Ben Thompson’s excellent Stratechery newsletter took this topic on this morning. Why is Facebook doing VR, he asks, while pointing out that Apple is the apparent winner?
I have been watching this space in much more detail than Ben, having cowritten a book on augmented reality and I have a few friends who are working inside Facebook on various efforts. Lately I’ve been visiting with people building Magic Leap and HoloLens and other AR glasses as well.
This gets to what I’ve been working on the past year: building a new theory of the next five years.
What is that thesis?
2019: six degrees of freedom VR for $400 on the VR front. On the AR front (the industry is moving toward calling these “XR” or “Spatial Computing Glasses” or “Mixed Reality Glasses”). Expected, a new version of Microsoft Hololens, which joins Magic Leap. Expected to be for high end users only, since the price of these will still be $3,000 and fairly big.
2020: Facebook enters the spatial computing glasses with something around $1,500.
2021: Apple releases its first spatial computing glasses at about $1,000. Microsoft, Snap, Samsung, Facebook, Google, and others jump in as well.
2022: Second pair of Apple finally fixes them enough that the mass market starts showing up.
Why am I still so bullish on spatial computing? Let’s start with the fact that you get as many virtualized screens around you as you want. Imagine working in an office where you just show up with a little keyboard and your glasses and you get glorious big screens around you. That, alone, will sell many of us on using these glasses. I have a Hololens already and it’s quite compelling to see virtualized screens in front of you. The problem is that Hololens is WAY too heavy and big to wear for long, way too expensive for most to try it, and the optics just aren’t very good so you can’t work as well as on a physical screen yet. I hear all of those problems get fixed by 2021.
This thesis comes from studying the immersive/spatial computing/VR/AR industry very closely for years. You can follow along by watching the best people and brands on this Twitter list I built.
So, this gets me to Thompson’s thesis and question. Why invest in VR today if it’s AR (er, spatial computing, XR, mixed reality glasses) that everyone is going to use?
This brings me to the video. It’s of a new SDK that’s coming from HTC in a few weeks. Most of you haven’t seen it, and wouldn’t know what’s important about it, but the developers I talk to say this technology is HUGELY important to where the glasses are going.
To get computing to properly interact with the real world, you will need many new technologies to be built BEFORE the glasses hit.
The video shows one of them my friend calls “a mesh unfucker.” Technical term, sorry for the harsh language, but that is what my friends call it so I’m going to stay with that term for now.
See, when you get the glasses they will have a sensor, or a few sensors, on the front that will see the world and then overlay a mesh of little polygons on top of that world.
The problem is that sensors will always introduce noise. You see that in the video by inaccuracies in the way the polygons lay on top of the real world. You then see the mesh unfucker do its work. Someone trained it, with deep learning, about the objects it sees there. The chair, for instance, and then the mesh unfucker “locks” the polygons onto the chair, so that the virtual human can properly sit in that chair.
This is really advanced tech that will be improved over time (Carlos Calva is building a company in this space, for instance, and he’s not alone, he’s one of the people who taught me about unfucking the mesh, more on his company when he gets out of stealth mode probably next year as a new Hololens and Oculus Quest arrives). The problem is that developers who are building these new technologies can’t always afford a HoloLens or a Magic Leap and certainly can’t afford 20 of them for a lab. Right now they are $2,500 each, while VR headsets are coming down to $400 for an Oculus Quest next spring).
Also, users can’t afford these headsets. Quick, how many of you bought a HoloLens. I know very few.
But many more users, for now, can afford a VR headset. So, now, think about how brands get built. How companies get built. How communities get there.
Facebook is using VR to “boot up” all of that into the glasses so that when glasses do arrive and do get popular, they will have a ton of cool things to do on them.
This is the fight of the next decade and billions of dollars are being spent.
So, will Apple’s program of secrecy work better to get all sorts of amazing apps built for the AR glasses? Or will Facebook’s more open strategy of working with VR developers get more apps?
I’m betting Facebook’s strategy will and I won’t be shocked if I’m more excited by Facebook’s ecosystem, even though Apple has so many more advantages, from 450 stores to a brand people are fine with putting on their face, to an non-ending marketing budget.
Note that developers are getting major new technologies from the VR ecosystem now, while Apple remains silent. This video is showing something that HTC Vives will have in a few weeks.
Yes, Apple has AR Kit, its own AR on the phone, but phone-based AR just doesn’t get developers and users to dream like putting on a VR headset.
That all said even if Apple wins, all this work and investment that Facebook is doing will pay off big time as it will have a ton of great things to do on the Apple headsets as well and will have all the learning that comes from making its own operating system and headsets (and Facebook is learning a LOT by doing this work) much of which will port nicely into Apple’s headsets as well.
For these big companies it’s about keeping relevant as a paradigm shift hits and this paradigm shift will be the biggest one yet for a whole lot of reasons.
Oh, and as to the future of jobs, well, there are teams coming up with mesh unfuckers as we speak. Imagine trying to explain that one to your family “well, I am training the mesh to reduce noise in it and more accurately get it to wrap to objects in the real world.”
I want that job and am learning as much as I can about how to train systems to do that. Hey, in two years you will want your mesh not to be fucked, no?