Sony VR Headset/Project Morpheus/PlayStation VR

Is VR really that much more demanding? I would have thought there'd be some drop off, but not that big. I guess you have to render two different views like you would 3D?
 
That's a little too harsh I think. PS4 is around 10x PS3 horsepower, it should be capable to extract more visual performance in its VR games.

Plus devs now for the first time have a great interest in creating 3D rendering pipelines that are fast and efficient.

I don't think Carmack's statement negates the fact that the PS4 would be capable of running games in 3D, 1080/60 and also have better graphics.
He's just saying that a game that ran at 60fps on PS3 will run at 1080p/60 in 3D with AA on PS4.

PS4-only games are not included in that statement, and we've already seen that PS4 is a lot more capable than just running PS3 games at 1080p/60/AA.

That's how I read it anyway.
 
Any guesses on the R&D cost associated with moving in this direction both in terms of capital and potential overhead from a planning and project management perspective? I ask bc while VR is interesting to me the additional resource allocation is a very real concern. Why are they continuing to pursue big risky plays in this space when there is little evidence to suggest VR will be widely adopted?
In case it turns out to be the next big thing. Technology has always withheld VR adoption because the experience has been poor. We're getting to the point where the technology is now an enabler, which means it'll be down to the consumers to decide if it's importnat or not. If they do, you really want to be at the forefront of the new market. If they don't, it's just another R&D expense.

I was kind of hoping Occulus Rift would beat Sony and Microsoft to market and get the lions share, because they're the ones that have brought people back around to VR again. All the press the last couple years, and the big kickstarter campaign. But it's hard to compete with the giants.
That might be unfair. OR was a public campaign because it was publicly funded. Companies like Sony can't go waving their R&D in public in case of giving away ideas to the competition. Reportedly they've been working on this for 3 years, and I'm sure they've dabbled with the idea before. They may have got some ideas from OR, but that's the problem with revealing your unfinished ideas - you leave yourself wide open to copycat products, unless you have a watertight patent situation. Which is why Sony et al don't say what they're working on until they're ready to as based on their evaluation of the competition.

There's the possibility that even without OR, Sony would be releasing this VR project. They had the HMZ, they probably tried VR on PS3, found the experience compelling but felt the technical limits (PS3 hardware, headset cost) meant a product would have to wait, and carried on with prototyping until they found something that was viable.

There is of course the possibility that Sony had no ideas, saw OR, and copied it using their established R&D talent and financial power to rush through a clone, but that strikes me as far less probable. I'm reminded of Move and claims Sony copied Wii, whereupon old R&D videos of glowing ball tracking on PS2 disproved these. Sony didn't release Move until they were ready (quite possibly prompted by Wii's success), but they had it waiting in the wings. Perhaps they've had a VR prototype in the works for a while, and it took OR's public success to motivate management to actually release it - something that we've heard Sony can be reluctant to do with it's R&D creations.
 
I wonder whether we'll see games that use the motion head tracking of the device, but drop the 3D aspect. The games would only then need to generate 1x 960x1080 view and duplicate it to the opposite view.

I remember using one of the devices back in the 90s, only I can't remember whether it was 3D or just one view.
 
Is VR really that much more demanding? I would have thought there'd be some drop off, but not that big. I guess you have to render two different views like you would 3D?
Yep. It's exactly the same requirements as rendering a stereoscopic game, so the same possible solutions and workarounds and costs. For true immersion, anything less than fully dual rendered images might be unacceptable.

Oh, only it won't be 2x1080p rendering because each eye is half res, so it'll be the equivalent of 1x1080p rendering regards drawing pixels, and I think ~2x regards geometry (don't know what reuse there is, but I imagine less than this reusing data for second eye). Or more like rendering dual 720p screens. So think PS4 rendering 720p at 60 fps - I can't see why something of the order of Infamous 3 won't be doable in VR.
 
I don't think Carmack's statement negates the fact that the PS4 would be capable of running games in 3D, 1080/60 and also have better graphics.
He's just saying that a game that ran at 60fps on PS3 will run at 1080p/60 in 3D with AA on PS4.

PS4-only games are not included in that statement, and we've already seen that PS4 is a lot more capable than just running PS3 games at 1080p/60/AA.

That's how I read it anyway.

VR requires two 1080P streams though right? This is 4.5X the pixels as 720P. Then throw in some MSAA (I assume Carmack thinks this is necessary for immersion). He also throws in the low latency caveat, which implies to me it must hold very tight to 60 FPS and not deviate, so that's another tough performance demand beyond a soft 60 FPS.

And if 60 FPS is a necessity, then a 30 FPS PS3 game would indeed get you to that 10X in performance needed.

Which isn't all bad I guess, there's some nice looking PS3/360 games certainly.

Of course I suppose Carmack has a certain bias, sure. Now Sony is out of the picture as a possible partner. Carmack will of course want to tout PC benefits.

This is the bugaboo of VR though, like 3D it's a major, major graphics downgrade. And as as I always say, there's no close second to graphics in core gaming importance.

But, what is to be seen is if the experience can be so compelling to overcome that.

About this project Morpheous, it seems like Sony doesn't exactly have the details down (since they claim specs are not at all final) but wanted to show something fast.

And with Sony, they've got a history of jumping in grand projects, only to half ass them and they die on the vine (Move, 3D to some extent, PSeye really)

I'm not sure how this works from a business side. I suppose it'll be a $300 peripheral. Will that fly? Only one way to find out. Mandatory bundling would certainly be suicide. Requiring Move and PSEye, which if I understand it does, is another major problem/expense, unless all that is included with the headset.

I dunno, I've been giving some thought to VR, and it's another one of those technologies that's hard for me to gauge it's business potential. Only one way to find out.
 
VR requires two 1080P streams though right? This is 4.5X the pixels as 720P. Then throw in some MSAA (I assume Carmack thinks this is necessary for immersion). He also throws in the low latency caveat, which implies to me it must hold very tight to 60 FPS and not deviate, so that's another tough performance demand beyond a soft 60 FPS.

And if 60 FPS is a necessity, then a 30 FPS PS3 game would indeed get you to that 10X in performance needed.
See my post above. Realistically the workload for a VR PS4 game is 720p60 * 2. I think Carmack's very wrong on this.
 
VR requires two 1080P streams though right?

No. Its 1080p SBS rendering.
http://i1.ytimg.com/vi/ConJqygVOqE/maxresdefault.jpg

Same number of pixels like one 1080p screen, but instead of rendering one view, engine has to render two "half screen" images. That can be done with the brute force way [calculate a lot of things twice] or simpler by shader reprojection [this however cannot work great with reflections/transparencies but is VERY FAST].
 
"This is a guy who builds spaceships." -- Mark Cerny

Wipeout would be insane though, no matter how much they improve the graphics. Why use drugs when you can do Zone Levels in VR.
 
I’ve tried to think up of some more alternative ways of displaying games with using a 1920x1080 screen split into two:

As stated above, devs could duplicate (or give some reprojection) to one of the views, so the game only need render once at 960x1080. I can imagine Crytek doing this.

Drop to two thirds the resolution to something approximating 900x800.

Go down to half the resolution, like 700x740.

I’d also hope that each view gets an elipse type shape instead of a perfectly spherical window, at least then it’s make better use of the resolution. I imagine a very specific lens would need to be applied. In my mind this would also increase the FOV to an extent.
 
Or more like rendering dual 720p screens. So think PS4 rendering 720p at 60 fps - I can't see why something of the order of Infamous 3 won't be doable in VR.

Time to port the PS3 portfolio. :yes: Should be really decent with affordable MSAA there.

(God of War III remade for 3D :O )
Dead or Alive 5+
 
Carmack tweets

Isn't that technically wrong?
Project Morpheus seems to use a single 5" 1920*1080p screen that is split vertically, so 960*1080p for each eye.

Therefore, a game that runs in PS4 VR is a game that runs at 60FPS 1920*1080p on the PS4. Which is.. almost all PS4 games with PS4 visuals?

Maybe there's an overhead for rendering two different points of view, but it shouldn't be all that much.
 
Time to port the PS3 portfolio. :yes: Should be really decent with affordable MSAA there.

(God of War III remade for 3D :O )
This for me is going to be the make-or-break for VR. What games are actually going to work? Are people going to get motion sickness (worse) with some game types like over-the-shoulder cameras? Seeing people try 3D and dislike it, VR is going to face the same problems, perhaps moreso. So at the end of the day, even a perfect VR solution may fail to be popular, because people just may not like it. But I'm sure there'll be core gamers placing racers and the odd FPS that'll take to it no matter what.

Maybe there's an overhead for rendering two different points of view, but it shouldn't be all that much.
Nope. Worst case would be 2x the workload. Reality should be less than that.

Incidentally, the EG article has over 800 comments already. I don't recall any story getting that much social feedback, suggesting this is actually a Big Thing. It Patsu's correct and PS4's original price target was lower than it released at, and if this thing isn't too expensive (screen + optics. No large ASIC?), maybe we'll be looking at a £400 VR system come Christmas? That could actually be really big with at least a year of strong 'gimmick' sales until we learn whether VR is a real thing or not.
 
Basically the same rendering technique that GG used in Shadowfall could be used for VR? That would be promising though they would have to introduce all those optimisations they were talking about after the launch. I don't remember the details but didn't they say that they figured out a shader optimisation that would allow treble the pixel throughput?

Anyway, looks like MS is scrambling to get on board the VR train as well now. Though it sound like they are heading down an AR route as opposed to VR. How very PS2 of them! ;) But I guess the $ put into the Kinnect R&D has to be justified somehow.
 
shahid7mu1d.png
 
Most games won't work, if simply converted to VR.
The people at Oculus seem to think that a game, in order to work, needs to be developed specifically for a VR experience.
The limitations with what you can do, are a lot, and anything else will pretty much result in a nausea inducing experience.
 
It Patsu's correct and PS4's original price target was lower than it released at, and if this thing isn't too expensive (screen + optics. No large ASIC?), maybe we'll be looking at a £400 VR system come Christmas? That could actually be really big with at least a year of strong 'gimmick' sales until we learn whether VR is a real thing or not.

If they release the add-on kit for £250-300 depending on the box content then I would be ordering right now!

I used VR\AR quite a bit in the VR Labs at University where we had a complete suit you put on that had force feed back units in the hands etc. So when you push against something in the VR it pushed back. Very cool. But the actual GFX were being pushed out by an Onyx2 rack mounted system.

It was all very compelling even then. I did work on the emergent properties of the environment and the general physics etc. The only downside was the smell of pine disinfectant and the sick bucket by the treadmill!!

I also have to own up to actually, for a short time, owning a Virtuality system way back in the silicon jurassic period!
 
This for me is going to be the make-or-break for VR. What games are actually going to work? Are people going to get motion sickness (worse) with some game types like over-the-shoulder cameras? Seeing people try 3D and dislike it, VR is going to face the same problems, perhaps moreso. So at the end of the day, even a perfect VR solution may fail to be popular, because people just may not like it. But I'm sure there'll be core gamers placing racers and the odd FPS that'll take to it no matter what.

Yeah... at the moment, I'm thinking racers, first person horror*, maybe even a "revival" of puzzle adventure/mystery/crime solving. Then there's on-rails games (House of the Dead?).

Poker/insert card game tournament (Too bad All-Stars kinda failed. An All-Stars Poker with Kratos across the table would be unnerving. :p)

*would be kinda cool for that new Alien game - also force you to look down to use your motion tracker ?
 
Back
Top