Google and LG creates VR AMOLED 120 Hz at 5500 x 3000
blurbusters.com
blurbusters.com
There are a few things that need to be accomplished before widespread adoption:
- Removal of wires. It restricts movement too much and removes immersion. The new HTC headset is a step toward this.
- Higher resolution screens. VR AMOLEDs like this are a step in the right direction.
- Prices for GPUs need to go down, and/or a few more years are needed for average computers to be able to render high frame-rates without breaking the bank.
- Headsets need to be lighter and smaller.
- Removal of sensor placement the room. This will be harder to do, but cameras/sensors built on the headsets themselves could potentially accomplish this.
The way I see it, we're in the iPhone 1 stage of VR right now. Imagine the iPhone X version: lighter, smaller, higher resolution, more colors, higher frame-rate, less hassle. These are all inevitabilities, and at that point it will become much easier to adopt the technology. We're also missing a true "killer app" that will get people to purchase a headset JUST for that. I think it will take some sort of truly massive MMO the likes of WoW to accomplish that.
The future is definitely exciting in this field. I hope hardware vendors don't give up and can see the light at the end of the tunnel.
Also, in a perfect world this will all just be a transition step before we get full on holodecks.
We're doing our own small part to try and make that happen with our automated 3D scanning platform (http://realityzero.one)
If it’s a cartoonish game, the imaging doesn’t need to be photo-realistic. If it’s ‘toon style or lit and textured by photography of real environments, the polygon count doesn’t need to be high.
What it does need, beyond the hardware, is immersive physics, and a gameplay-justified reason for why you can’t run through the furniture/wall/cable that you can no longer see. The game “I Expect You to Die” does that perfectly because you’re sitting down the whole time.
You could immediately do a number of useful (read:productivity enhancing) things with the iPhone 1 - play your music, make calls, browse the web, take pictures, send emails. It deprecated a lot of what needed to be done on traditional phones, desktops, and laptop computers. It would last a day without charging, you wouldn't have to hook it up to a GPU or strap it to your body to use it, and it wouldn't give you motion sickness. Cellular technologies developed quickly to support the bandwidth needed for even better user experiences. Shipments jumped from 1M in 2007 to 20M in 2009.
There are very few polished games or apps available for VR 2 years after the "new" generation of VR headsets was released in 2016 by Oculus and HTC, and total headset unit shipments for the entire market (excluding phone-mounting headsets such as Gear) are probably in the low single digit millions for 2017. It hasn't yet deprecated any traditional dedicated communications technology or functions.
I wish I could get excited about the future in this field, but I really just don't see what the killer app will be for VR. Facebook thinks it will be virtual meetings for the enterprise and hanging out virtually with friends/family for the consumer market...I am very skeptical but want to be proven wrong as VR is one of the last platforms pushing hardware and software innovation forward at the moment.
Previous VR endeavours, like 90s, early 2000s would then be the Pac-Man and Donkey Kong in this analogy :-)
We're most definitely there. Here's the Original DOOM, modded for VR:
Not true. Examples: Brass Tactics, Robo Recall, In Death, Lone Echo. There are already more quality games in the Oculus Store and Steam than most people will have the time to play.
High res stereo at 120Hz is never going to have to same graphics as the latest high budget big game release. Current GPUs are already very powerful, but if people have the expectations of getting the same graphics when they use VR they are going to be very frustrated.
At some point your GPU runs out of things to do.
The reality is that greater horsepower allows for greater abstraction, and easier to program APIs. Increasing developer productivity 2x reduces performance 10-100x, or something like that. So there’s never “enough” performance for the same reason there’s never “enough” powerful/usable APIs.
Audio used to be really difficult to process in real-time but now it's trivial. There's only so much audio processing you can do before it's ridiculous and pointless.
The same goes for video. Once you have, say, a 40K display for each eye at 244Hz there's no point in going for more pixels or faster refresh rates. If a GPU can handle that, easily, then that GPU will probably be best put to use doing other things in addition to rendering graphics.
Memory is not tied to your senses, we can always find uses for more. Audio and video are, and at some point it's as good as real.
Its quite possible that with progress made in brains wavelengths readings, the true vr in form of iphone1 will be a head cap you put on and your view/smell/touch perceptions are overwritten by cap’s sensor. That would trully be regular vr/phone versus true vr/smartphone/iphone1
That sort of headset would be such a massive breakthrough in neuroscience that the VR aspect of it would be tiny in comparison.
As we finally realize, how not very real time our OS and software / hardware. After all these years of abstraction and slight delay added to everywhere in stack. We finally have a motive to unwind / improve those.
May be some day we could have Sword Art Online Style Game to move VR forward. Link Start.
Inside out tracking is a reality in consumer devices now. All Windows MR devices that shipped late last year have 6DOF inside out tracking via cameras on the front of the headset with no external sensors. Moving forward there will be more devices from other vendors that use inside out tracking. Qualcomm has shown prototypes, Google + HTC were working on a tango device that got cancelled, HTC is working on an inside out standalone for the Chinese market, Oculus has shown standalone inside out tracked prototypes, etc.
They are very, very good. I've owned every major HMD since the DK2 came out in 2014, and I would say the Samsung Odyssey is the best one to date. The inside-out tracking is fantastic and just as good as Lighthouse (in practical usage, not theoretically). When you consider that there is no setup involved, it makes it a no brainer that this is the way forward.
Maybe they could do inside-out tracking on the controllers?
Agreed, there is a bit of an occlusion issue when your arms are behind/above the HMD. I feel like they could probably solve this though, with inductive tracking like Sixsense [0] integrated into the controllers and used in conjunction with IMU/camera data.
VR was already too expensive. Now it’s completely out of the question for most people. Adding an absurdly high res and high refresh HMD to the mix right now doesn’t seem like a good idea.
And those things were true.
Games and people adapted.
Personally, I feel something work that can pick up nerve impulses as well as deliver modest feedback, tactile and or electrical will close much of this gap.
New input paradigms will advance too, just as they currently are for touch.
Touch today is getting good. The finger in the way problem is being chipped away.
BTW, GPU is a problem, but we're expecting Frame Rate Amplification Technologies to solve the problem. Basically improved versions of Oculus Spacewarp that can do large framerate multiplication factors with zero parallax artifacts (unlike today).
I covered this topic near the bottom of a different article about the journey to 1000 Hz displays at https://www.blurbusters.com/1000hz-journey
The gist is that within five to ten years, we'll have many tricks to increase framerates with the same number of transistors, without needing to reduce detail levels or make textures/edges blurry, without input lag, and without interpolation artifacts.
So if that delay is 200ms, we know it makes people sick, for example. A delay of 0ms would be "zero lag" IMO.
When you're dealing with a head moving, and very brief slices of time, inertia plays a large role and allows for fairly accurate prediction. After rendering the frame they check head position again, update their prediction for head position at time of display, and move/warp the frame slightly to match. This does require rendering a slightly larger view.
I remember when Oculus cracked the 20 ms mark and got down into imperceptible lag, it was very exciting. They bragged at the time that their predictive models would let them get down to 0 ms eventually, but I'm not sure if they've hit that yet.
Mice and head trackers can already run at 1000 Hz. It's the GPU that cannot keep up.
Instead of black-box interpolators (e.g. Sony MotionFlow), a smart interpolator can be made to know the high-frequency controller inputs in realtime, and doesn't even need to use guesswork-based interpolation for everything.
Just shift everything around based on the high-refresh 1000Hz controller input. (In other words, "reprojection").
Also, knowing more data about the source (e.g. near-zero-lag controller input stream) eliminates lots of interpolation guesswork. It's much like how H.264 (video compression) is heavily interpolation-based mathematics during the video codec, but it had full awareness of the source video material, to successfully compress it virtually artifact-free.
So basically, you are simply giving a smart interpolator full awareness of things like geometry & input at a higher rate than the GPU renders. To avoid guesswork on those kinds of items.
Things like future multilayer Z-buffers can help solve a lot of parallax-reveal problems of trying to create intermediate frames, and there are future tweaks they are working on to eliminate reprojection artifacts. Like artifacts or reprojection distortions around edges of objects in front of objects. So adding intermediate frames with full parallax effects can eventually become artifact free because of the GPU's knowledge-in-advance of what-behind-what. Basically, more advanced reprojection algorithms that can create near-flawless intermediate GPU frames (without lookforward) without a full polygonal rerender.
Prediction helps (as it does for Oculus), but remember, we have controllers that already go at ultra high frequencies, and it is expected headtrackers will eventually become ultra high frequency too -- and that extra data can reduce the need to do lookforward prediction.
It's all very complex, with many researchers working on multiple solutions, but it can reduce the average processing-power-required per extra frame, and it can theoretically allow high reprojection ratios without lookforward lag (e.g. theoretical future 10:1, such as multiplying 100fps to 1000fps, at least with 1000Hz input devices like 1000Hz gaming mice, and 1000Hz head trackers).
Several VR scientists have indeed advocated the need for 1000Hz eventually, someday in humankind, as there are confirmed tangible immersion benefits to getting that high and beyond.
That's why I wrote that article full of motion demos explaining the visual science concepts of why 1000Hz displays are needed. It will be useful for passing a theoretical future Holodeck Turing Test (not telling apart a VR headset versus transparent ski goggles in a reality-versus-VR blind test), in terms of Morarity-style or Matrix-style "it's real" VR.
Many tricks layers upon each other, to achieve what's being achieved today, and this creativity will only continue. Lagless lookbehind-only interpolation (utilizing ultra-high-Hz controller input to reproject new 3D position). Foveated rendering too, yes. Realtime beamtracing with realtime denoising (NVIDIA scientist paper), perhaps. Maybe even all piled on top of each other simultaneously, perhaps.
Similar techniques are used in video compression - encoding the exact value of every pixel is expensive, but you can trade bandwidth for processing by encoding transformations of a previous frame. A modern compressed video consists mainly of these interpolated frames, with only a minority of frames containing a full image. This interpolation can use data from both past and future frames (B frames) but can also use just the data in previous frames (P frames). This works extremely well most of the time, but there are some edge cases:
As someone who has had an oculus since the consumer version was released, my main problems are:
1) Eye strain. Even though you have a 3d effect you're still looking at something a few inches from your eye and that disconnect causes eye pain and headaches after playing for more than an hour or so.
2) Locomotion. I've yet to find any way of moving around in VR space that doesn't either make you nauseous or pull you right out of the realism of the experience.
Locomotion is less of a problem than people who think its a problem is, and I think that stems from a lot of people in VR being hardcore gamers, and exploration of large spaces being a core mechanic and selling point for 3D video games for the last 25 years. I don't think omni-directional treadmills or vestibular stimulation or anything inconvenient like that will catch on for locomotion, and I think we'll either use various forms of teleport, or sliding (traditional 3D) locomotion for the foreseeable future, and people will mostly be ok with it. It's possible that with wireless and/or standalone headsets that redirected walking, will be a popular option. You could imagine a headset where Chaperone\Guardian builds on SLAM used for inside out tracking, and can give apps information about the layout of your house, allowing for large procedurally created virtual spaces. This still doesn't solve the problem for people that don't have medium sized private spaces to play in, and also makes it harder to do multiplayer games where players are playing in very different spaces.
There's going to be a black bar across the top middle of my field of vision?
Alone for remodeling my flat (kitchen, bathroom) or building a house.
I would also like to train on a virtual lathe before using a real one.
I might buy the new htc vive and i will see it as an early adopter beta hype thing because there is still work to do, but in general it feels already really good.
That Valve Portal Demo, wow that frightned me a little bit :)
We already have 3d video - I'd be surprised if the concept couldn't be expanded, leaving the unit to merge streams of background environment and 3d animated sprites.
I assume this is targeted to the gaming market because at those specs unless you sell it at a loss the MSRP is going to be at least $500. I wonder if this signals Google entring into the VR arena as a publisher to compete with Oculus. I know that also VR is used for Advertisements and they have engagement for properly setup ads for VR as 5 minutes or more so maybe they will be pushing on that front. From what I have seen from Google their VR has mostly been focused on mobile so them focusing on the high end market is really interesting.
Clarity isn't as important as getting rid of the screen door effect.
The other bonus here is that with the huge resolution bump, other use cases become available. Reading text in VR is tricky with the shipping headsets right now, but with this, a virtual desktop could be realistic.
Higher resolution screen with upscaled input does not sacrifice clarity for blur/no SDE. Upscaled input won't have as high detail as a higher resolution input, but the performance gains of the upscaled input far outweigh the negatives.
Conceivably, a higher resolution display would have a worse SDE than a lower one. If the pixels became smaller, then the gaps between pixels could get worse even if there are more pixels per inch. For example, [1] shows a game of Tetris played on a building and the problem with it is not that the resolution is low, but that the pixels have too gaps between them. Doubling the resolution would only help a little but make each pixel a big square instead of a dot would help immensely. A diffusive screen over that building could help potentially quite a lot. This is obviously an absurdly exaggerated scenario.
[1] https://mashable.com/2014/04/06/building-tetris-philadelphia...
The microlens can also be designed to blur the RGB together to provide a true white pixel from the approx three coloured subpixels.
Whenever we crumple our eyes up to squint we're creating a diffraction grating that works a treat for the myopic amnesiac I am when without my spectacles.
The ground glass of the focusing screen in my Nikon is diffuse. This particular diffusion assists with guaging the lens focus, bit only for smaller apertures than f/2.0, so that expensive 85mm f/1.4 prime lenses need autofocus. Dunno how I managed for two decades professionally before the digital switch rendered great optics practically redundant*
So what's preventing the projection of the high resolution display information into a kind of focusing screen like the old fashioned Single Lens Reflex cameras won the market with in the fifties and sixties?
Is it more physical constraints of the imposition of the display into the field of vision for infinity focus?
If long ago my amateur interest would have carried my thinking further, sadly I have no applicable reading even on the subjects, so forgive me if this is a beginner supposition in error.
I certainly see the opportunity to directly encourage the hardware development, even if the equipment will be unwieldy to use, even more so than today, for specialist fields such as fabric design and magazine printing. I have been involved in the latter for the entirety of my career, two decades of that running the company I hoped would open up enormous markets and that field even defenestrated Google in 2004, leading to the most epic narrative and pivot story if I manage to get things kicking again. * Surely there is a very good overlap between the current generation of customers for whom the budget simply isn't the obstacle, and the same customers who will allocate whatever you want to charge in the future, if you can deliver real advantages? I'm optimistic about VR, but not in any kind of short term horizon. I founded my own business on a thirty year plan.. that still could be met, and made decades one and two, kinda just the latter. Maybe this is what we all need: real long term benefit analysis and commitment. This isn't a factor of the youthfulness of the Valley to which short termism is often ascribed,when flighty CEOs are discounted, I was very young and junior by a generation and some to my co-founder, when I realized how I had found my calling. Surely the VR sector has plenty of similar thinkers young enough to put such time to their dreams?
*I kinda continue a bit in my profile, where I may be more bold soon. I welcome any feedback or inquiry, I may be less than the epitome of clarity when I come close to professional engagement with ideas I can't believe.. the industry hasn't destroyed my hopes because of politics not technology, not yet anyhow..
The one consistent thing about Google's VR/AR strategy so far is that it's extremely inconsistent. Different groups within Google are taking a lot of shots on different technologies and form factors with different partners. Some examples include Google Glass (built by Google X, now revived as Glass X for Enterprise) which was supposed to be picked up by Tony Fadell of Nest who then ended up leaving Google altogether, Google Cardboard (a side project by 2 smart Google France engineers for a conference), Google Dream w/ Daydream View, Google Tango (built by ATAP group and shut down after announcement of ARCore), and now ARCore w/ Asus and LG.
There is no discernible Google VR/AR strategy other than - let's see what bubbles up from different dev groups and, when needed, react to market forces (ARCore was a direct reaction to Apple's ARKit). IMO until Google really "focuses" on VR/AR with a dedicated group and strategy this new tech will remain a sideshow without much traction.
Clay Bavor's presentation at Google I/O[1] and SID Display Week[2] seems to outline their strategy concerning AR/VR pretty well.
>Some examples include Google Glass
Released in 2013 and pivoted to enterprise use. Now under the control of the Google hardware division.
>Tony Fadell of Nest who then ended up leaving Google altogether
Yes, he did leave Google, but only after being assigned to the roof.
>Google Cardboard (a side project by 2 smart Google France engineers for a conference)
A low cost VR solution developed in 2014 by two Googlers during their 20% innovation time off. I believe it's also the most widely used VR solution in the world.
It is not a must, but it is a mostly. Most people want something approaching realism by default. Stylism is okay but it isn't the primary use case.
The video game sales show this. Top selling games are mostly ultra realistic, Witcher, Call of Duty, Modern Warfare, Battlefront, etc. Yes there is Nintendo's low end strategy but most are ultra realistic.
A significant benefit of higher res VR displays is not being able to physically see the pixels / screen door effect of the screen. You still get this benefit by rendering at a lower resolution and upscaling to the display resolution. Certainly you will be able to see 'render pixels', but that's a much less significant problem and just reduces the realism of the scene.
The brain corrects for things like that automatically. Might work just as well here.
Source: https://www.roadtovr.com/google-shares-new-research-foveated...
(the one you linked came up first for me too when searching around, took a little more digging to find the original article).
Technically though the fov is refering to fovea, not field of view.
Yeah. If the new panel is presented as 3x or 4x 1080p (two foveal insets and one or two backgrounds), and tolerates 30 fps, I could run it off the old integrated graphics I'm typing at now.
> The primary issues in this space is latency
For immersive VR, yes. But as one moves away from gaming, and immersive, and VR, design constraints relax a lot.
There's so much ambient confusion about the shape of the broader design space, and misattribution of constraints.
I usually use my WMR (and Vive before that) on an old laptop with Intel integrated graphics. A duct-taped-on 30 fps camera-passthrough AR serves for balance, permitting <30 fps rendering for a software development 3D "desktop". Latency, and its variance, is just not a challenge here.
I can even do subpixel rendering on the WMR. Lens focus is so bad, subpixels are only worth bothering with for like (0.5 kP)^2/eye (of the ~1 kP square/eye usable area, of the 1.4 kP square/eye panel). There's no point in even rendering full native resolution over much of the visual field. And not caring about immersion, there's no need for barrel and chromatic correction, nor for blending away seams between resolutions.
Such a vast amount of effort is spent on dealing with "horrible immersion-breaking visual artifacts", which for some parts of design space, are like "meh, sooo don't care". Remember the skeuomorphic user interface fad of a decade ago? A calender app should look like a leather-bound calendar book? And... remember all the claims that calendar apps wouldn't be viable for years, because of the GPU demands of faithfully rendering light reflection off the leather. Yeah, me neither. That would have been silly. Just like expecting a 3D/HMD professional "desktop" environment to prioritize immersion.
Vulkan allows game developers to spread the load across all GPUs, if they wish to. You can do SLI with 8 1080Tis, an intel iGPU and an RX VEGA all together without issues
In contrast to SLI, where the driver frequently was the cause of issues.
But many of the modern engines have started splitting their render tasks so that some can be put onto additional GPUs (or physics tasks, or animation tasks, ...)
What motivates people to write these silly comments attacking cryptocurrency investors? Jealousy?
For that you'd need multiple displayport cables.
By the time GPUs able to render at this resolution and framerate become mainstream I'm sure we'll have the cables to connect them. The big problem with VR is that you want very long cables to be able to move around unless you manage to stuff everything into a backpack. Very high bandwidth and long cables don't always play very nice together. Worst case scenario there's always optical fiber...
I would have loved Apple to capture on this, as they stand to be the company that makes software and hardware well integrated. The problem is Apple has never gave a damn about gaming. They say they do, but they dont. And It isn't their priority, it isn't in their DNA. ( I bet none of the VPs are gamers of any sort ).
Nintendo could be another fit. But they are always lacking in Hardware and VR is very much hardware limited.
Also, with high enough resolutions, you don't need anti-aliasing. This is pretty close to 8K, which is probably around the time that anti-aliasing stops mattering.
That being said my high-end gaming PC (using a GTX1080) struggles to maintain native framerate in demanding games without reprojection on the Rift, I can't imagine what kind of futuristic computer you'll need to drive 5500x3000@120Hz in VR. Let's hope the cryptocurrency mining gold rush will have subdued by then.
I think that this kind of headset is meaningful for both VR power users and developers but Google and Facebook should focus on non gaming content to attract a wider audience.
Facebook is in the odd position where they could make a WebVR social killer app over night but they won't because they really badly want to become the Apple of VR with their Oculus Store. I think that they will regret this strategy in the future. That or they will have to cash billions out once more to acquire a company that found out what to do with this tech.
With higher resolution you could correct for more lens distortion, but you still need lenses, and you won't be able to fix lens flare on fresnel lenses.
Something that can give immediate immersion when put on, scaled down, lightweight/comfortable, portable and can last around 5 hours would be a great first step.
https://www.youtube.com/playlist?list=PLL-lmlkrmJalNqp7Q_dLA...
A custom high bandwidth driver IC was fabricated. Foveated driving logic for VR and AR applications was implemented.
If you are interested in modelling the behaviour of a modern fabric loom, which can be a four storey high proposition with thousands of spindles feeding air guided bobbins, where the weight of the threads in your weave design affects the entire process, the way your thread hangs requires a different tension, how it stretches affects the feed and speed with which the bobbin can fly through the weft (vertical line frame into which the pattern is woven) and the elasticity and friction tension the weft sometimes intentionally... I am willing to bet it would suit me if the visual rendering is put on ASICS or FPGAs, while the physical modelling for cloth behaviour might be a more general purpose solution.
A single 4.3 inch isn't big enough for VR, right?