Review At $3,000 Microsoft’s Hololens isn’t going to be in too many gamers’ Christmas stockings this year, being more of a chance for corporates to work out the business case for augmented reality.
I did manage to get my hands on, and head in, a Hololens set. I did so thanks to Kazendi who – for £17,500 – can knock out a fixed-price prototype that let CTOs and CMOs to get their head around what can actually be done with Microsoft’s headset.
My impression of Hololens was somewhat unusual for a Microsoft Version 1.0 product in that it actually (mostly) works, though there are gaps you need to paper over in speech output – and, as you’d expect, you have to elbow past Microsoft updates that break your code.
Back when I coded for Google Glass, the smaller system could be handled by a single development environment, whereas Microsoft has left behind the fading Silverlight framework largely in favour of Unity3D. There you do most of the heavy lifting development such as creating holograms, using Visual Studio for scripting and integration.
Unity is pretty popular amongst game devs and there’s enough use in education for the pool to be growing, even though fewer corporate programmers use it. If you’re more comfortable with DirectX or XAML then they’re available, meaning that you can easily adapt 2D apps to appear as panes in your 3D environment without too much pain and of course you get the rest of the Microsoft stack.
The GUI (or should it be HUI?), still called the Shell, is easy enough – just look at an icon and click with your fingers – but it is more taxing than using a mouse because you have to learn new habits on moving your head relative to your eyes. Things move out of its field of view, which is about two thirds of your normal vision.
It's not quite immersive, a fact that’s important given this is supposed to be a work tool where full immersion gets pretty tiring if you try doing it all day. Although the Hololens has Kinect, which originated for Xbox, in its ancestry the two aren’t yet playing nicely together – but you do have gesture support within a visibility cone. Gaze is analogous to input focus in Windows and can be paired to speech input so that your app can directly get the context for the object you’re talking about. Having the microphone in close fixed proximity to your mouth helps it guess what you’re saying.
Handling this as well as cameras and projection would is why the Hololens has hardcore built-in processing capabilities so the display can keep up with your movements – even when the upgrades to your reality are in complex environments and the colours are as bright as you might want. As we’re still early in the lifecycle, you can’t get directly at as much of this as you might like for dynamic display maps, with your code largely dealing with a series of what amounts to snapshots rather than streams. Communications aren’t fragile, but with this much data flying about, you might want to check your Wi-Fi router is near and newish.
As for battery life, the Hololens will run for a couple of hours between charges. If you use the official Microsoft adaptor then you’ve got a similar wait, but if you charge via a USB port then you’ll have eight hours to fill.
“Disruption” is the current corporate buzzword, and we’ve learned from Register CTO roundtables that their job involves warming the board up with low cost/risk pilots before embarking on a large multi-year middleware revamp.
One example is evaluating the insurance risks for buildings where an expensive inspector has to wander round taking notes and reconciling them with last year. The Hololens prototype lets you capture the locations of fire extinguishers, pipes, trip hazards and so on together with enough imagery to give them context. Integrating this with back office data such as plans of buildings, managing safety certificates and storing the much greater volume of data this generates in turn is a formidable task, so to get decision-makers on board and work out what it is you really want needs a few iterations – or, brutally, to decide it simply isn’t worth the effort.
This, it would seem, is a good scenario for augmented reality; that is, for the Hololens. For applications like this, the Hololens suite can work. Using the test software, it can capture where you drop the holograms using Wi-Fi. If it’s a space your Hololens has seen before, it can be loaded from memory – otherwise you spend a minute or two looking at walls to capture the spatial map.
Apps can access a simplified feed where features such as the walls and floor have been identified by simple geometry and size, which works fine if (unlike my office) walls are at 90 degrees to the floor. It can also spot sofas and tables but for anything more complex you are into the rather more intricate task of parsing the mesh
That lets you place things on or around objects and keep their position and visibility updated as you move around (and have code to move), but the sheer volume of the mesh means that changing objects rather than decorating them is likely to tax the whole system as well as needing serious programming effort.
If you’re thinking about an AR/VR for your firm, then it would seem that Hololens is is the path of least resistance, at least for corporate developers, as they have APIs that look familiar and work well enough with the existing Microsoft toolset.
To make best use of it you’ll need some graphics skills that you might need to rent for the purpose, but it will work – and provided you don’t mind looking a bit daft wearing Microsoft’s gear, it is the sort of thing that’ll help the board appreciate that in-house IT isn’t a dinosaur after all. ®