will take off. It is one of a large number of issues yet to be solved. Additionally, while Avegant says they have solved it economically, what is economical is relative. It still has added weight, power, processing, and costs associated with it and it has negative impacts on the image quality; the classic “squeezing the balloon” problem.

Even if V/A added nothing and cost nothing extra, there are still many other human factor issues that severely limit the size of the market. At times like this, I like to remind people the the Artificial Intelligence boom in the 1980s (over 35 years ago) that it seemed all the big and many small companies were chasing as the next era of computing. There were lots of “breakthroughs” back then too, but the problem was bigger than all the smart people and money could solve.

BTW, it you want to know more about V/A and related issues, I highly recommend reading papers and watching videos by Gordon Wetzstein of Stanford. Particularly note his work on “compressive light field displays” which he started working on while at MIT. He does an excellent job of taking complex issues and making them understandable.

Generally Skeptical About The Near Term Market for AR/MR

I’m skeptical that with or without Avegant’s technology, the Mixed Reality (MR) market is really set to take off for at least 5 years (an likely more). I’ve participated in a lot of revolutionary markets (early video game chips, home/personal computers, graphics accelerators, the Synchronous DRAMs, as well as various display devices) and I’m not a Luddite/flat-earther, I simply understand the challenges still left unsolved and there are many major ones.

Most of the market forecasts for huge volumes in the next 5 years are written by people that don’t have a clue as to what is required, they are more science fiction writers than technologist. You can already see companies like Microsoft with Hololens and before them Google with Google Glass, retrenching/regrouping.

Where Does Avegant Go Business Wise With this Technology?

Avegant is not a big company. They were founding in in 2012. My sources tell me that they have raise about $25M and I have heard that they have only sold about $5M to $10M worth of their first product, the Avegant Glyph. I don’t see the Glyph ever as being a high volume product with a lot of profit to support R&D.

A related aside: I have yet to see a Glyph “in the wild” being using say on an airplane (where they would make the most sense). Even though the Glyph and other headsets exist, people given a choice still by vast percentages still prefer larger smartphones and tablets for watching media on the go. The Glyph sells for about $500 now and is very bulky to store, whereas a tablet easily slips into a backpack or other bag and the display is “free”/built in.

But then, here you have this perhaps “key technology” that works and that is doing something that Magic Leap has raised over $1.4 Billion dollars to try and do. It is possible (having not thoroughly tested either one), that Avegant’s is better than ML’s. Avegant’s technology is likely much more cost effective to make than ML’s, particularly if ML’s depends on using their complex waveguide.

Having not seen the details on either Avegant’s or ML’s method, I can’t say which is “best” both image wise and in terms of cost, nor whether from a patent perspective, whether Avegant’s is different from ML.

So Avegant could try and raise money to do it on their own, but they would have to raise a huge amount to last until the market matures and compete with much bigger companies working in the area. At best they have solved one (of many) interesting puzzle pieces.

It seems obvious (at least to me) that more likely good outcome for them would be as a takeover target by someone that has the deep pockets to invest in mixed reality for the long haul.

But this should certainly make the Magic Leap folks and their investors take notice. With less fanfare, and a heck of a lot less money, Avegant has as solution to the vergence/accommodation problem that ML has made such a big deal about.

Near-Eye Bird Bath Optics Pros and Cons – And IMMY’s Different Approach

Why Birdbaths Optics? Because the Alternative (Waveguides) Must Be Worse (and a teaser)

The idea for this article started when I was looking at the ODG R-9 optical design with OLED microdisplays. They combined an OLED microdisplay that is not very bright in terms of nits with a well known “birdbath” optical design that has very poor light throughput. It seems like a horrible combination. I’m fond of saying “when intelligent people chose a horrible design, the alternative must have seemed worse

I’m going to “beat up” so to speak the birdbath design by showing how some fundamental light throughput numbers multiply out and why the ODG R-9 I measured at CES blocks so much of the real world light. The R-9 also has a serious issue with reflections. This is the same design that a number of publications considered among the “best innovations” of CES; it seems to me that they must have only looked at the display superficially.

Flat waveguides such as used by Hololens, Vuzix. Wave Optics, and Lumus as well as expected from Magic Leap get most of the attention, but I see a much larger number of designs using what is known as a “birdbath” and similar optical designs. Waveguides are no secret these days and the fact that so many designs still use the birdbath optics tells you a lot about the issues with waveguides. Toward the end of this article, I’m going to talk a little about the IMMY design that replaces part of the birdbath design.

As a teaser, this article is to help prepare for an article on an interesting new headset I will be writing about next week.

Birdbath Optics (So Common It Has a Name)

The birdbath combines two main optical components, a spherical mirror/combiner (part-mirror) and a beam splitter. The name  “birdbath” comes from the spherical mirror/combiner looking like a typical birdbath. It is used because it generally is comparatively inexpensive to down right cheap while also being relatively small/compact while having  good overall image quality. The design fundamentally supports a very wide FOV, which are at best difficult to support with waveguides. The big downsides are light throughput and reflections.

A few words about Nits (Cd/m²) and Micro-OLEDs

I don’t have time here to get into a detailed explanation of nits (Cd/m²). Nits is the measure of light at a given angle whereas lumens is the total light output. The simplest analogy is to water hose with a nozzle (apropos here since we are talking about birdbaths). Consider two spray patterns, one with a tight jet of water and one with a wide fan pattern both outputting the exact same total amount of water per minute (lumens in this analogy). The one with the tight patter would have high water pressure (nits in this analogy) over a narrow angle where the fan spray would have lower water pressure (nits) over a wider angle.

Additionally, it would be relatively easy to put something in the way of the tight jet and turn it into a fan spray but there is no way to turn the fan spray into a jet. This applies to light as well, it is much easier to go from high nits over are narrow angle to lower nits over a wide angle (say with a diffuser) but you can’t go the other way easily.

Light from an OLED is like the fan spray only it covers a 180 degree hemisphere. This can be good for a large flat panel were you want a wide viewing angle but is a problem for a near eye display where you want to funnel all the light into the eye because so much of the light will miss pupil of the eye and is wasted. With an LED you have a relative small point of light that can be funneled/collimated into a tight “jet” of light to illuminate an LCOS or DLP microdisplay.

The combination of light output from LEDs and the ability to collimate the light means you can easily get tens of thousands of nits with an LCOS or DLP illuminated microdisplay were OLED microdisplays typically only have 200 to 300 nits. This is major reason why most see-through near eye displays use LCOS and DLP over OLEDs.

Basic Non-Polarizing Birdbath (example, ODG R-9)

The birdbath has two main optical components, a flat beam splitter and a spherical mirror. In the case a see-through designs, the the spherical mirror is a partial mirror so the spherical element acts as a combiner. The figure below is taken from an Osterhaut Design Group (ODG) patent which and shows simple birdbath using an OLED microdisplay such as their ODG R-9. Depending on various design requirements, the curvature of the mirror, and the distances, the lenses 16920 in the figure may not be necessary.

The light from the display device, in the case of the ODG R-9 is a OLED microdisplay, is first reflect away from the eye and perpendicular (on-axis) to the curved beam splitter so that a simple spherical combiner will uniformly magnify and move the apparent focus point of the image (if not “on axis” the image will be distorted and the magnification will vary across the image). The curved combiner (partial mirror) has minimal optical distortion on light passing through.

Light Losses (Multiplication is a Killer)

A big downside to the birdbath design is the loss of light. The image light must make two passes at the beam splitter, a reflective and transmissive, with a reflective (Br) and transmissive (Bt) percentages of light. The light making it through both passes is Lr x Lt.  A 50/50 beam splitter might be about 48% reflective and transmissive (with say a 4% combined loss), and the light throughput (Br x Bt) in this example is only 48% x 48%= ~23%. And “50/50” ratio is the best case; if we assume a nominally 80/20 beam splitter (with still 4% total loss) we get 78% x 18% = ~14% of the light making through the two passes.

Next we have the light loss of the spherical combiner. This is a trade-off of image light being reflected (Cr) versus being transmitted  (Ct) to the real world where Cr + Ct is less than 1 due to losses. Generally you want the Cr to be low so the Ct can be high so you can see out (otherwise it is not much of a see through display).

So lets say the combiner has Cr=11% and the Ct=75% with about 4% loss with the 50/50 beamsplitter. The net light throughput assuming a “50/50” beam splitter and a 75% transmissive combiner is Br x Cr X Bt = ~2.5% !!! These multiplicative losses lose all but a small percentage of the display’s light. And consider that the “real world” net light throughput is Ct x Bt which would be 48% x 75% = 36% which is not great and would be too dark for indoor use.

Now lets say you want the glasses to be at least 80% transmissive so they would be considered usable indoors. You might have the combiner Ct=90% making Cr=6% (with 4% loss) and then Bt=90% making Br=6%. This gives the real world transmissive about 90%x90% = 81%.  But then you go back and realize the display light equation (Br x Cr X Bt) becomes 6%x6%x90% = 0.3%. Yes, only 3/1000ths of the starting image light makes it through. 

Why the ODG R-9 Is Only About 4% to 5% “See-Through”

Ok, now back to the specific case of the ODG R-9. The ODG R-9 has an OLED microdisplay that most like has about 250 nits (200 to 250 nits is commonly available today) and they need to get about 50 nits (roughly) to the eye from the display to have a decent image brightness indoors in a dark room (or one where most of the real world light is blocked). This means they need a total throughput of 50/250=20%. The best you can do with two passes through a beam splitter (see above) is about 23%.  This forces the spherical combiner to be highly reflective with little transmission. You need something that reflects 20/23=~87% of the light and only about 9% transmissive. The real world light then making it through to the eye is then about 9% x 48% (Ct x Bt) or about 4.3%.

There are some other effects such as the amount of total magnification and I don’t know exactly what their OLED display is outputting display and exact nits at the eyepiece, but I believe my numbers are in the ballpark. My camera estimates for the ODG R-9 came in a between 4% and 5%. When you are blocking about 95% of the real world light, are you really much of a “see-through” display?

Note, all this is BEFORE you consider adding say optical shutters or something like Varilux® light blocking. Normally the birdbath design is used with non-see through designs (where you don’t have the see-through losses) or with DLP® or LCOS devices illuminated with much higher nits (can be in the 10’s of thousands) for see through designs so they can afford the high losses of light.

Seeing Double

There are also issues with getting a double image off of each face of plate beam splitter and other reflections. Depending on the quality of each face, a percentage of light is going to reflect or pass through that you don’t want. This light will be slightly displaced based on the thickness of the beamsplitter. And because the light makes two passes, there are two opportunities to cause double images. Any light that is reasonably “in focus” is going to show up as a ghost/double image (for good or evil, your eye has a wide dynamic range and can see even faint ghost images). Below is a picture I took with my iPhone camera of a white and clear menu through the ODG R-9. I counted at least 4 ghost images (see colored arrows).

As a sort of reference, you can see the double image effect of the beamsplitter going in the opposite direction to the image light with my badge and the word “Media” and its ghost (in the red oval).

Alternative Birdbath Using Polarized Light (Google Glass)

Google Glass used a different variation of the birdbath design. They were willing to accept a much smaller field of view and thus could reasonably embedded the optics in glass. It is interesting here to compare and contrast this design with the ODG one above.

First they started with an LCOS microdisplay that was illuminated by LEDs that can be very much brighter and more collimated light resulting in much higher (can be orders of magnitude) starting nits than an OLED microdisplay can output. The LED light is passed through a polarizing beam splitter than will pass about 45% P light to the LCOS device (245). Note a polarizing beam splitter passes one polarization and reflect the other unlike a the partially reflecting beam splitter in the ODG design above. The LCOS panel will rotate the light to be seen to S polarization so that it will reflect about 98% (with say 2% loss) of the S light.

The light then goes to a second polarizing beam splitter that is also acting as the “combiner” that the user sees the real world through. This beam splitter is set up to pass about 90% of the S light and reflect about 98% of the P light (they are usually much better/more-efficient in reflection). You should notice that they have a λ/4 (quarter wave = 45 degree rotation) film between the beam splitter and the spherical mirror which will rotate the light 90 degrees (turning it from S to P) after it passes through it twice. This  λ/4 “trick” is commonly used with polarized light. And since you don’t have to look through the mirror, it can be say 98% reflective with say another 3% loss for the λ/4.

With this design, about 45% (one pass through the beamsplitter) of the real world makes it through, but only light polarized the “right way” makes it through which makes looking at say LCD monitors problematical. By using the quarter wave film the design is pretty efficient AFTER you loose about 55% of the LED light in polarizing it initially. There are also less reflection issues because all the films and optics are embedded in glass so you don’t get these air to glass index mismatches of off two surfaces of a relatively thick plate that cause unwanted reflections/double images.

Google Glass design has a lot of downsides too. There is nothing you can do to get the light throughput of the real world much above 45% and there are always the problems of looking through a polarizer. But the biggest downside is that it cannot be scaled up for larger fields of view and/or more eye relief. As you scale this design up the block of glass becomes large, heavy and expensive as well as being very intrusive/distorting in looking through a big thick piece of glass.

Without getting too sidetracked, Lumus in effect takes the one thick beam splitter, and piece-wise cuts it into multiple smaller beam splitters to make the glass thinner. But this also means you can’t use the spherical mirror of a birdbath design with it and so you require optics before the beam splitting and the light losses of the the piece-wise beam splitting are much larger than a single beamsplitter.

Larger Designs

An alternative design would mix the polarizing beamsplitters of the Google Glass design above with the configuration of ODG design above.  And this has been done many times through the years with LCOS panels that use polarized light (an example can be found in this 2003 paper). The spherical mirror/combiner will be a partial non-polarizing mirror so you can see through it and a quarter waveplate is used between the spherical combiner and the polarizing beam splitter. You are then stuck with about 45% of the real world light times the light throughput of the spherical combiner.

A DLP with a “birdbath” would typically use the non-polarizing beam splitter with a design similar to the ODG R-9 but replacing the OLED microdisplay with a DLP and illumination. As an example, Magic Leap did this with a DLP but adding a variable focus lens to support focus planes.

BTW, by the time you polarized the light from an OLED or DLP microdisplay, there would not be much if any of an efficiency advantage sense to use polarizing beamsplitters. Additionally, the light from the OLED is so diffused (varied in angles) that it would likely not behave well going through the beam splitters.

IMMY – Eliminating the Beamsplitter

The biggest light efficiency killer in the birdbath design is the combined reflective/transmissive passes via the beamsplitter. IMMY effectively replaces the beamsplitter of the birdbath design with two small curved mirrors that he correct for the image being reflected off-axis from the larger curved combiner. I have not yet seen how well this design works in practice but at least the numbers would appear to work better. One can expect only a few percentage points of light being lost off of each of the two small mirrors so that maybe 95% of the light from the OLED display make it to the large combiner. Then you have the the combiner reflection percentage (Cr) multiplying by about 95% rather than the roughly 23% of the birdbath beam splitter.

The real world light also benefits as it only has to go through a single combiner transmissive loss (Ct) and no beamsplitter (Bt) loses. Taking the OGD R-9 example above and assuming we started with a 250 nit OLED and with 50 nits to the eye, we could get there with about an 75% transmissive combiner. The numbers are at least starting to get into the ballpark where improvements in OLED Microdisplays could fit at least for indoor use (outdoor designs without sunshading/shutters need on the order of 3,000 to 4,000 nits).

It should be noted that IMMY says they also have “Variable transmission outer lens with segmented addressability” to support outdoor use and variable occlusion. Once again this is their claim, I have not yet tried it out in practice so I don’t know the issues/limitations. My use of IMMY here is to contrast it with the classical birdbath designs above.

A possible downside to the the IMMY multi-mirror design is bulk/size has seen below. Also noticed the two adjustment wheel for each eye. One is for interpupillary distance to make sure the optics line up center with the pupils which varies from person to person. The other knob is a diopter (focus) adjustment which also suggests you can’t wear these over your normal glasses.

As I have said, I have not seen IMMY’s to see how it works and to see what faults it might have (nothing is perfect) so this is in no way an endorsement for their design. The design is so straight forward and a seemingly obvious solution to the beam splitter loss problem that it makes me wonder why nobody has been using it earlier; usually in these cases, there is a big flaw that is not so obvious.

See-Though AR Is Tough Particularly for OLED

As one person told me at CES, “Making a near eye display see-through generally more than double the cost” to which I would add, “it also has serious adverse affects on the image quality“.

The birdbath design wastes a lot of light as do every other see-through designs. Waveguide designs can be equally or more light wasteful than the birdbath. At least on paper, the IMMY design would appear to waste a less than most others. But to make a device say 90% see through, at best you start by throwing away over 90% of the image light/nits generated, and often more than 95%.

The most common solution to day is to start with LED illuminated LCOS or DLP microdisplay so you have a lot of nits to throw at the problem and just accept the light waste. OLEDs are still orders of magnitude in brightness/nits away from being able to compete with LCOS and DLP with brute force.

 

CES 2017 AR, What Problem Are They Trying To Solve?

Introduction

First off, this post is a few weeks late. I got sick on returning from CES and then got busy with some other pressing activities.

At left is a picture that caught me next to the Lumus Maximus demo at CES from Imagineality’s “CES 2017: Top 6 AR Tech Innovations“. Unfortunately they missed that in the Lumus booth at about the same time was a person from Magic Leap and Microsoft’s Hololens (it turned out we all knew each other from prior associations).

Among Imagineality’s top 6 “AR Innovations” were ODG’s R-8/R-9 Glasses (#1) and Lumus’s Maximus 55 degree FOV waveguide (#3). From what I heard at CES and saw in the writeups, ODG and Lumus did garner a lot of attention. But by necessity, theses type of lists are pretty shallow in their evaluations and I try to do on this blog is go a bit deeper into the technology and how it applies to the market.

Among the near eye display companies I looked at during CES include Lumus, ODG, Vuzix, Real Wear, Kopin, Wave Optics, Syndiant, Cremotech, QD Laser, Blaze (division of eMagin) plus several companies I met with privately. As interesting to me as their technologies was there different takes on the market.

For this article, I am mostly going to focus on the Industrial / Enterprise market. This is were most of the AR products are shipping today. In future articles, I plan to go into other markets and more of a deep dive on the the technology.

What Is the Problem They Are Trying to Solve?

I have had an number of people asked me what was the best or most interesting AR thing I saw at CES 2017, and I realized that this was at best an incomplete question. You first need to ask, “What problem are they trying to solve?” Which leads to “how well does it solve that problem?” and “how big is that market?

One big takeaway I had at CES having talked to a number of different company’s is that the various headset designs were, intentionally or not, often aimed at very different applications and use cases. Its pretty hard to compare a headset that almost totally blocks a user’s forward view but with a high resolution display to one that is a lightweight information device that is highly see-through but with a low resolution image.

Key Characteristics

AR means a lot of different things to different people. In talking to a number of companies, you found they were worried about different issues. Broadly you can separate into two classes:

  1. Mixed Reality – ex. Hololens
  2. Informational / “Data Snacking”- ex. Google Glass

For most of the companies were focused on industrial / enterprise / business uses at least for the near future and in this market the issues include:

  1. Cost
  2. Resolution/Contrast/Image Quality
  3. Weight/Comfort
  4. See-through and/or look over
  5. Peripheral vision blocking
  6. Field of view (small)
  7. Battery life per charge

For all the talk about mixed reality (ala Hololens and Magic Leap), most of the companies selling product today are focused on helping people “do a job.” This is where they see the biggest market for AR today. It will be “boring” to the people wanting the “world of the future” mixed reality being promised by Hololens and Magic Leap.

You have to step back and look at the market these companies are trying to serve. There are people working on a factory floor or maybe driving a truck where it would be dangerous to obscure a person’s vision of the real world. They want 85% or more transparency, very lightweight and highly comfortable so it can be worn for 8 hours straight, and almost no blocking of peripheral vision. If they want to fan out to a large market, they have to be cost effective which generally means they have to cost less than $1,000.

To meet the market requirements, they sacrifice field of view and image quality. In fact, they often want a narrow FOV so it does not interfere with the user’s normal vision. They are not trying to watch movies or play video games, they are trying to give necessary information for person doing a job than then get out of the way.

Looking In Different Places For the Information

I am often a hard audience. I’m not interested in the marketing spiel, I’m looking for what is the target market/application and what are the facts and figure and how is it being done. I wanting to measure things when the demos in the boths are all about trying to dazzle the audience.

As a case in point, let’s take ODG’s R-9 headset, most people were impressed with the image quality from ODG’s optics with a 1080p OLED display, which was reasonably good (they still had some serious image problems caused by their optics that I will get into in future articles).

But what struck me was how dark the see-through/real world was when viewed in the demos. From what I could calculate, they are blocking about 95% of the real world light in the demos. They also are too heavy and block too much of a person’s vision compared to other products; in short they are at best going after a totally different market.

Industrial Market

Vuzix is representative of the companies focused on industrial / enterprise applications. They are using with waveguides with about 87% transparency (although they often tint it or uses photochromic light sensitive tinting). Also the locate the image toward the outside of the use’s view so that even when an image it displayed (note in the image below-right that the exit port of the waveguide is on the outside and not in the center as it would be on say a Hololens).

The images at right were captured from a Robert Scoble interview with Paul Travers, CEO of Vuzix. BTW, the first ten minutes of the video are relatively interesting on how Vuzix waveguides work but after that there is a bunch of what I consider silly future talk and flights of fancy that I would take issue with. This video shows the “raw waveguides” and how they work.

Another approach to this category is Realwear. They have a “look-over” display that is not see through but their whole design is make to not block the rest of the users forward vision. The display is on a hinge so it can be totally swung out of the way when not in use.

Conclusion

What drew the attention of most of the media coverage of AR at CES was how “sexy” the technology was and this usually meant FOV, resolution, and image quality. But the companies that were actually selling products were more focused on their user’s needs which often don’t line up with what gets the most press and awards.

 

ODG R-8 and R-9 Optic with a OLED Microdisplays (Likely Sony’s)

ODG Announces R-8 and R-9 OLED Microdisplay Headsets at CES

It was not exactly a secret, but Osterhout Design Group (ODG) formally announce their new R-8 headset with dual 720p displays (one per eye) and R-9 headset with dual 1080p displays.  According to their news release, “R-9 will be priced around $1,799 with initial shipping targeted 2Q17, while R-8 will be less than $1,000 with developer units shipping 2H17.

Both devices use use OLED microdisplays but with different resolutions (the R-9 has twice the pixels). The R-8 has a 40 degree field of view (FOV) which is similar to Microsoft’s Hololens and the R-9 has about a 50 degree FOV.

The R-8 appears to be marketed more toward “consumer” uses with is lower price point and lack of an expansion port, while ODG is targeting the R-9 to more industrial uses with modular expansion. Among the expansion that ODG has discussed are various cameras and better real world tracking modules.

ODG R-7 Beam Splitter Kicks Image Toward Eye

With the announcement comes much better pictures of the headsets and I immediately noticed that their optics were significantly different than I previously thought. Most importantly, I noticed in the an ODG R-8 picture that the beam splitter is angled to kicks the light away from the eye whereas the prior ODG R-7 had a simple beam splitter that kicks the image toward the eye (see below).

ODG R-8 and R-8 Beam Splitter Kicks Image Away From Eye and Into A Curved Mirror

The ODG R-8 (and R-9 but it is harder to see on the available R-9 pictures) does not have a simple beam splitter but rather a beam splitter and curve mirror combination. The side view below (with my overlays of the outline of the optics including some that are not visible) that the beam splitter kicks the light away from the eye and toward partial curved mirror that acts as a “combiner.” This curve mirror will magnify and move the virtual focus point and then reflects the light back through the beam splitter to the eye.

On the left I have taken Figure 169 from ODG’s US Patent 9,494,800. Light from the “emissive display” (ala OLED) passes through two lenses before being reflected into the partial mirror. The combination of the lenses and the mirror act to adjust the size and virtual focus point of the displayed image. In the picture of the ODG R-8 above I have taken the optics from Figure 169 and overlaid them (in red).

According to the patent specification, this configuration “form(s) at wide field of view” while “The optics are folded to make the optics assembly more compact.”

At left I have cropped the image and removed the overlay so you can see the details of the beam splitter and curved mirror joint.  You hopefully can see the seam where the beam splitter appears to be glued to the curved mirror suggesting the interior between the curved mirror and beam splitter is hollow. Additionally there is a protective cover/light shade over the outside of the curved mirror with a small gap between them.

The combined splitter/mirror is hollow to save weight and cost. It is glued together to keep dust out.

ODG R-6 Used A Similar Splitter/Mirror

I could not find a picture of the R-8 or R-9 from the inside, but I did find a picture on the “hey Holo” blog that shows the inside of the R-6 that appears to use the same optical configuration as the R-8/R-9. The R-6 introduced in 2014 had dual 720p displays (one per eye) and was priced at $4,946 or about 5X the price of the R-8 with the same resolution and similar optical design.  Quite a price drop in just 2 years.

ODG R-6, R-8, and R-9 Likely Use Sony OLED Microdisplays

Interestingly, I could not find anywhere were ODG says what display technology they use in the 2014 R-6, but the most likely device is the Sony ECX332A 720p OLED microdisplay that Sony introduced in 2011. Following this trend it is likely that the ODG R-9 uses the newer Sony ECX335 1080p OLED microdisplay and the R-9 uses the ECE332 or a follow-on version. I don’t know any other company that has both a 720p and 1080p OLED microdisplays and the timing of the Sony and ODG products seems to fit. It is also very convenient for ODG that both panels are the same size and could use the same or very similar optics.

Sony had a 9.6 micron pixel on a 1024 by 768 OLED microdisplay back in 2011 so for Sony the pixel pitch has gone from 9.6 in 2011 to 8.2 microns on the 1080p device. This is among the smallest OLED microdisplay pixel pitches I have seen but still is more than 2x linearly and 4x in area bigger than the smallest LCOS (several companies have LCOS pixels pitches in the 4 micron or less range).

It appears that ODG used an OLED microdisplay for the R-6 then switched (likely for cost reasons) to LCOS and a simple beam splitter for the R7 and then back to OLEDs and the splitter/mirror optics for the R-8 and R-9.

Splitter/Combiner Is an Old Optic Trick

This “trick” of mixing lenses with a spherical combiner partial mirror is an old idea/trick. It often turns out that mixing refractive (lenses) with mirror optics can lead to a more compact and less expensive design.

I have seen a beam splitter/mirror used many times. The ODG design is a little different in that the beam splitter is sealed/mated to the curved mirror which with the pictures available earlier make it hard to see. Likely as not this has been done before too.

This configuration of beam splitter and curve mirror even showed up in Magic Leap applications such as Fig. 9 from 2015/0346495 shown at right. I think this is the optical configuration that Magic Leap used with some of their prototypes including the one seen by “The Information.

Conclusion/Trends – Turning the Crank

The ODG optical design while it may seem a bit more complex than a simple beam splitter, is actually probably simpler/easier to make than doing everything with lenses before the beam splitter. Likely they went to this technique to support a wider FOV.

Based on my experience, I would expect that ODG optical design will be cleaner/better than the waveguide designs of Microsoft’s Hololens. The use of OLED microdisplays should give ODG superior contrast which will further improve the perceived sharpness of the image. While not as apparent to the casual observer, but as I have discussed previously, OLEDs won’t work with diffractive/holographic waveguides such as Hololens and Magic Leap are using.

What is also interesting that in terms of resolution and basic optics, the R-8 with 720p is about 1/5th the price of the military/industrial grade 720p R-6 of about 2 years ago. While the R-9 in addition to having a 1080p display, has some modular expansion capability, one would expect there will be follow-on product with 1080p with a larger FOV and more sensors in a price range of the R-8 in the not too distant future and perhaps with integration of the features from one or more of the R-9’s add-on modules; this as we say in the electronics industry, “is just a matter of turning the crank.”

Microvision Laser Beam Scanning: Everything Old Is New Again

Reintroducing a 5 Year Old Design?

Microvision, the 23 year old “startup” in Laser Beam Scanning (LBS), has been a fun topic on this blog since 2011. They are a classic example of a company that tries to make big news out of what other companies would consider to not be news worthy.

Microvision has been through a lot of “business models” in their 23 years. They have been through selling “engines”, building whole products (the ShowWX), licensing model with Sony selling engines, and now with their latests announcement “MicroVision Begins Shipping Samples to Customers of Its Small Form Factor Display Engine they are back to selling “engines.”

The funny thing is this “new” engine doesn’t look very much different from it “old” engine it was peddling about 5 years ago. Below I have show 3 laser microvision engines from 2017, 2012, and 2013 to roughly to the same scale and they all look remarkably similar. The 2012 and 2017 engine are from Microvision and the 2013 engine was inside the 2013 Pioneer aftermarket HUD. The Pioneer HUD appears use a nearly identical engine and within 3mm of the length of the “new” engine. 


The “new” engine is smaller than the 2014 Sony engine that used 5 lasers (two red, two green, and one blue) to support higher brightness and higher power with lower laser speckle shown at left.  It appears that the “new” Microvision engine is really at best a slightly modified 2012 model, with maybe some minor modification and newer laser diodes.

What is missing from Microvision’s announcement is any measurable/quantifiable performance information, such as the brightness (lumens) and power consumption (Watts). In my past studies of Microvision engines, they have proven to have much worse lumens per Watt compared to other (DLP and LCOS) technologies. I have also found their measurable resolution to be considerably less (about half in horizontally and vertically) than they their claimed resolution.

While Microvision says, “The sleek form factor and thinness of the engine make it an ideal choice for products such as smartphones,” one needs to understand that the size of the optical engine with is drive electronics is about equal to the entire contents of a typical smartphone. And the projector generally consumes more power than the rest of the phone which makes it both a battery size and a heat issue.

Everything VR & AR Podcast Interview with Karl Guttag About Magic Leap

With all the buzz surrounding Magic Leap and this blog’s technical findings about Magic Leap, I was asked to do an interview by the “Everything VR & AR Podcast” hosted by Kevin Harvell. The podcast is available on iTunes and by direct link to the interview here.

The interview starts with about 25 minutes of my background starting with my early days at Texas Instruments. So if you just want to hear about Magic Leap and AR you might want to skip ahead a bit. In the second part of the interview (about 40 minutes) we get into discussing how I went about figuring out what Magic Leap was doing. This includes discussing how the changes in the U.S. patent system signed into law in 2011 with the America Invents Act help make the information available for me to study.

There should be no great surprises for anyone that has followed this blog. It puts in words and summarizes a lot that I have written about in the last 2 months.

Update: I listen to the podcast and noticed that I misspoke a few times; it happens in live interviews.  An unfathomable mistake is that I talked about graduating college in 1972 but that was high school; I graduated from Bradley University with a B.S. in Electrical Engineering in 1976 and then received and MSEE from The University of Michigan in 1977 (and joined TI in 1977).  

I also think I greatly oversimplified the contribution of Mark Harward as a co-founder at Syndiant. Mark did much more than just have desigeners, he was the CEO, an investor, and and the company while I “played” with the technology, but I think Mark’s best skill was in hiring great people. Also, Josh Lund, Tupper Patnode, and Craig Waller were co-founders. 

 

ODG R-9: A Peak Behind the Video Curtain

Introduction

With all the hype about Hololens and Magic Leap (ML), Osterhout Design Group (ODG) often gets overlooked. ODG has not spent as much (but still spending 10’s of millions).  ODG has many more years working in field albeit primarily in the military/industrial market.

I don’t know about all the tracking, image generation, wireless, and other features, but ODG should have the best image quality of the three (ODG, Hololens, and ML).  Their image quality was reasonably well demonstrated in a short “through the optics” video ODG made (above and below are a couple crops from frames of that video). While you can only tell so much from a YouTube video (which limits the image quality), they are not afraid to show reasonably small text and large white areas (both of which would show up problems with lesser quality displays).

Update 2016-12-26: A reader “Paul” wrote that he has seen the “cars and ball” demo live. That while the display was locked down, the cubes were movable in the demo. Paul did not know where the computing was done and it could have been done on a separate computer. So it is possible that I got the dividing line between what was “real” and preplanned a bit off. I certainly don’t think that they detected that there was a clear and a black cube, and much of the demo had to have been pre-planned/staged. Certainly it is not a demonstration of what would happen if you were wearing the headset. 

Drawn To Contradictions

As I wrote last time, I’m not a fan of marketing hyperbole and I think calling their 1080p per eye a “4K experience” is at best deliberately confusing. I also had a problem with what Jame Mackie (independent) reporter said about the section of the video starting at 2:29 with the cars and balls in it and linked to here. What I was seeing was not what he was describing.

The sequence starts with a title slide saying, “Shot through ODG smart-glasses with an iPhone 6” which I think is true as far as it what is written. But the commentary by Jame Mackie was inaccurate and misleading:

So now for a real look at how the Holograms appear, as you can see the spatial and geometric tracking is very good. What really strikes me is the accuracy and positioning.  Look how these real life objects {referring to the blocks} sit so effortlessly with the Holograms

I don’t know what ODG told the reporter or if he just made it up, but at best the description is very misleading. I don’t believe there is any tracking being done and all the image rendering  was generated off-line.

What Real Virtual Reality Looks Like

Before getting into detail on the “fake” part of the video, it is instructive to look at a “real” clip. In another part of the video there is a sequence showing replacing the tape in a label maker (starting at 3:25).

In this case, they hand-held the camera rig with the glasses. In the first picture below you can see on the phone that that they are inserting virtual a virtual object, circled in green on the phone, and missing in the “real world”. 

As the handheld rig moves around the virtual elements moves and track with the camera movement reasonably well.  There is every indication that what you are seeing is what they can actually with tracking in an image generation. The virtual elements in three clips from the video are circled in green below.

The virtual elements are in the real demonstration are simple with no lighting effects or reflections off the table. Jame Mackie in the video talks as if he actually tried this demonstrations rather than just describing what he thinks the video shows.

First Clue – Camera Locked Down

The first clue that Cars and Balls video was setup/staged video is that the camera/headset never moves. If the tracking and everything was so good, why not prove it by moving rig with the headset and camera.

Locking the camera down makes it vastly easier to match up pre-recorded/drawn material. As soon as you see the camera locked down with a headset, you should be suspicious of whether some or all of the video has been faked.

Second Clue – Black Cube Highlights Disappeared

Take a look at the black cube below showing the camera rig setup and particularly the two edges of the black cube inside the orange ovals I added. Notice the highlight on the bottom half of each edge and how it looks like the front edge of the clear plastic cube. It looks to me like the black cube was made from a clear cube with the inside colored black. 

Now look at the crop at left from the first frames showing the through the iPhone and optics view. The highlight on the clear cube is still there but strangely the highlights on the black cube have disappeared. Either they switched out the cube or the highlights were taking out in post processing. It is hard to tell because the lighting is so dim.

Third Clue – Looks Too Good – Can’t Be Real Time

2016-12-16 Update: After thinking about it some more, the rending might be in real time. They probably knew there would be a clear and black  box and rendered accordingly with simpler rendering than ray tracing. Unknown is whether the headset or another computer did the rendering. 

According to comments by “Paul” he has seen the the system running. The Headset was locked-down which is a clue that is some “cheating” going on, but he said the blocks were not in a fixed location. 

Looking “too good” is a big giveaway. The cars in the video with all their reflections were clearly using much more complex ray-tracing that was computed off-line. Look at all the reflections of the cars at left. There are both cars reflecting off the table and off the clear cube the flashing light on the police car also acts like a light source in the way it reflect off the cube.

4th Clue: How Did The Headset Know The Cube Was Clear?

One of the first things that I noticed was the clear cube. How are the cameras and sensors going to know it is clear and how it will reflect/refract light? That would be a lot of expensive sensing and processing to figure this out just to deal with this case.

5th Clue: Black Cube Misaligned

On the right is a crop from a frame where the reflection of the car is wrong. From prior frames, I have outlined the black cube with red lines. But the yellow care is visible when it should be hidden by the black cube. There also a reflection in the side of the cube around where the render image is expecting the black cube to be (orange line shows the reflection point).

How It Was Done

2016-12-26 Updates (in blue): Based on the available evidence, the video is uses some amount of misdirection. The video was pre-rendered using a ray tracing computer model with a clear cube and a perfect black shiny cube on a shiny black table being modeled.  They knew that a clear and black cube would be in the scene and locked down the camera. They may have use the sensors to detect where the blocks are to know how to rendering the image. 

They either didn’t have the sensing and tracking ability or the the rendering ability to allow the camera to move.

Likely the grids you see in the video are NOT the headset detecting the scene but exactly the opposite; they are guides to the person setting up the “live” shot as to where to place the real cubes to match where they where in the model. They got the black cube in slightly the wrong place.

The final video was shot through the optics, but the cars and balls where running around the a clear and black cubes assuming they would be there when the video was rendered. No tracking, surface detection, or complex rendering was required, just the ability to playback a pre-recorded video.

Comments

I’m not trying to pick on ODG. Their hype so far less than what I have seen from Hololens and Magic Leap.  I don’t mind companies “simulating” what images will look like provided they indicate they are simulated effects. I certainly understand that through the optics videos and pictures will not look as good as simulated images. But when they jump back and forth between real and simulated effects and other tricks, you start to wonder what is “real.”

ODG R-9 (Horizon): 1080p Per Eye, Yes Really

Lazy Reporting – The Marketing Hyperbole’s Friend

While I have not ODG’s R-9 in person yet, I fully expect that it will look a lot better than Microsoft’s Hololens. I even think it will look better in terms of image quality than what I think ML is working on. But that is not the key point of this article.

But there is also a layer of marketing hyperbole and misreporting going on that I wanted to clear up. I’m just playing referee hear and calling it like a see them.

ODG 4K “Experience” with 2K (1080p) Per Eye


2016-12-28 Update – It appears I was a bit behind on the marketing hype vernacular being used in VR. Most VR displays today, such as Oculus, take a single flat panel and split it between two eyes. So each eye sees less than half (some pixels are cut off) of the pixels. Since bigger is better in marketing, VR makers like to quote the whole flat panel size and not the resolution per eye. 

ODG “marketing problem” is that historically a person working with near eye displays would talk in in terms of “resolution per eye” but this would not be as big by 2X as the flat panel based VR companies market. Rather than being at a marketing hype disadvantage, ODG apparently has adopted the VR flat panel vernacular, however misleading it might be. 


I have not met Jame Mackie nor have I watched a lot of his videos, but he obviously does not understand display technology well and I would take anything he says about video quality with a grain of salt. If should have understood that ODG’s R-9 has is not “4K” as in the title of his YouTube video: ODG 4K Augmented Reality Review, better than HoloLens ?. And specifically he should of asked questions when the ODG employee stated at about 2:22, “it’s two 1080p displays to each eye, so it is offering a 4K experience.

What the ODG marketing person was I think trying to say was that somehow having 1080p (also known as 2K) for each eye was like having a 2 times 2K or “4K equivalent” it is not. In stumbling to try and make the “4K equivalent” statement, the ODG person simply tripped over his own tongue to and said that there were two 1080p devices per eye, when he meant to say there were two 1080p devices in the glasses (one per eye). Unfortunately Jame Mackie didn’t know the difference and did not realize that this would have been impossible in the R-9’s form factor and didn’t follow up with a question. So the  false information got copied into the title of the video and was left as if it was true.

VRMA’s Micah Blumberg Asks The Right Questions and Get The Right Answer – 1080p Per Eye

This can be cleared up in the following video interview with Nima Shams, ODG’s VP of Headworn: “Project Horizon” AR VR Headset by VRMA Virtual Reality Media“. When asked by Micah Blumberg starting at about 3:50 into the video, “So this is the 4K headset” to which Nima Sham responds, “so it is 1080p to each eye” to which Blumberg astutely makes sure to clarify with, “so we’re seeing 1080p right now and not 4K” to which Nima Sham responds, “okay, yeah, you are seeing 2K to each eye independently“.  And they even added an overlay in the video “confirmed 2K per eye.” (see inside the read circle I added).

A Single 1080p OLED Microdisplay Per Eye

Even with “only” 1080p OLED microdisplay per eye with a simple optical path the ODG R-9 should have superior image quality compared to Hololens:

  1. OLEDs should give better contrast than Hololens’ Himax LCOS device
  2. There will be no field sequential color breakup with head or image movment as there can be with Hololens
  3. They have about the same pixels per arc-minute at Hololens but with more pixels they increase FOV from about 37 degrees to about 50 degrees.
  4. Using a simple plate combiner rather than the torturous path of Hololens’ waveguide, I would expect the pixels to be sharper and with little visible chroma aberrations and no “waveguide glow” (out of focus light around bright objects). So even though the angular resolution of the two is roughly the same, I would expect the R-9 to look sharper/higher resolution.

The known downsides compared to Hololens:

  1. The ODG R-9 does not appear to have enough “eye relief” to support wearing glasses.
  2. The device puts a lot of weight on the nose and ears of the user.

I’m not clear about the level of tracking but ODG’s R-9 does not appear to have the number of cameras and sensors that Hololens has for mapping/locking the real world. We will have to wait and see for more testing on this issue. I also don’t have information on how comparable the level of image and other processing is done by the ODG relative to Horizon.

Conclusion

Micah Blumberg showed the difference between just repeating what he is told and knowing enough to ask the right followup question. He knew that ODG had a 4K marketing message was confusing and that what he was being told was at odds with what he was being told so he made sure to clarify it. Unfortunately while James Makie got the “scoop” on the R-9 being the product name for Horizon, he totally misreported the resolution and other things in his report (more on that later).

Lazy and ill informed reporters are the friend and amplifier of marketing hyperbole. It appears that ODG is trying to equate dual 1080p displays per eye with being something like “4K” which is really is not. You need 1080p (also known as 2K) per eye to do stereo 1080p, but that is not the same as “4K” which which is defined as 3840×2060 resolution or 4 times the spatial resolution of 1080p. Beyond this, qualifiers of like “4K “Experience” which has no real meaning are easily dropped and ill informed reporters will report it as “4K” which does have a real meaning.

Also, my point is not meant to pick on ODG, they just happen to be the case at hand. Unfortunately, most of the display market is “liars poker.” Companies are fudging on display specs all the time. I rarely see a projector that meets or exceeds it “spec” lumens. Resolutions are often spec’ed in misleading ways (such as specifying the input rather than the “native” resolution). Contrast is another place were “creative marketing” is heavily used. The problem is that because “everyone is doing it” people feel they have to just to keep up.

The problem for me comes when I have to deal with people that have read false or misleading information. It gets hard to separate truth from marketing exaggeration.

This also goes back to why I didn’t put much stock in the magazine reports about Magic Leap looked. These reports were made by people that were easy to impress and likely not knowledgeable about display devices. They probably could not tell the display resolution by 2X in each direction or would notice even moderately severe image problems. If they were shown a flashy looking demo they would assume it was high resolution.

One More Thing – Misleading/Fake “True Video”

It will take a while to explain (maybe next time), I believe the James Makie video also falsely indicates at 2:29 in the video (the part with the cars and the metal balls on the table), that what is being shown is how the ODG R-9 works.

In fact, while the images of the cars and balls are generated by the R-9, there tracking of the real world and the reflections off the surfaces are a well orchestrated FAKE. Basically they were playing a pre-rendered video though the glasses (so that part is likely real). But clear and black boxes on the table where props there to “sell the viewer” that this was being rendered on the fly.  There also appears to be some post-processing in the video. Most notably, it looks like the black box was modified in post production. There are several clues in the video that will take a while to explain.

To be fair to ODG, the video does not claim to not be fake/processed, but the way it is presented within Jame Makie’s video is extremely misleading to say the least. It could be that the video was taken out of context.

For the record, I do believe the video starting at 4:02 which I have analyze before is a genuine through the optics video and is correctly so identified on the video. I’m not sure about the “tape replacement” video at 3:23, I think it may be genuine or it could be some cleaver orchestrating.