Month: February 2019

HoloLens 2 announcement Quick Recap

James Ashley3 Comments

Basic sensors look the same between HL1 and HL2, which is nice. Still 4 monochrome cameras for SLAM, 1 TOF for spatial mapping. For me personally, knowing the size of the battery and the refresh rate of the TOF with more consistent power is really huge. Also curious about burn in from the battery-bun. Doesn’t it get hot? And the snapdragon 850 is just an overclocked 845. That’s going to be a bit hotter too? Also curious why MEMS? It’s not cheaper or lighter or much smaller than LCOS so, I think, must be the reason for the FOV improvement (right?) but  Karl Guttag complains this will make for more blurry digital images (complain? who Karl?).

Piano playing stage demo was cool but there was noticeable awkwardness in the way the hand gestures were performed (exaggerated movements) — which is actually really familiar from working with magic leap’s hand tracking. Either that needs to get better by the summer or there’s going to be some serious disappointment down the road.

Anyone seen any performance comparisons between the NVidia Tegra X2 (Magic Leap’s SOC) and Qualcomm 850 (HoloLens 2’s SOC)? I know the 845 (basically same as 850) was regarded as better than X1, but most assumed X2 would leapfrog. Haven’t found anything conclusive, though.

Doubling of FOV between HL1 and HL2 might have been miscommunicated and meant something different from what most people thought it meant. It turns out the FOV of the HL2 is actually very close to the FOV of the Magic Leap One, both of which are noticeably bigger than the original HoloLens but still significantly smaller than what everyone says they want.

My friend and VR visionary Jasper Brekelmans calculated out the HL2 FOV in degrees to be 43.57 x 29.04 with a diagonal of
52.36. Magic Leap is 40 x 30, with a diagonal of 50 (thank you magic leap for supporting whole numbers).

From now on, whenever someone talks about the field of view in AR/VR/MR/XR, we’ll all have to ask if that is being measured in degrees or steradians. Oh wells.

Internet bad boy Robert Scoble turns out to probably have one of the most interesting takes on the HoloLens 2 announcement. I hope his rehabilitation continues to go well. On the other hand it was a really bad week for tech journalists in general.

Unity even made an announcement about HoloLens 2 and even though they have been working on their own in-house core tools for XR development, way down deep in the fine print they are saying that you will need to use the Mixed Reality Toolkit v2 to develop for HoloLens – which is very eeeeenteeresting.

Coolest thing for me, outside of the device itself, was the Azure Spatial Anchors announcement. No one is really paying attention to Azure Spatial Anchors, yet, but this is a game changer. It means implementing Vernor Vinge’s belief circles. It means anyone can build their own Pokemon Go app. And it works on ARKit, ARCore and HoloLens –> so the future of XR is cross-platform.

Mike Taulty, who I can’t say enough great things about, as usual has dived in first and written up a tour of the new service.

Crap, my boss just came back from lunch. Gotta work now.

The Fork in Mixed Reality

James Ashley

Futuristic Spaceman - Programmer Reading Projected Information

Yogi Berra gnomically said, “when you come to a fork in the road, take it.”  On the evening of Friday, February 1st, 2019 at approximately 9 PM EST, that’s exactly what happened to Mixed Reality.

The Mixed Reality Toolkit open source project, which grew out of the earlier HoloLens Toolkit on github, was forked into the Microsoft MRTK and a cross-platform XRTK (read the announcement). While the MRTK will continue to target primarily Microsoft headsets like the HoloLens and WMR, XRTK will feature a common framework for HoloLens, Magic Leap, VR Headsets, Mobile AR – as well as HoloLens 2 and any other MR devices that eventually come on the market.

So why did this happen? The short of it is that open source projects can sometimes serve multiple divergent interests and sometimes they cannot. Microsoft was visionary in engineering and releasing the original HoloLens MR Headset. They made an equally profound and positive step back in 2016 by choosing to open source the developer SDK/Framework/Toolkit (your choice) that allows developers to build Unity apps for the HoloLens. This was the original HoloLens Toolkit (HLTK).

While the HLTK started as a primarily Microsoft engineering effort, members of the community quickly jumped in and began contributing more and more code to the point that the Microsoft contributions became a minority of overall contributions. This, it should be noted, goes against the common trend of a single company paying their own engineers to keep an open source project going. The HLTK was an open source success story.

In this regard, it is worth calling out two developers in particular, Stephen Hodgson and Simon Jackson, for the massive amounts of code and thought leadership they have contributed to the MR community. Unsung heroes barely captures what they have done.

In 2017 Microsoft started helping to build occluded WinMR (virtually the same as VR) devices with several hardware vendors and it made sense to create something that supported more than just the HoloLens. This is how the MRTK came to be. It served the same purpose as the HLTK, to accelerate development with Unity scripts and components, but now with a larger perspective about who should be served.

In turn, this gave birth to something that is generally known as MRTK vNext, an ambitious project to support not just Microsoft devices but also platforms from other vendors. And what’s even more amazing, this was again driven by the community rather than by Microsoft itself. Microsoft was truly embracing the open source mindset and not just paying lip service to it as many naysayers were claiming.

But as Magic Leap, the other major MR Headset vendor, finally released their product in fall, 2018, things began to change. Unlike Microsoft, Magic Leap developed their SDK in-house and threw massive resources at it. Meanwhile, Microsoft finally started throwing their engineers at the MRTK again after taking a long hiatus. This may have been in response to the Magic Leap announcement or equally could have been because the team was setting the stage for a HoloLens 2 announcement in early 2019.

And this was the genesis of the MR fork in the road: For Microsoft, it did not make sense to devote engineering dollars toward creating a platform that supported their competitors’ devices. In turn, it probably didn’t make sense for engineers from Google, Magic Leap, Apple, Amazon, Facebook, etc. to devote their time toward a project that was widely seen as  a vehicle for Microsoft HMDs.

And so a philosophical split needed to occur. It was necessary to fork MRTK vNext. The new XRTK (which is also pronounced “Mixed Reality Toolkit”) is a cross-platform framework for HoloLens as well as Magic Leap (Lumin SDK support is in fact already working in XRTK and is getting even more love over the weekend even as I write).

But XRTK will also be a platform that supports developing for Oculus Rift, Oculus Go, HTC Vive, Apple ARKit, Google ARCore, the new HoloLens 2 which may or may not be announced at MWC 2019, and whatever comes next in the  Mixed Reality Continuum.

So does this mean it is time to stick a fork in the Microsoft MRTK? Absolutely not. Microsoft’s MRTK will continue to do what people have long expected of it, supporting both HoloLens and Occluded WinMR devices (that is such a wicked mouthful — I hope someone will eventually give it a decent name like “Windows Surface Kinect for Azure Core DotNet Silverlight Services” or something similarly delightful).

In the meantime, while Microsoft is paying its engineers to work on the MRTK, XRTK needs fresh developers to help contribute. If you work for a player in the MR/VR/AR/XR space, please consider contributing to the project.

Or to word it in even stronger terms, if you give half a fork about the future of mixed reality, go check out  XRTK  and start making a difference today.