The State of Consumer Virtual Reality in Q1'2017 — an early adopter perspective
The consumer releases of Oculus Rift and HTC Vive have been available for less than a year, and I’m “that guy” — the enthusiastic early adopter who has picked up a headset and is eagerly demoing it to everyone who’s willing to come by and strap on the head-mounted display. While I have tried to explain and describe VR with accessible concepts, I sense a collective feeling of meh due to a variety of reasons including, but not limited to:
- too expensive, too hard to set up a VR PC (and aren’t we in a post-PC era?)
- the awkwardness of strapping on a tethered headset
- no killer app
- experiences are not shared and social
Clearly, if tech-savvy Bay Area techies cannot be bothered to spend a few hours in VR, VR has an accessibility and marketing problem. In some ways, this is worse than the 2007–2009 era of smartphone skepticism, during which many consumers late-adopted because the phones were either 1) not available on their carrier or 2) out of their price range. Time solved the carrier and cost problems in a couple of years, but VR has much more to overcome. I am a believer, but I recognize there is a chance that VR is yet another consumer technology that is destined to fail.
So, what specifically needs to improve? The following analysis is how I’d want the industry to evolve.
Higher resolution, please.
1080p is not enough!
When you strap on a VR headset, the hope is that you can sink into the content quickly and be immersed in the storytelling. With this current generation of hardware, this is not possible because the headsets and the GPUs that power them only support up to 1080p resolution per eye.
Both the Rift and Vive goggles are spec’d at 2160 horizontal x 1200 vertical resolution (2.6M pixels). When you strap on the headset, the center divide cuts that display in half, so you get 1080 x 1200 per eye. (Note: in the Vive and Rift, the single eye VR viewports are taller than they are wide, so the 1080 pixels is actually the horizontal resolution, unlike your desktop monitor)
You’d think that if you held up a 1080p desktop display or your phone in front of your face, that image ought to be crisp enough, but in VR, the viewport needs to account for your frontal vision as well as your peripheral vision, which means the 1080 x 1200 pixels is “stretching” to cover more of your field of view than a 2D screen in front of you.
The effect of this is pixelation of your video content comparable to the days of pre-HD videos of 320p or 480p resolution that older millennials may recall. Basically, it’s like trying to watch online video with RealPlayer in 2001, when you might go for a coffee break before returning to hit ▶️. *shudder*
For publishers, this means that publishing your content in 1080p is not enough. We need to get to 2x, 4x or even 8x more resolution density in the coming years before viewers can be convinced that there are no pixels in front of their eyeballs.
So, what does it take to support 2, 4, or 8 times the resolution of the current generation goggles? Growth in GPU performance, and lots of video RAM!
How much? Let’s avoid the overloaded definitions of marketing-speak “4K”, “8K” definitions in the following calculations, and simply do naive napkin math for “doubling the resolution” over the course of two generational iterations:
- twice the current resolution in each direction is 4320 x 2400 (4x total pixels of current gen, or 10.4M pixels per frame)
- double again in each direction is 8640 x 4800 resolution (16x total pixels of current gen, or 41.5M pixels per frame)
Assume 24-bit “true color”, that’s 3 bytes to represent each pixel. So respectively, the per-frame data is 31MB and 124MB.
Assume 90 frame per second, which is the refresh threshold at which your body does not vomit with motion sickness. The naive memory requirements are respectively 2.8GB and 11.2GB per second of content. (in practice, programs will optimize/cheat this in many ways)
Assume the GPU is calculating some N frames ahead, you can do your own napkin math for how much video RAM is needed to accommodate this future. (The above estimations do not even take into account the need for games to actually calculate and draw the content in each frame!)
So, ballpark estimation, we’re not getting to the holy grail of retina VR displays until GPUs can support 4x then 16x the pixel count of today’s goggles, and there is enough video RAM to buffer the uber high resolution content.
I am not a GPU engineer or industry expert, so I’ll just hand-wave the rate of improvement as doubling in performance every calendar year, that puts the 4x hardware 2 years away (2019), and the 16x hardware 4 years away (2021) in my most wildly optimistic scenario. Of course, I doubt Nvidia and AMD are iterating this quickly, so this may be a ten-year journey!
Faster internet, please.
Home broadband is not “broad” enough for VR streaming!
In the Oculus app store, several apps provide a YouTube-like browsing experience of early VR video content. (Hulu, JauntVR, Within to name a few) When you attempt to watch the compressed 1080p-ish videos within, you constantly encounter download buffering and the artifacts of compression, both of which are visual defects that take the user out of immersion.
I have a gigabit urban high-speed internet connection, my 1080p YouTube videos do not buffer, but on the same network, almost all streaming VR content does, because each piece of VR content is gigabytes in size. Simply, broadband is not broad enough to stream real-time HD VR video.
With the same napkin math as above, streaming VR content will require our internet service providers to 4x, 8x and 16x their throughput. Good luck with that, maybe we’ll get it in 2025?
Or, pray that Middle-Out Compression leaps from TV to reality. 😂
In the meantime, maybe content delivery can settle on several possible workarounds:
- iTunes Podcasts-like asynchronous subscribe-and-download model where the content arrives in your VR inbox in its leisurely pace
- IRL VR theaters/arcades that can play pre-loaded content
- Netflix-DVD reborn: discs in the mail! The post office once again becomes a pragmatic way for you to get gobs of bytes in a timely and reliable manner
Native content, please.
Most early content is not imaginative enough.
When the iOS app store first opened, most entertainment apps and games in the store were just lazy ports of apps that started as 2D content. It took a few months before games that took full advantage of the touch controls emerged. (My personal obsessions were Frenzic and Flight Control.) This breakthrough is defined by one key attribute: the controls and mechanics of play simply do not make sense or is outright impossible in its previous medium: 2d display with mouse and keyboard input.
VR entertainment is experiencing the same early scarcity-of-imagination problem. So far, the only game that echoes my excitement of Frenzic and Flight Control is The Climb. It’s a game that simply does not make sense in the previous medium (2D gaming of any input), but in VR — it is self-evident and transcendental.
Down with gimmicky demos, toys, and concepts!
The online press covering VR is thirsty for stories, so they will promote gimmicky VR apps and toys that are gimmicky but ultimately useless beyond demos.
The most guilty category is “VR exercise”, which tries to duplicate real world experience, but somehow manage to provide a worse experience that literally no one I know would want.
- This man is cycling around the UK in virtual reality using Google Street View
- VRZoom — the whole company/concept
Let me explain: current generation goggles are not very “airy”, which means any moderate amount of movement will make the average person’s facial warmth and perspiration sweat/steam into the headset, fogging up the lenses. I’m relatively fit, and I’ve had this happen sitting down after taking a brisk walk around the house.
Any exercise app in VR, no matter how moderate, has an audience of 3: you, your obligated BFF/SO, and the thirsty writer who needs a story. Just stop it until we get light-weight, untethered headsets. VR goggles in 2017 are not ready for reenacting battle scenes from The Matrix or a simulating Tour de France. Developers, you ought to focus your energy on developing experiences that works better in VR than in the real world.
Dishonorable mention: selling stuff assuming your target user is a teenager from the 90s: http://uploadvr.com/cyberpowerpc-amd-499-rift/. You want this VR thing to go mainstream or what? Then make it blend into the home as discreetly as possible, for adults.
Better sharing, please.
I still have not figured out how I can share a recording of my VR experiences with anyone, in the same way that screen recordings and video games can be shared on Twitch and YouTube.
You can find picture-in-picture videos on YouTube of players sharing a) a 2d projection of their VR screen and b) video of their body and movements. They all seem like laborious, manual productions beyond the motivation of casual intents to share. What we need is for Oculus and Vive to integrate with PC cameras and provide an export functionality of play sessions. You should be able to watch these VR replays on cheaper headsets such as Samsung Gear or any cheap Google Cardboard.
Epic demos gets the users. Please make this happen.
So, as of January 2017, I still think VR is gonna be the next awesome medium and would recommend it to any early adopter who wants to live in the near future of their peers. However, if you don’t want to be VR buddies just yet, I get it, I’ll catch you in The Matrix in 2020.