This is the opposite direction that we're wanting to go however. We're moving towards rendering less of the view in full quality and eventually with eye tracking we'll only be rendering the fullest quality for the fovea (maybe the size of a quarter held at arm's length.) Moore's Law has a handful of cranks left and then that credit card expires, so any proposed future tech needs to dovetail with optimizations that bring us closer to how the brain receives information. The optic nerve supposedly shuffles no more than ~10mbit/s of data between the retina and the brain, which is already several orders of magnitude below what we're pumping across the HDMI cable to the HMD right now, so we should be looking for technologies that allow us to generate only what the eyes/brain needs and little more than that.
Rift, Vive, and Virtual Reality
- Thread starter idsn6
- Start date
This is the opposite direction that we're wanting to go however. We're moving towards rendering less of the view in full quality and eventually with eye tracking we'll only be rendering the fullest quality for the fovea (maybe the size of a quarter held at arm's length.) Moore's Law has a handful of cranks left and then that credit card expires, so any proposed future tech needs to dovetail with optimizations that bring us closer to how the brain receives information. The optic nerve supposedly shuffles no more than ~10mbit/s of data between the retina and the brain, which is already several orders of magnitude below what we're pumping across the HDMI cable to the HMD right now, so we should be looking for technologies that allow us to generate only what the eyes/brain needs and little more than that.
Yeah, all of those are in the realm of possibility a couple years out. I'd actually rather have had this generation be cheaper "DK3"s and waited for another iteration before trying to go full consumer, but oh well. 2k x 2k per eye would probably get us enough resolution to make virtual cinema a selling point for most people. The antiquating of previous VR hardware is probably going to continue every few years until we're at the point of jacking spikes into the back of our heads. I wonder what the chances are that the HMD bom price will actually go up over the next five years as the core components increase in complexity and number.
Silent_Buddha
Legend
Yes, VR is a best case scenario for foveated rendering as you don't need to worry about multiple viewers and the screen is relatively close to the eye which should help with tracking accuracy.
I'm hoping that will appear in gen 2 or gen 3 devices. I'm wondering how CPU intensive eye tracking would be? I admit not having looked into any research related to that.
And directly responding to the other post. I expect prices for future generations of HMD to increase in price. Whether it's gradual increases or sharp increases will depend on how much and what level of technical advances are included.
If they are smart, they'll continue to sell "dumb" VR (current generation) which will benefit from price reductions making them more accessible to the general consumer. If foveated rendering comes into play, however, they'll need to modify early designs to accommodate foveated rendering or developers will be less inclined to use it.
Regards,
SB
The big issue with the eye tracking is the camera speed/latency, and the speed and low inertia of the eye. Maybe something akin to a gaming mouse optical sensor that tracks a narrow patch of the eye and spits out an X/Y, rather than a high resolution, wide FOV camera spitting full images of the eyeball for the CPU to try and sort out where the pupil is, what shape, etc.
edit:
I guess the nice thing would be that you could use both eyes in conjunction to error correct (since there's a pretty limited range in which they can operate independently, can't diverge beyond parallel, etc.)
edit:
error correct and some sort of early reject perhaps - (quick and dirty, early best-guesses for where the eyes are approximately pointing). All of the efficiency gains from foveated rendering are going to be engine/renderer side anyways, so perhaps if you had sufficient granularity/programmability in the pipeline you could composite the pixels/raster stage to resolve detail progressively. Path-tracing monte carlo style would fit very well with this.
edit:
could maybe use something super fast like a mouse sensor for small motions, and then do your dead reckoning against a typical 60Hz IR camera that tracks the pupil. That'd give you something akin to the 1KHz polling rate that we get from the IMUs for the HMD.
edit:
I guess the nice thing would be that you could use both eyes in conjunction to error correct (since there's a pretty limited range in which they can operate independently, can't diverge beyond parallel, etc.)
edit:
error correct and some sort of early reject perhaps - (quick and dirty, early best-guesses for where the eyes are approximately pointing). All of the efficiency gains from foveated rendering are going to be engine/renderer side anyways, so perhaps if you had sufficient granularity/programmability in the pipeline you could composite the pixels/raster stage to resolve detail progressively. Path-tracing monte carlo style would fit very well with this.
edit:
could maybe use something super fast like a mouse sensor for small motions, and then do your dead reckoning against a typical 60Hz IR camera that tracks the pupil. That'd give you something akin to the 1KHz polling rate that we get from the IMUs for the HMD.
Last edited:
Vr Lens Lab the creators of VR covers has a kickstarter going.
https://www.kickstarter.com/projects/451454651/vr-lens-lab-glasses-for-virtual-reality-headsets
http://vr-lens-lab.com/
If you wear glasses these look really good as a replacement. The kickstarter doesn't give you prescription but you can add the lenses after the kickstarter ( due to kickstarter rules)
I backed for two pair. One for my rift and one for my vive. I am getting the plano ones since I had lasik but will be using them as a screen protector for $33 I figured I couldn't go wrong
https://www.kickstarter.com/projects/451454651/vr-lens-lab-glasses-for-virtual-reality-headsets
http://vr-lens-lab.com/
If you wear glasses these look really good as a replacement. The kickstarter doesn't give you prescription but you can add the lenses after the kickstarter ( due to kickstarter rules)
I backed for two pair. One for my rift and one for my vive. I am getting the plano ones since I had lasik but will be using them as a screen protector for $33 I figured I couldn't go wrong
While it'd be ideal to not waste on anything that's not seen, sometimes it can take more effort to do that than trying to do it the "dumb" way.
Personally I think the future of VR is mobile. The current "must be tethered to something large" is not going to be mainstream. However, for a mobile VR solution to be able to render a convincing world with high frame rate is, IMHO, going to be many years away. It seems to me that it will need to be at least partly cloud assisted. The network latency will become a serious problem and you'll need something that can tolerate higher latency.
For example, you don't really need to render everything locally. Far away objects don't really move much when you move your head. A hybrid system could receive semi-background objects from the cloud and only render nearby objects. That'd make rendering a convincing world with a mobile system more manageable.
cheapchips
Veteran
I think it's likely they'll be hybrid headsets for a while, regardless of what's driving the graphics. Mobile still lacks positional tracking and IPD adjustment.
I'm sure they'll move together eventually but the latter seems insurmountable in phone based headsets?
Once wireless transmission to PCs is cracked you already have a battery in the headset. Might as well chuck a mobile CPU in there for when you're out of the house
Or
You have a phone based headset, might as well chuck a wireless PC connector in there.
Silent_Buddha
Legend
While I tentatively agree with this to some degree, it's not in the way you stated.
While lateral motion of the user's head will not product much virtual movement in distant objects, rotational movement of the head will still product significant movement of distant objects making cloud assisted rendering not a good fit.
Perhaps something like reflections can be rendered in the cloud. There will still be a delay but if you can do ray marched or ray traced reflections in the cloud to be used based on the user's perspective, it might still be better than screen space based reflections.
Foveated rendering is still going to be the best way forward to not only increased perceived image quality but to also reduce GPU rendering overhead.
On the other hand, using non-local cloud resources to assist in physics rendering can help in reducing CPU resources needed while also increasing the belieability of physics reactions within the world. Similar to Crackdown. Immediate physics that impact the player are done locally (the ground under their feet exploding, or bullets hitting them). While physics that aren't of direct impact on the player are rendered in the cloud (a building exploding takes a long time for things to start moving in real life, hence makes a good candidate for remote computation). This leads to not only reduced CPU overhead for physics calculations but also greatly boosts the robustness of the physics simulation.
Regards,
SB
I guess I don't see it as a matter of "smart" engineering philosophy, but of necessity. Unless someone pulls a rabbit out of a hat akin to the invention of CMOS, we may never see modern desktop performance on mobile. We can't confidently assume that we've got more than an order of magnitude (if even that) of performance scaling left, so whatever new use cases, sizable spec increases, size/form factor reductions that come at huge costs in transistor budget or TDP, we need to accommodate those with equal thought paid to efficiency. If something new bumps the perf footprint by 10x, then we have to be looking for an optimization that provides a countering 1/10x. If we're not doing that at this stage, then we may as well be talking fantasy holodecks, the matrix, singularity, etc.
Mobile makes a lot of sense, yep. Economically, mobile VR (particularly phone add-on VR) stands to have far greater penetration due to ease of use and lower cost. What that mobile VR will look like in 5 years however I'm not sure. I suspect that demand for higher quality and much costlier GearVR-equivalents for smartphones will arrive once inside-out tracking is sorted out and if/when the resolution scales up. I think the safest prediction to make is that within 5 years we'll have low cost[relative to Rift/Vive] smartphone VR that's capable of producing an uncompromised Imax equivalent experience for movie viewing, and that will be enough to justify 50-100 million people paying $100-200 for an add-on to their smartphone.
Beyond that... I suspect we're more likely to see a wireless "tethered" (PC/console powered) full VR experience before we see viable all-in-one HMDs. In addition to that I think that the huge and frequent generational leaps in VR hardware will make it incredibly hard for any company to feel comfortable planting a flag and committing to a fully integrated, very high cost HMD platform.
Yeah right now you can absolutely get away with a static (or animated) skybox due to the lack of parallax beyond a certain distance (which is even more limited due to the low effective resolution). Similarly there's normal mapping and parallax mapping that are tied to distance. Where normal maps can be used outside that visible parallax boundary, within that boundary you need to use POM or just plain higher geometry. The wrinkle in this is the fact that as resolution scales up we may find the lack of pixelation vasoline on the lenses will remove the veil on artifacts (not just graphical, but brain/inner-ear/perceptual) we didn't realize were there before. It's an exciting yet shitty spot to be in when your engineering is effectively beyond the leading edge of well understood science (in this case brain science), where you pretty much have to build fully functioning, low tolerance prototypes and then test them to find out what issues lay ahead.
Watching a VR stream right now gave an interesting similar example - the streamer mentioned that he is no longer able to play driving sims in VR with the CV1 or Vive without feeling sick as he could with DK1/DK2. It turns out the experience now has hit a sort of uncanny valley where his brain and body associate the sim experience with his real world driving experience, and the lack of negative Gs during heavy braking produces a severe motion sickness that he's never felt previously.
Last edited:
So they started updating orders.
Here is mine
Order Status
Pre-order
Tracking #
(TBA when your order ships)
Order #
613000029xxxx
Estimated Ship Date
5/16/2016 - 5/26/2016
I am not happy at all. I have a vive coming and unless they give me something good free I will be canceling my rift order. I started trying to order as soon as the preorders went up and I still had a march shipping date when I started my order. Now its estimated for the end of May
Here is mine
Order Status
Pre-order
Tracking #
(TBA when your order ships)
Order #
613000029xxxx
Estimated Ship Date
5/16/2016 - 5/26/2016
I am not happy at all. I have a vive coming and unless they give me something good free I will be canceling my rift order. I started trying to order as soon as the preorders went up and I still had a march shipping date when I started my order. Now its estimated for the end of May
And of course in some cases there's no good news, in which case they don't provide the bad news until the last possible moment. Heh.
Argh. oculus folks really suck with their communication and prediction on shipping things. I got my order in around 6 minute mark. First estimate for shipping was march 28th... Then there was the mysterious component shortage and april 12th deadline. Now I see my order pushed to something happens possibly between 4/25/2016 - 5/5/2016
I wish they would just come out with the reason to the delays. Good thing there is still possibility to cancel. If vive becomes available I might as well jump the ships and abandon the idea of owning a rift.
I wish they would just come out with the reason to the delays. Good thing there is still possibility to cancel. If vive becomes available I might as well jump the ships and abandon the idea of owning a rift.
I have a vive coming also.
However if you watch the video they recommend the vive but they say that while using the vive they wish they could put the rift on and use the vive motion controllers.
You don't even need an HMD to use the Vive controllers: http://i.imgur.com/LCjFuUy.gifv
They both need to get their stuff out sooner than later. Before the 2nd generation of headsets hits (PSVR, Oculus 2, HTC Dive), because people can change ships then, if they've bought into the proprietary stores, then changing headsets will not be encouraged.
Now: HTC Vive is less comfortable with worse image quality than Oculus: people buy Oculus. But if they have the option to buy HTC2 in a few months, if it has improved optics and comfort, they would still choose HTC (if both were sold out anyway)
Now: HTC Vive is less comfortable with worse image quality than Oculus: people buy Oculus. But if they have the option to buy HTC2 in a few months, if it has improved optics and comfort, they would still choose HTC (if both were sold out anyway)
We're not going to see any second generation HMDs for several years, and the if/when of those HMDs is going to be dependent on how well these ones fare in hardware and software sales.
Until we have concrete sales numbers from Oculus or HTC, it's going to be very hard to make any early projections about VR hardware's trajectory on the PC. Personally I think the biggest concern right now isn't Rift vs Vive, but rather the health of indie studios and their launch title sales amidst a very very gradual hardware roll out. Launch content on new platforms/consoles generally make their money not by virtue of their quality, but by having a captive audience with very limited selection. If the bulk of Rift buyers end up spending their 2-3 month wait in the pre-order queue reading reviews and watching live streams, then these games are not going to hold up well when it comes time for people to buy them. I can already say for myself that there are several games that I may not buy that I would have otherwise had I received my HMD by now. More damning than that will be the multiplayer-only titles that live or die on having an active population of players.
Until we have concrete sales numbers from Oculus or HTC, it's going to be very hard to make any early projections about VR hardware's trajectory on the PC. Personally I think the biggest concern right now isn't Rift vs Vive, but rather the health of indie studios and their launch title sales amidst a very very gradual hardware roll out. Launch content on new platforms/consoles generally make their money not by virtue of their quality, but by having a captive audience with very limited selection. If the bulk of Rift buyers end up spending their 2-3 month wait in the pre-order queue reading reviews and watching live streams, then these games are not going to hold up well when it comes time for people to buy them. I can already say for myself that there are several games that I may not buy that I would have otherwise had I received my HMD by now. More damning than that will be the multiplayer-only titles that live or die on having an active population of players.
If it makes you guys feel any better, this is mine:
Estimated Ship Date
7/18/2016 - 7/28/2016
And I only pre-ordered on the second day! While I'm super eager to get my hands on a rift, my PC isn't ready yet and I need enough time for the next gen parts to launch and hit stable prices before I can re-build so I'm not over concerned about this date. In fact there still won't be anything particularly interesting from Intel by this point but I really do need to upgrade if only for the USB3 ports.
Also, my Birthday is on the 20th of July so hopefully that's gonna be one happy birthday for me!
Incidentally, what is this about them covering delivery charges? Are they saying they will no longer charge us for that now? The email isn't too clear. i.e:
We'll also cover all shipping and handling costs for orders placed through 11:59 PDT on April 1, 2016.
