As with my visit to Oculus Connect last September, hands-on time with the Crescent Bay prototype of the Oculus rift and a chat with Oculus VR's VP of Product offered some new insights into the state of virtual reality tech and the challenges for the upcoming consumer version release. No, we didn't find out when CV1 would be available, nor could Nate Mitchell spill the beans on the exact technical specifications of even this prototype. But our conversation--and the way he answered some of my questions--allows us to infer some details about curious topics like screen resolution, optics, and virtual reality input. Here are eight things that I took away from this chat, with the full video interview below.
Downplaying Screen Resolution
Everyone is curious about what the screen resolution of the consumer release will be, with VR enthusiasts hoping for a display as high-res and dense as possible. Development Kit 2 uses a 1080p AMOLED display, but the pentile subpixel configuration and screen door effect (SDE) is still noticeable. These effects are reduced in Crescent Bay, and the prevailing thought is that the prototype must use a 1440p or better screen, given Gear VR's use of the Samsung Note 4. I'm not 100% sure that's the case, now. We knew that due to the constraints of mobile graphics bandwidth, Gear VR renders its games and software at lower than native resolution (eg. 720p) and then scales up to 1440p. Nate said that on the PC side, the Oculus Rift--and Crescent Bay--does the opposite: it will use supersampling and render software at a higher than native res and then downscale it to fit the screen. It's like what Apple does with the iOS for its iPhone 6 Plus.
So here's a thought: maybe Crescent Bay actually uses a 1080p display, and the single GTX 980 card renders some demos at a higher than 1080p resolution and downsamples to fit into that screen. Supersampling would reduce aliasing, and more demanding software could still just be rendered at 1080p to guarantee 90Hz output. This theory would explain why Oculus reps (including Nate) have mentioned the 1080p resolution when discussing PC system performance in other interviews, and why some press have reported that Crescent Bay runs at a lower res than Gear VR. I also noticed that the mirrored game window on the 27-inch panel connected to the PC was not filling the full monitor--it looked about 1080p to me. Of course, this doesn't mean than CV1 won't use a 1440p display, but Oculus is seems to want to downplay talk of screen resolution as they recognize its tradeoffs with performance. Don't expect CV1 to run a 4K screen.
Text Readability will be a Challenge
Alongside discussion of screen resolution is the acknowledgement that text rendering will be a challenge for Oculus Rift. Anyone who's tried to play a game like Elite: Dangerous with the Development Kit 2 knows that reading HUD or on-screen text is difficult, especially in the peripheral. If Crescent Bay is indeed running on a supersampled 1080p screen, it would explain why the demos were almost devoid of any text. I looked for text in the demo sequences, and could only find examples in a distant billboard shown in the skyscraper demo the loading screens. Nate told us that text readability could be solved with different or larger typefaces, but that assumes that developers will be taking those guidelines into consideration when making games for both high-res desktop screens and VR.
The FOV Target is Around 110 Degrees
This is something we've known for a little while--field of view is one of the benchmarks Oculus CEO Brendan Iribe talked about during his Connect keynote--but it's good to hear a confirmation that the target FOV for CV1 will be around 110 degrees. Nate confirmed that Crescent Bay and DK2 have around the same FOV--both less than that of DK1. FOV is partially gated by the size of the display panel used (7-inches in DK1 and 5.5-inches in DK2) and the optics. Again, there are tradeoffs--a larger panel is more susceptible to SDE, and a wider FOV requires more rendering on the PC side (assuming more pixels).
Optics Remain Key: Pixel Smearing Confirmed
I don't think enough attention is paid to the optical improvements made in Crescent Bay over DK2. The lenses used in this prototype (which don't have IPD or focal adjustments) do a lot to help make the screen image look better than previous versions. SDE is almost completely eliminated though something Nate called "pixel smearing"--which may be another name for the diffusion effect that some Crescent Bay users have described. Instead of seeing a grid of pixels, I notice almost a "linen" texture to the image--a pattern of very thin lines that have an almost fluid orientation. An analogy t describe it: instead of the game image looking like it was printed on graph paper, it looks like it as printed on high thread-count bed sheets.
IPD adjustment is something we're told Oculus is still tinkering with, so you don't have to worry there.
The Ergonomics are Close
What will CV1 look like? Probably not like Crescent Bay, but we do know that it'll be similar in terms of how the headset is work. The headstrap and backplate design in CB is something Oculus seems happy with, especially since they have said that CB is a good weight for what they want to get to in CV1. It's a LOT lighter than DK2, and the balance feels good--much better than the understandably front-heavy design of Gear VR. My feeling that Oculus has locked in the strap and balance design for CV1, and that unseen prototypes are now figuring out the physical face interface--how the goggles actually press against your face. Other HMD makers like Avegant are designing swappable nose bridges for different users, so maybe that's one solution. CB also didn't work work over my large-ish glasses, and Oculus has said that they want CV1 to be able to be work over glasses.
Wireless HDMI isn't Practical
With the Crescent Bay demo being a standing experience, I was curious whether Oculus was considering a wireless solution for CV1. Nate quickly dismissed the idea, confidently saying that wireless HDMI just isn't practical today for consumer VR, both for economic and performance costs. It's not something they're actively tinkering with for CV1.
Input Input Input
I think this was the big--if unintentional--reveal at CES. After Brendan Iribe mentioned to TheVerge that an input solution for the Rift was in the works, we had to pick Nate's brain on it. He didn't specify what kind of input devices they may be working on--whether it's hand tracking or some kind of gamepad--but the way he described the gamepad experience today leads me to believe that it won't be something like Valve's Steam Controller. Gamepads are "a great start for today" and allows "good" gameplay experiences (emphases mine), but the smile he gave when he said that a different input solution "may come sooner" makes me think that there will be something to see before CV1 is released. As to that that input model is? We can only guess, but note that the first example that Nate listed when we talked about the iteration of VR UI was gesture recognition--with both his hands held up. Take from that what you will.
Any input solution will need to get to the hands of developers if it's to make the CV1 release, so my bet is that we'll see a dev kit for input at GDC--one year after the surprise announcement of DK2. This is super exciting!
Consumer Version 2 May be Iterative Release
The last question I asked Nate was actually less about Consumer Version 2 (because even thinking about that is silly at this point) and more about figuring out Oculus' plans for product segmentation. We figure that CV1 will be the best that Oculus has to offer--there's only one opportunity to make a first impression with the public--but consider that by the time it's released it'll already be dated technology, from a R&D standpoint. Oculus iterates fast, and my fear is that the userbase will become segmented in the year or two after release if subsequent SKUs offer new functionality critical to the experience. There are rumors about multiple versions of the Oculus CV (eg. a low cost entry model or one not specifically for games), and I don't know how often users will be expected to upgrade. Early adopters have already picked up two developer kits at this point, and CV1 would make a third hardware purchase. My hope is that the strategy for VR products is along the line of Apple's annual release of iPhones--a iterative update one year followed by a big leap the next. The Oculus experience should be more about just the hardware, and they're in a good position to create a strong hardware and software bond to differentiate the Rift from the impending flood of VR devices.