Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Thee Business Insider (BI) Headline states, “Inside Microsoft’s mixed reality mess, where confusion, rivalries, and canceled projects have roiled the company’s metaverse strategy.” (behind paywall). Other news sources have also reported problems at Hololens such as Windows Central’s story saying “History seems to be repeating itself as Microsoft’s mixed reality ambitions are a chaotic disaster” (the title of the Windows Central story uses more graphic language).
This will be just a quick take article with some of my comments and perspective.
As I was writing this article, I became aware that Alex Kipman had tweeted a response to an Engadget article which sites the BI article and says that Hololens 3 has been canceled.
Quoting Kipman’s Tweet:
“don’t believe what you read on the internet. #HoloLens is doing great, and if you search said internet they also said we had canceled #HoloLens2… which last I checked we shipped with success [)-)”
The BI article summarized the situation with Kipman as:
“Inside Microsoft’s mixed reality unit run by Hololens cocreator and Microsoft technical fellow Alex Kipman, more than 20 current and former employees who spoke to Insider, described confusion and strategic uncertainty as different factions argue about its future.”
Kipman’s response is a non-denial and more of a deflection. Sure, there are false reports all the time, but he did not deny either the specific allegation that the HL3 was canceled or the much broader issues reported in BI and elsewhere of turnover and turmoil within the Hololens group, with many people specifically criticizing Kipman’s leadership. My own sources say that the HL3 program was at least dramatically delayed in mid 2021.
As this blog demonstrated in Hololens 2 Display Evaluation (Part 2: Comparison to Hololens 1), Kipman is willing to look in a camera and lie about specifications (see below). Specifically, in the MWC19 video which introduced the Hololens 2, Kipman claimed that Hololens 2 had “47 pixels per degree,” but when I checked it, the Hololens 2 was less than 20 pixels per degree in terms of measurable. Kipman also played marketing games when he said that Hololens 2 had double the FOV of Hololens 1 when FOV was typically measured linearly (see for example, Microsoft Significantly Misrepresented HoloLens 2’s Field of View at Reveal) He later “clarified” that he meant “in-area,” which squares the value of the linear measurement. When a person fudges and lies about numbers you can later measure, their credibility on thing you can’t measure is completely lost.
Microsoft’s marketing appeared to try and spread FUD about my extensive analysis of the Hololens 2 image quality problems and I responded in: Hololens 2 Display Evaluation (Part 6: Microsoft’s FUD on Photographs)
The Hololens program started with a “fib,” calling what they were doing “Holograms” when there were no holograms anywhere in Hololens, not even in the optical elements. It indicated that the people running Hololens were putting hype ahead of substance and that all their claims might be suspect. Because Microsoft is so big, others decided it if you could beat them, join them, and the whole AR industry was soon misusing the term “Hologram” to mean 3-D stereo vision. Now, what word do you use for true “holograms?”
Then came Hololens 2, and they over-promised. They flat lied about the display’s resolution, and the overall image quality was horrible. They wasted a lot of money trying to get laser beam scanning to work and went backward from Hololens 1 in display quality (see: Hololens 2 Display Evaluation (Part 2: Comparison to Hololens 1). I expect better out of an established company.
Hololens 2 did greatly improve ergonomics and comfort (a lesson Magic Leap 2 missed), as I wrote about in Hololens 2 First Impressions: Good Ergonomics, But The LBS Resolution Math Fails!. Most importantly, they keep enough eye relief to be worn with typical glasses and added the ability to flip up the display. In many ways, Hololens 2 showed wearing comfort, including eye relief for glasses, and SLAM tracking is more important than a good display, at least in their enterprise applications.
It has always been clear that the Hololens program was spending much more than the revenue from product sales. The word was it was a loss-leader for Azure web services, and the so-called $22B US Army contract for “Hololen 2” was mostly for the web services contract. There are so many things wrong with the Hololens 2, including the variant Microsoft made for the US Army, that any rational person with military experience and technical knowledge should know it was ridiculous to say it would be deployable with landed troops anytime soon. Late last year the program was reported as being delayed.
Many other news sources focus on BI’s reported cancellation of Hololens 3 in mid-2021. I heard reports as far back as 2020 that Microsoft was shopping around for an LCOS display for use in the Hololens 3. Hololens 2 uses Laser Beam Scanning (LBS), which resulted in a horrible image (see my Hololens 2 Display Evaluation Series). Then In 2021, those same sources said that Microsoft had delayed or canceled the program. I was unsure if they had decided to stick with LBS or not. It looks like we now know.
LBS has been the fool’s gold of small displays for headsets and small projectors for near 30 years, at least since Microvision’s founding in 1993. This blog has been reporting on the problems with LBS since 2011. Microvision was quick to brand me a “Soothsayer” in a December 2011 8-K statement to the SEC, which gave this blog a lot of publicity in its early days and inspired me to start a Soothsayer series.
Microvision has had a history of “one and done” design wins including Sony Projector, Celluon Projector, Pioneer HUD, Ragentek Projector Phone.
According to SEC filings, Microvision has lost since its founding ~$617M. Microsoft, which hired key technical people from Microvision and paid a license to Microvision, likely lost several hundreds of millions more trying to make the dreadful display in the Hololens 2. About the same time Microsoft’s licensing deal was made, Microvision strongly “pivoted” to being primarily an automotive LIDAR company, having little history in LIDAR before, after seeing the valuation of LIDAR companies. I have always found Microvision’s main product was selling stock.
I recently put out a story about LBS for AR in AWE 2021 Part 2: Laser Scanning – Oqmented, Dispelix, and ST Micro. I went in with low expectations and frankly it was less than I expected. It more served to show how far away they are from other display technologies.
The Hololens program has always been an “R&D project that escaped the lab.” It looks like Microsoft’s management saw Facebook buying Oculus plus all the noise that Magic Leap was making and threw money at an R&D program to turn out a R&D project in a product-like case.
My first reaction was that the Hololens 2 program’s choice of LBS for the display technology proved that the Hololens program was an R&D effort out of control and they didn’t know what they were doing. It likely cost Microsoft hundreds of millions of dollars, caused a whole new set of problems for the waveguides, increased the size and weight, increased the unit cost, and greatly delayed the whole effort. Hololens 1 uses a mediocre and moderately low-resolution LCOS display. Microsoft could have had their pick of much better devices that would have delivered more than four times the resolution of what the HL2 gave.
I remain skeptical that laser scanning display will prove a viable display technology until someone can prove it is anywhere close to being competitive with other display technologies.
Based on the BI story, by mid-2021, the upper management of Microsoft realized that the Hololens program was spending a lot of resources with little chance of ever turning the corner from being a niche market product. I don’t give Kipman’s non-denial, denial, much credibility given his past record of playing with the truth.
Doesn’t seem like new information, though. It’s also odd that at the end of the article it asks for more Microsoft employees to contact them as sources, as if they don’t have as many sources or the newest ones as they’re claiming. We knew people were changing jobs with crazy offers from other companies. We also knew months ago that IVAS FOV was being changed, cyber security WIFI tests had been postponed, and with both COVID and the rushed Afghan withdrawal there wasn’t much urgency, so the bigger combat simulation field tests and full production got moved to the next year. All old news, and “not there yet” doesn’t come off as much more than taking old comments out of context. Last year Microsoft employees said Hololens 3 would be a more-mobile consumer version of Hololens 2. With the IVAS system so much more capable, that no longer makes sense. There’s also no reason that those consumers or academics who want the current Hololens 2 can’t just buy one already. Either Microsoft ought to wait until IVAS is finished or just do their own version of a Quest or Pimax using conventional VR technology if they want in on that bandwagon. I also have to agree with the idea that AR is more obviously an enterprise and military application, but VR spreads more easily into consumer spaces, too. If you’re going to do VR for the consumer segment, again, just get together with Samsung and build a VR headset. As for people leaving or protesting over Hololens technology being used by the military, again, old news and I think those Washington State hippy peaceniks are out of touch with reality. Microsoft is probably better off without employees who can’t understand the utility of technology that could improve effectiveness, shorten wars, and lower civilian casualties. They’re probably the same crowd that wants to de-fund the police.
Wrt “putting hype ahead of substance”: Another horrible one is “The only device with a Holographic Processing Unit”, which is really just a (good) DSP…
When do you think IVAS will be ready if not this year? What tech are they still waiting for?
Whether IVAS is “ready” this year is more of a political than a technical decision. It has so many faults, from blocking 60% of the ambient light to being too heavy, too fragile, to interfering with the use of weapons like rifles. The Navy seals I work with were rolling on the floor with laughter when they heard that soldiers would have to hold the guns differently. I certainly don’t see troops walking around in the open using it in combat ever. It could be used in training where people’s lives are not on the line.
Whether the program continues is more of a political matter. It would not be the first time that the government bought a headset for the troops that sat in boxes (scroll down about halfway for the story about Microvision units sitting in boxes: https://archive.seattletimes.com/archive/?date=20071014&slug=favorfactory14m)
While I’m not diminishing the points you raise, I was curious to hear your perspective on if there are technical limitations to what $30,000 (cost per headset) can buy versus what the Army has started they intend to do with augmented reality. Are the updates they’re giving about FOV and brightness going to be the last of the technical hurdles before this is reasonably functional?
Things like weight and how one shoots seem like tradeoffs needed to field a functional technology at this point. The question of “are these trade offs worth it” is political, as you say.
I believe that article is referring to the Land Warrior program – worth reading this article: https://www.popularmechanics.com/military/a1590/4215715/
They are AR goggles with a clear lens intended to be used with military, search & rescue…. How do you get high image brightness without LBS? Which is more important – clarity, or being able to see the image at all?
Microsoft Hololens 2 are not particularly bright. Microsoft rated the HL2 at 500 nits and that is only their “peak in the near center brightness.” Lumus has demonstrated waveguides that are much more transparent that have up to 7,000 nits. Both Digilens and Avegant have demonstrated getting to 2,000 nits (albeit at only 30-degrees) with LCOS but will be going to 50 degrees.
I think you may be repeating misinformation that LBS has magical properties that nothing else can do.
I think you have fallen into the trap of being overly critical. I think that, at least, your criticism of Kipman’s use of the word hologram is unjustified.
The word is not restricted to a hologram produced on film using a laser. Researchers have demonstrated the use of other technologies that produce true holograms that show the object being displayed from all angles. A HMD can function as a hologram because it can give the user the ability to view all of an object from all angles. Though, a single HMD doesn’t show all angles at once, a single person cannot see all angles at once. Multiple HMDs for multiple observers can see a virtual object from any angle they want.
You have plenty of solid criticisms, so I would drop the hologram criticism, and guard against other similar arguments.
First, I was aiming to criticize “Hologram” at Microsoft in general. I don’t particularly appreciate changing the meaning of well-established words for marketing purposes. Now we have to put the word “true” in front of anything that is truly a hologram to distinguish it from the marketing concept. I was trying to say that it made Hololens look deceitful from the start. So what was Microsoft supposed to call their true Hologram-based glasses that they presented at Siggraph 2017 (https://www.microsoft.com/en-us/research/publication/holographic-near-eye-displays-virtual-augmented-reality/)?
I see it differently than you. I think you are trying to rationalize the misuse of the term and suggest you review the definition (The Wikipedia article does a good job of explaining true vs false Hololgrams). A true Hologram utilizes the interference of light to produce the image. I can tell you that the technical people were pretty sheepish about it when asked.
I was similarly critical of Magic Leap for calling what they did a “light field.” Once again, it was not a light field but a couple of focus planes. Companies like Creal and Looking Glass make true light fields devices.
I was calling Kipman for flat-lying was over the “47 pixels per degree.” That is clearly a lie as can be seen both in the pictures I took and by the human eye when comparing Hololens to any true display with that resolution.
In your LBS cartoon you mention Braun. I suppose you meant to write Bosch. Clearly, they dipped their feet into the topic and then went silent shortly after.
Oops. Thanks, I did mean Bosch. It has been fixed after about 18,000 people have viewed the article. I don’t think Bosch has given up yet. They still have a laser projection module on the Bosch-Sensortec web page (https://www.bosch-sensortec.com/applications-solutions/wearables-hearables/).
How would one create infinite focus with microLED ?
I assume you are talking about a focus-free projector. You would have to have a very high f-number projection lens or small aperture (essentially a pinhole).
Unfortunately, a MicroLED outputs roughly Lambertian light, which would be extremely inefficient as only the straightest light rays from the LEDs would make it out. So it would be highly impractical. MicroLEDs are a bad starting point for making a focus-free projector.
Wow, you still short Microvision? Even after institutions loading the boat for the last 6 months? Even after you were wrong about Hololens 2 not having MVIS? At what point do you realize that it’s ok to be wrong?
First of all, you start with a lie that I have shorted Microvision (I have never shorted any stock).
Yes, I could not believe at first that the people at Hololens would be so dumb as to go with LBS, and as I have shown the results are terrible. HL2 has lousy image quality and a very large projection engine as I have proven (see my Hololens 2 image quality series). Also, Hololens 2 had terrible problems getting LBS to work with their waveguides and other optics. Even with the license to Microsoft, Microvision has continued to lose about $1M/month rain or shine. Microvision has now “pivoted” to LiDAR seemingly to boost the stock price.