>>47691752 Anything that is well reviewed when it comes out. Other than that, I am expecting the Gear VR 3 non-innovator edition to be quite good, in addition to CV1 and Vive. I'm also interested in seeing where OSVR goes, especially now that Wearality might partner with them to provide high FoV lenses.
>>47692223 I'm not worrying too much about resolution to be honest. We've already seen that the Gear VR 2 has a near perfect resolution that can satisfy nearly everyone. The consumer products, probably from any company that comes out in winter, or 2016, will have better displays anyway.
>>47691752 Google cardboard. Actually, better software for it.
Yes, we actually have all the shit we need for VR glasses in smartphones already.
Gyro, compass, camera in front and back (it could fucking do eye tracking and at least a 2D overlay of the real world... if 3d world needs to be a thing, I guess we'll get 2 cameras on smartphones), high resolution displays, gps with fucking magnet sensors, computing power, a sound card, and what not.
I don't really need another expensive device with much the same stuff *again*, and much fewer choices to pick from.
>>47692289 You seem to forget that VR requires a lot of horsepower if you want any high fidelity scenes. Of course, eventually mobile VR will dominate, but currently it's shit, and no, software alone will not fix the problem. In designing a VR HMD, you have to take into consideration the design of the whole system. Phones that weren't made specifically with the intention of being used as a VR HMD from the very beginning will most likely not allow you to take advantage of the sensors and cameras in the way you're describing. That's why the Gear VR is currently an add-on that has it's own sensors, in addition to correctly matched lenses and distortion. Cardboard will improve, and will eventually achieve that without needing additional outside components communicating with the phone, but that's at least a year away until phone companies decide to push for VR.
>>47692417 >You seem to forget that VR requires a lot of horsepower if you want any high fidelity scenes. I don't. I'm just not really expecting that to be the thing that matters.
We'll just progress our VR at the speed of pretty pedestrian vidya game GPU - mobile or desktop.
It's not like those don't leave any options.
> will most likely not allow you to take advantage of the sensors and cameras in the way you're describing. But they will. Some have sensors that are too crappy, but actually many can pretty accurately and rapidly tell which way they are oriented in 3D space.
And it's not like you'll doubt my statement that the front-facing camera could handle at least just doing a spherical 2D projection upon which you can overlay a bunch of 3D VR already, right?
>That's why the Gear VR is currently an add-on that has it's own sensors, in addition to correctly matched lenses and distortion. ... the correctly matched lenses (for phones of +- the common sizes around 4.5-5.5") are a thing with cardboard already.
> but that's at least a year away until phone companies decide to push for VR Again, cardboard's software isn't good enough yet, that's the primary problem I have with it right now.
Not my phone's sensors. Not the lenses. Those work.
My $140 phone's sensors seem to update at least 5-10 times per second (just looking at cpu-z, I can't seem to directly get a number for the actual rate at which the sensors work), and they seem reasonably accurate.
'tis not like I can comfortably move my head a lot more anyways.
Maybe they wouldn't be good enough for a game where you want to play no-scope long range sniper or perhaps when you need to do something accurately while jumping?
But overall, I kinda don't get your concerns for now.
>>47692727 To compare, the Rift's gyros and accelerometers update at 1000 Hz, and even that is something they'd like to improve in the future. 5-10 Hz is simply inadequate for VR, the tracking won't be good enough to fool your brain.
>>47691752 >Gear VR may be limited in content but it is still fucking amazing
Daily reminder that Samsung pays people to shill on forums.
>Samsung paid 500 fake fans to attend Chinese press event for the launch of its new flagship phones http://www.phonearena.com/news/Report-Samsung-paid-500-fake-fans-to-attend-Chinese-press-event-for-the-launch-of-its-new-flagship-phones_id67937
>Samsung fined $340,000 for faking online comments
>After being caught paying for false praise and negative comments about competitors, Samsung has been fined just over $340,000. The issue first arose internationally in April, when Taiwan's Fair Trade Commission (FTC) announced it was opening an investigation into the allegations. That investigation found the allegations were true: the FTC says Samsung used a "large number of hired writers and designated employees" to post in Taiwanese forums. The commission does add that the company did this through a third-party marketing company, just as Samsung originally claimed. Two local marketing firms were fined a combined total of over $100,000 for their part in the marketing ploy.
>>47692569 We've observed that graphical fidelity doesn't equate to necessarily realistic scenes, but to say it won't matter is pretty naive. Obviously, everything will improve, but it's really not here yet.
>But they will... 3D VR already, right? They will in the future, but it's uncertain when that will happen. If you want eye tracking, you need a way of making the camera look through the lenses, which would be a hack at best with today's phones and an external holder. The overlays and pass through is possible already and has been demonstrated but it's not really a VR experience.
>correctly matched lenses... area a thing with cardboard already. No, no they are not. You could literally slap on DK1 or DK2 lenses and it would look better, as long as your phone is around 5.6" and the software uses the correct barrel distortion + chromatic aberration. Future versions of Google Cardboard's API will allow this to happen easily though.
Also, it's not cardboard's software that's the problem. It's a general problem with designing phones that work with the OS to give you low-level access to the phone's sensors and hardware, which all have to meet a required spec in order for it to be good for VR.
>>47692774 Wait, since when are we even able to recognize stuff that happens in 1/1000th of a second?
Or actually, how noticeably can your muscles alter the direction your head is moving in 1/1000th of a second?
And again. I'll state that I do not find 5-10hz unacceptable at all, simply based on how much you can move your head in 1 second it's probably pretty much good enough for everything but games and situations where you move a lot. Like, running, jumping, ducking. The latter of which, I can already tell you, are not going to be huge for entertainment, even after VR hits.
>>47692850 You need to educate yourself on VR and the neuroscience involved. I recommend reading the Oculus Best Practices Guide. You need to get motion to photon latency down to below 20 ms in order for there to be no disconnect between what you do and what your brain interprets. VR is arguably about much more than entertainment that only involves small trivial movements. It would be pretty sad if it was. Also, have you even tried better VR than Cardboard? Have you actually achieved presence before?
>>47692842 >but to say it won't matter is pretty naive. Obviously, everything will improve, but it's really not here yet. No really, we don't care for our current games, and for "practical" VR such as simple navigation or assistance on how exactly to open up a machine and do maintenance on it or all that stuff, it really really won't matter already - you can draw far more stuff than you need to elegantly visually express everything accurately.
> If you want eye tracking, you need a way of making the camera look through the lenses, which would be a hack at best with today's phones and an external holder. Hm, you're probably right about that one, but if it is a thing the cameras shipped on phones will be adapted for it in like half a year. So they get fish lenses or whatever is necessary, and a precise mounting point for another one.
Not a big deal.
> The overlays and pass through is possible already and has been demonstrated but it's not really a VR experience. It's not the nicest VR experience, but it works for many things. People won't be ready with much software before the camera standards for a "cardboard" + smartphone type solution are ready anyhow.
> Future versions of Google Cardboard's API will allow this to happen easily though. All right, the software part isn't good enough yet and the lenses aren't high quality. Won't mind if they do both better, and I mentioned I'd like better software already a few times, eh.
>>47692859 That's some heavy bias there. What happens when you wear a pair of goggles where you can't see the display, or there is none, and when you wear it, it transports you to another place? It's going to get there. You'd be surprised what strapping a screen to your face can do, because the current VR HMDs are not actually just screens strapped to your face.
>>47692842 > designing phones that work with the OS to give you low-level access to the phone's sensors and hardware ... cpu-z can access it on Android, so I guess that's done already? Not that I see the problem anyways, it's Linux we're talking about.
> which all have to meet a required spec in order for it to be good for VR Just take the values the sensor gives you and use them. People will decide if they need a phone with better sensors, add-on sensors or whatever.
I bet most won't need anything extra really soon... or anything extra for VR software as-is, even.
>>47692926 Exactly, all these things will come in the following year or two, but it's still a matter of time, and when VR is literally still starting out, we need it to be as good as it can get, with trade-offs in other places as needed. But you still seem to underestimate how compelling moving applications are. So far, the most compelling first demos are ones where you interact and move around. After the public has been sufficiently wowed by VR, we can introduce easier to use applications, which will already be there.
>>47692850 It's not just about the raw frequency and how small changes we can perceive. Something very important to VR is latency, and to improve latency VR headset drivers predict things like head direction and position ahead of time so that the apparent latency is near 0 instead of ~25ms that is usual for motion to photon latency. But to do this they need extremely high sensor refresh rates, far higher than than the actual refresh rate of the screen, because prediction data derived from samples of which most are already several frames old, is worthless.
>And again. I'll state that I do not find 5-10hz unacceptable at all
Have you tried an Oculus rift DK2 or newer, so that you have anything to compare it with?
>>47692965 That's not low-level access. You simply read the sensor data. You need access to low level processes in order to pull off low latency. But with linux you do get to have low level access to things like the frame buffer. It's trivial problem in the coming times with updates to the Android OS. We'll see how phone manufacturers account for minimum specs for VR.
>>47693036 This. Someone who has not achieved even morsels of "presence" can not be talking about what should or should not be good for VR. Presence is what VR is mostly about. Otherwise we wouldn't need VR. We could just settle with immersive 3D displays and call it a day, but we know how well that turned out.
>>47693012 Hm. Maybe. I was never opposed to photorealistic vidya and porn and stuff either, but really, it doesn't seem *necessary* overall.
VR that looks about like normal 2D games would on either platform is just fine and "compelling" enough. Pretty sure that's exactly what's realistically going to happen, too. It's not like many people will start building compute clusters to have their VR better.
>>47693036 > Something very important to VR is latency, and to improve latency VR headset drivers predict things like head direction and position ahead of time so that the apparent latency is near 0 instead of ~25ms that is usual for motion to photon latency. I'm having trouble believing you that you really need remotely near 1000hz worth of accurate(-ish?) sensor data for that.
I realize the 5-10hz that I have access to isn't perfect yet, but 1000hz seems far away...
>Have you tried an Oculus rift DK2 or newer, so that you have anything to compare it with? No, I have tried cardboard. I'm sure these toys are a bit better in a bunch of ways, but cardboard seemed okay enough for the "useful" kind of VR and slower games - maybe not for very precise reaction games, though.
>>47693166 >1000hz seems far away What are you even talking about? We already get that with good IMUs these days.
>VR and slower games >only tried cardboard It's funny because you have no idea what you're missing, and anyone who has tried both Cardboard and the DK2 or better will tell you cardboard is not as good or close to achieving presence as better VR.
You also need to educate yourself on how low latency VR works. High sensor refreshes add so much to it. Without it, you can't have good asynchronous timewarp, prediction, and positional tracking (because positional tracking works by using IMU data in conjunction with correction and fusion by absolute positional tracking solutions). Please go research what prediction and asynchronous time work are, and how they work.
And also educate yourself on what presence is, although I doubt your current mindset would allow you to take that information for granted without experiencing it for yourself.
>>47693166 >I'm having trouble believing you that you really need remotely near 1000hz worth of accurate(-ish?) sensor data for that.
How so? 1000 Hz is 1 ms each sample, 13 samples per frame on the DK2, it's not like that is some huge overkill to work with. To accurately derive a sampled function for future predictions, you need as many recent samples as possible, each one is precious. In fact even 1000Hz is not enough to predict the most sudden direction changes, only relatively gradual changes.
>>47693306 When are people going to use their phones other than when making calls? When are people going to use their gaming consoles/handhelds other than in planes anyway, right? People couldn't possibly use things outside of one specific situation right? Fuck your bait.
>>47693306 Many people who have one seem to use it for watching movies instead of on their TVs or computers at home. I have watched a few movies on my DK2 and it's good, just need a little more resolution, and without actually having tried one I assume the 1440p gearvr is enough to make it objectively better than TVs/monitors for movies.
>>47693259 >What are you even talking about? We already get that with good IMUs these days. No. It seems far away from a necessary spec.
Not far away from technically possible.
> educate yourself on what presence is Looking through the internets, science seems quite absent on that. It seems to be more like a marketing term, though Valve apparently did an evaluation (doesn't seem like that can be called a scientific study, though).
Nowhere is anything to show you need anywhere near 1000hz.
>>47693289 None of this is relevant to anywhere near proving that 10hz is very far off the required tracking rate for head orientation.
Intuitively, I think it's not even as much as the visual display rate that needs to be exactly tracked for it to be realistic. It's not like we alter the movement of our heads much between frames on, say, a 60fps frame rate.
>>47692938 No, its not going to 'transport' you anywhere, you will be sitting in your dumb cheat with headphones on and a screen on your face. You won't have any sort of actual physical interaction with anything, you will be sitting there using your mouse and keyboard
>>47693418 >None of this is relevant to anywhere near proving that 10hz is very far off the required tracking rate for head orientation.
>Intuitively, I think it's not even as much as the visual display rate that needs to be exactly tracked for it to be realistic. It's not like we alter the movement of our heads much between frames on, say, a 60fps frame rate.
Well sorry, but your intuition is wrong. Don't take it from me, listen to the pros: http://blogs.valvesoftware.com/abrash/latency-the-sine-qua-non-of-ar-and-vr/
>>47693418 >far away from a necessary spec That's because you have no idea what's necessary for good VR. Your cardboard in it's present form is hardly VR.
>science seems quite absent That's because it's an emerging technology and old implementations were shit no one bothered? There is actually one study I know about done by a UCLA researcher about it, where they have determined "place cells" in your brain that activate based on your sensory input.
>Nowhere is anthing to show you need anywhere near 1000hz. But there's a lot. You just haven't looked deep enough because searching for these things is kind of hard at the moment. You kind of have to know about the whole pipeline, and that takes some time to learn. Go look up and watch the various 1+ hours talks about these things.
>>47693458 No you don't, stop acting like you do. Keyboard and mouse is so behind from the innovations being developed today. It's not even funny how archaic it seems now. So is the display strapped to your face meme.
>>47693526 What innovations? Moving your hands around in front of you? That has almost no real world applications for anything other than tech demos. NOTHING here is going to 'transport' you anywhere else, your dumb PR statements have no effect on me because I have actually used one of these stupid things.
>>47693471 >of course it does. i think you're overestimating how willing people will be to have something strapped to their face in public, passthrough camera or not.
Why use it in public? VR is about immersing you in some place else, so of course it's not something you'd want to do publicly unless you really want to be immersed in some world. You're overestimating how much people want to do things, not just in VR, not in public.
>>47693548 >I have actually used one of these stupid things. Which one? STEM? Lighthouse? Haptic Grip? Dexmo? I doubt you've even tried one of those. If you're judging based on some year old hardware with limited implementation, then you're biased into thinking that's the best we can do, when technology is advancing at a rapid pace.
In addition, the prototype technology today has varying success in achieving presence for people because everyone has different eyes and structure that can actually impede the effect if you don't do REAL calibration, which currently is too limited on the prototype hardware available for people to demo.
>>47693495 >Well sorry, but your intuition is wrong. Don't take it from me, listen to the pros: http://blogs.valvesoftware.com/abrash/latency-the-sine-qua-non-of-ar-and-vr/ Sorry, but where does this say 1000hz is required in any shape or form?
They pretty much say that the real obstacle to getting their preferred ~60hz display rate with as little as possible in terms of delay is video rendering.
The sensors they have now that can give them an accurate reading have 4ms latency after polling them (would be enough if they determined their position at 60hz if they just finished right before polled, though).
Actually, that's also just what Valve wants. I cannot see where have they studied with any finality how much they actually need for that sensor, only that:
> you are capable of moving your head at hundreds of degrees/second
Which presumably means full-body movement, not head-only movement.
Honestly, I don't even care if VR isn't full body movement for now, but even full body movement won't apparently require more than like 60hz, as far as I can tell from what they're looking at.
>>47693727 No part of these dumb VR gimmicks involve an input solution anyone would would use. For any sort of computing environment, you will use a mouse/keyboard, and for gaming you will use a mouse/kb as well and maybe a controller.
>>47693709 90-110 degree FoV is the sweet spot in terms of hardware performance we can squeeze out of today's systems, but we can make trade-offs if we want higher FoV. Look at Wearality. Look at Infiniteye.
Sorry, I have to go now. I would love to continue this discussion actually, but I doubt this thread will be alive till then. Can you give me some way to contact you guys? Throwaway e-mail or something is fine.
>>47693741 >They pretty much say that the real obstacle to getting their preferred ~60hz display rate with as little as possible in terms of delay is video rendering. PS: It's more specifically the ~36ms delay.
Of which there is only *4ms* delay on their actual current sensor of choice's readout - the rest is on video rendering.
They actually poll it at 60hz from all I can tell there.
>>47693709 looking 'at' a screen and 'through' goggles are two different things. the point of having one lens for each eye looking at a separate view is that it's more like looking through goggles than at a screen.
>>47693810 And that's why you need to improve everything for better VR. But that won't happen when you don't even start with the shittiest first implementation. Would you rather wait 10 or 20 more years for something we can get now in 1 or two years, just because the VR and R&D never started or became popular?
>>47693823 I haven't followed the link or the current discussion really, so I might be wrong. Valve's lighthouse tracking system if I remember right does work at 60 Hz, just like the DK2's camera. The real trick is to do sensor fusion with the IMUs to get positional tracking. Actually, all the camera/lighthouse scanning systems should do is correct for drift that comes from deriving position from acceleration using the IMU data.
>>47693741 No, you're wrong, 60Hz sensors would not be able to do any prediction at all because of too few samples in a meaningful recent timeframe. Not to mention 5-10Hz, that is not even enough to do one sample per frame, meaning you'd need interpolation which increases latency over the baseline 25 ms and horribly affect accuracy as well.
More here, on why they decided they needed the 1000Hz IMU: https://www.oculus.com/blog/building-a-sensor-for-low-latency-vr/
And by the way, even the refresh rate of 60Hz is bad, which is another reason phone VR is bad. Oculus have said that at least 90 Hz is needed for good VR.
>Which presumably means full-body movement, not head-only movement.
I don't know what you mean by this, "moving your head at hundreds of degrees/second" is moving your HEAD at hundreds of degrees/second, what does the body have to do with this?
VR is going to ruin so many peoples lives, all the rehab centers, all the sickly people in their rooms curled up virtualing away in their dream worlds. As one life is ruined another grows considerably in another... it's interesting to see this all unfold.
>>47693946 By the time VR ruins people's lives, do you think the government would stay still?
But it's funny you mention rehab centers and sickly people. VR might be a savior to some people who are currently stuck in hospitals or in facilities because of some disability. For example, look at all the old people just staring out the window and waiting for their death in retirement homes where they have no choice but to be in for some reason or another, mostly due to the defects that come with age. With VR, they can have a lot more things to do, and have a much more meaningful elderly life.
Every piece of technology created has consequences, i'm just pointing it out which offends you for some reason. For example, look at video games, particularly World of Warcraft where people have died playing it from exhaustion.
>>47694008 But don't you guys want to keep this going, or are you people really that satisfied with being stubborn in the things you discuss? Why even have meaningful and long discussions then, if you don't want to be proven wrong with good justification?
You do prove a good example about the elderly, however for most "normal" people they will struggle to find balance between real life and virtual life. I say this because just take a look at how bad society currently is when it comes to simple things such as eating, taking pharmaceuticals, or finances.
>>47694014 People have died from pretty much anything ever. You're not clever or insightful for spewing bullshit about the possible negative consequences of some new technology as if it's at all significant
>>47694014 Not that guy, but I agree. The problem I think he has is that there are too many people these days talking about how bad something can turn out, as if that was the only thing that could happen.
>>47694034 Or, you know, how bad life is outside of the first world countries. If we get into talking about social justice and technology, it would be better to look at the overall global picture first, and then move into the micro-dynamics.
I'm just saying man, virtual reality can lure untold millions if not billions of lives if it gets out of hand. The question we must ask ourselves is whether or not we wish to abandon this world for a mankind created virtual world. Should we just forget about the universe we inhabit waiting to be explored?
>>47693878 I see. Yea, thanks. > I haven't followed the link or the current discussion really, so I might be wrong. It was a response to some anon telling me that my 5-10hz (observed in CPU-Z, might actually be faster in reality) smartphone sensor as used with Google cardboard is quite profoundly insufficient for VR, and that you should use (about?) 1000hz on the sensor.
I highly doubted that based on, human perception and head movement speeds and actual experience with Google cardboard, asked for some evidence, and got that link. Which says nothing of 1000hz (but rather 60hz).
>>47693909 > meaning you'd need interpolation which increases latency over the baseline 25 ms There is no baseline of 25ms for head tracking.
There is actually no value whatsoever that I've seen for head-only movement after which even just "empirically" the tracking becomes weird, never mind an actual scientific one.
> More here, on why they decided they needed the 1000Hz IMU There is literally nothing as to why they needed it. They just say they decided to make their own fast sensor.
>I don't know what you mean by this, "moving your head at hundreds of degrees/second" is moving your HEAD at hundreds of degrees/second, what does the body have to do with this? I'm pretty sure you have to also turn your body to be anywhere near ergonomically okay with turning your heads "hundreds" of degrees a second....
>I'm pretty sure you have to also turn your body to be anywhere near ergonomically okay with turning your heads "hundreds" of degrees a second....
You know that when you turn your head from left to right, you're turning 180 degrees, in less than a second? That equates to the speed of "hundreds" of degrees a second. Not to mention that we make small, but fast head movements, that also equal to the SPEED of "hundreds" of degrees a second. We're not talking about literally turning hundreds of degrees at a time here.
>>47693909 > 60Hz sensors would not be able to do any prediction at all because of too few samples in a meaningful recent timeframe. There are 60 samples if it's synchronized. And I guess it's otherwise on average half a sample's worth of difference wrong.
>>47694275 I don't know what you're talking about to be honest. I'm just pointing out that we make very fast head movements all the time, no matter how small. You can turn your head 180 degrees easily if you start out facing left, you know? Where did I mention turning your head 180 degrees so that you face behind you? You can do a 180 degree turn in less than a second easily, with no ergonomic drawback. I feel as if you also do not know intuitively how long a second lasts. It might be slower than you think.
>>47694424 Right, right, but I was talking about movement speed, not actually moving 180 degrees. We do faster than 180 degree rotations of the head all the time. That's what I was trying to say. I never said anything about body movement either, because I wasn't the original guy you were talking to, so don't even bring that up.
>>47691752 I hate the form factor; Mobile VR caught on because companies realized that it's the easiest cash-grab for the VR scene. I don't see why Oculus themselves are dead set on creating an all in one HMD; that's more akin to AR than VR. Mobile VR in its current state is just a fad companies are latching onto until "true" VR headsets actually launch and become affordable.
Even then, why are people afraid of having some sort of breakout box?
Imagine if someone made a single strapped 60% all in one mechanical keyboard whose keys are backlit with IR LEDs (for pic related especially if Leap Motion's Dragonfly sensor takes off) and also utilized something like Nvidia's Tegra X1 processor; it could even have some sort of eGPU slot for hotswapping. Let's not forget that you'll need a place to store the controllers (if there are any) and I don't think fanny packs are going to make a comeback any time soon.
>>47694424 I just did an experiment where I set a timer and did 20 180 degree head rotations at a comfortable and constant speed, and the speed our heads would normally turn if you were to look at something. I got a result of 9 seconds which comes out to around 400 degrees per second.
>>47694456 >>47694503 If you do 180 degree head rotation in under a second while sitting, you've moved your head left-right and back like 2.25-3 times in that second (60-80 degrees are typical turn range, apparently).
Again, you'll not do this anywhere near ergonomically without using the rest of your body, too, right?
>>47694497 I agree. But we also have to question if the non-freedom we get from products like these that offer potentially compelling benefits is, or isn't worth it, just like how we, as information and security professionals and academics (because that's the kind of people that go on /g/, not stupid consumers), question the morals of people who use Google in their everyday lives, as if it isn't a potentially dangerous and large scale proprietary technology.
>>47694537 I feel like you're confusing something here. For example what if I want to look at the cup to the right of my computer screen right now. I have to turn my head 20 degrees, and I do it at the speed of ~400 degrees per second. So what's the problem with that? Again, I was never talking about doing literally 180 degree turns, but the speed of doing turns.
>>47694617 Then it appears as if we've always been living in a dystopia, right from the day homo sapiens decided using secret farming methods was a good idea. Our society is inherently non-free, as long as people decide to keep anything secret, and as long as we don't integrate fully integrate and plug our brains into a network of other brains.
>>47694720 OK, let's simplify a bit. I turn my head 150 degrees back and forth, and I do 20 turns that are slower from the first experiment. I get 15 seconds, which comes out to 200 degrees per second. That still satisfies the condition of "hundreds" of degrees per second. Acceleration is also quite fast to get up to the 200 degree per second speed.
60 Hz polling of an IMU is not bad for VR if all you can do is just update the position on every frame if it is rendered at 60 AND displayed at 60 fps. You need more than 60 Hz for an IMU if you want positional tracking. You need it if you want higher refresh rates without latency (because without higher refresh rates, you'd have the old 60 Hz update speed without the benefit of a new frames). You need higher than 60 Hz refresh rate of the display because you want low-persistence. You want more than 60 Hz IMU for timewarp and for prediction, if you want to display higher than 60 Hz refresh rates, which is something you want for VR, since it will allow low persistence. Not to mention that we can still perceive a doubling of render speed quite easily up until a high limit which I forgot, but 60 Hz is not perfect. It's only good enough for limited applications that don't take advantage of VR, or presence. In that case, go ahead.
>>47694882 >200 degrees per second That is still *more* than the equivalent of 2.25-3 times full left-right movements a second. Sorry, but how is this going to be ergonomic?
As for the rest, again, I was never talking about how fast the device *internally* needs to work.
Just about how many times a second you need to be able to pretty reliably get a measurement *from* it with reasonably low latency.
And all the rest is mostly "you want" and nothing in the links so far has showed any of this scientifically.
>You need higher than 60 Hz refresh rate of the display because you want low-persistence.
Is not even in Valves blog or anywhere I can find it. They only say that you might get 16ms extra due to persistence in the worst case at 60hz on LCD, but a fuck's worth about this tells you this will destroy immersion or cause problems. Only that they'd like to target 20 ms.
Basically, I remain unconvinced. It's all *at most* documented by Valve's blog which seems to be a mixture of empirical observations and marketing goals. Alas, I guess we'll see when a range of actual devices to test stuff is available.
>>47695281 >That is still *more* than the equivalent of 2.25-3 times full left-right movements a second. Sorry, but how is this going to be ergonomic? 2.25-3 times the full left-right movement of the head per second equates to roughly 405-540 degrees per second, if 1 left-right movement = 180 degrees. So I guess we're still not talking in the same terms here.
How fast the device needs to work internally is determined by how fast the sensors can work. That's a given. The getting the data points is trivial. As far as I know there have been no papers written about this because it's so new and people have yet even had the time to conduct full studies on it, but it's not exactly necessary. If you do not know the visual difference between low-persistence and full persistence in VR, then it might be hard for you to understand. It's easy to google though, so do that first. We'd like to target 20 ms because it's doable, and because the lower latency you can get, the more you can prevent motion sickness and unrealistic visual feedback. You remain unconvinced because you don't have enough knowledge about the technology and you don't have an incentive to know about that knowledge because you have not seen it yourself. 16 ms latency is because you divide 1 second by 60 fps, not anything to do with low-persistence. You want higher refresh rates because low-persistence means you only display for 2-3 ms, and the rest is blank, so you get (16-(2 to 3)) ms that shows nothing. What do you think this looks like? At 60 fps, it is actually doable, but not ideal, as your visuals will look dimmer, unless you increase the persistence of the frame, but that will mean increasing the blur of the frames.
Thread replies: 107 Thread images: 11
Thread DB ID: 56098
All trademarks and copyrights on this page are owned by their respective parties. Images uploaded are the responsibility of the Poster. Comments are owned by the Poster.
This is a 4chan archive - all of the content originated from them. If you need IP information for a Poster - you need to contact them. This website shows only archived content.
If a post contains personal/copyrighted/illegal content you can contact me at firstname.lastname@example.org with that post and thread number and it will be removed as soon as possible.