Wowfunhappy
Member
How do I add applications to Vive Home?
-
Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.
HTC Vive Launch Thread -- Computer, activate holodeck
- Thread starter taco543
- Start date
- Status
GuitarAtomik
Member
Assuming I understand your question, they should just be there. Even my Revive stuff showed up. It may take a little bit of time to populate if it's the first time you're jumping in though.
Project cars is fantastic in VR, but I can completely understand not enjoying the resolution. But this is also the first generation, and resolution will be one of the highest priority things to improve.
They appear automatically once you play them once, as far as I can tell.
GuitarAtomik
Member
CrossVR confirmed that Edge of Nowhere is using Denuvo which makes it incompatible with the current Revive. He seems to be working on a work around though.
Thanks. Based on your responses I will just hold on to it. The stuck pixel I have is visible in bright areas with large coverage only, particularly when there is some blue. The sky in Fantastic Contraption, for example, is an area where I sometimes catch it. Most of the time it's completely invisible even when I look for it, however.
I still don't understand this direction of thinking? How are we going to push these higher resolution displays? And if there is a way, why are those people with those solutions not cranking the hell out of SS & AA to make the current headsets look much better? PCars look pretty darn good to me with DS9 AA. I cant run it with my 970, but if there is something that can at 90fps I will open my wallet. If there isn't (I doubt even the 1080 can), then what do we need HMDs with higher res displays for?
I think the assumption is that higher res HMDs will appear along better GPUs.
artsi
Member
Like we have been adapting to increasing graphics demands since forever: we are going to use high performance GPU's.
1080ti, 1170 / 1180 after that, etc.
It's not like the new VR gen is coming tomorrow.
Zalusithix
Member
Higher resolution displays will have to happen alongside foveated rendering. There's simply not enough power to dedicate to rendering the periphery at high resolutions. Even when we get better hardware. Combine foveated rendering with eye tracking to get the resolution where you need it.
Fried Food
Banned
Raising the quality with the current headsets will only get people so far. There is only so much you can do with the resolution we are working with. As far as perceivable visual quality goes.
So 2 years from now? We are just now getting new GPUs.
Iced_Eagle
Member
Rendering optimizations and tricks is the biggest thing. New hardware enhancements in both GPU's and HMD's in future generations of headsets will also help. For example, foveated rendering is probably one of the most well known upcoming tricks that requires eye tracking hardware in the HMD. With that, you render full resolution in a circle where the eye is looking, a few degrees out from that is a lower resolution circle and the outer is lowest edge. The eye sees clearest at the center and it fades out from there, so if you know where the eye is looking, you can reduce the amount you need to render with ideally no perceivable image quality difference.
You also have things like improved time warping capabilities and naturally, users are constantly upgrading their hardware. VR is going to be a big push from the entire industry, so expect focused enhancements there from basically every part of the stack, from OS kernels on up (MS made changes to the Win10 kernel to improve VR as an example by removing buffering at portions to reduce latency).
New gen GPUs of course. Plus advancements in VR tech like foveated rendering and support of VR GPU tech like multi projection and SLI.
The 2nd gen of HMDs will probably come out around the same time as the 1180Ti or 1280 too,(I see new iterations of HMDs coming out every 2 years or so) so theres plenty of time for tech to meet the demands of a higher resolution HMD.
artsi
Member
I would not expect sooner than that, personally.
My bet is 2-3 years from now until they're in our hands.
Maybe some kind of refresh model with small improvements in between (ergonomics, etc.) but no real generational leap until that.
Wrestlemania
Banned
Yeah, this is my thinking. Next revision of Vive will be a big improvement to HMD design but little or no improvement to the screens or optics generally, and it'll be out within a year. Next big revision will be a year later.
So software enhancements are the way forward then. Even today's headsets have sweetspots and periphery that can be rendered at different levels. Whenever someone mentioned VR is low res and blurry, there is always a response of "Wait for Gen2, HMDs will have better screens". My point is, better screens are only a portion of the problem. All the other problems, when they are solved, will actually improve the experience in these gen 1 headsets also.
Haint
Member
If the 480 and 10XX are what 2 node shrinks (3 years delayed) and new architecture achieve, it's probably safe to rule out exponential GPU advances brute forcing 4k+ 120hz+ HMD's anytime soon, particularly in a mainstream ~$300 card. We're at least 4 or 5 years away another GPU node shrink as even Intel can't pull off 10nm. Foveated rendering/eye tracking combined with VR specific efficiency's (i.e. Nvidia's Pascal VRWorks shit) are where the real gains will be.
Yeah of course. We're just getting 1080 now on a small chip (almost half the area of the Xbox one APU) and no hbm. In two years we should have much larger chips and more powerful chips, and hbm2 for more memory bandwidth. Combined with systems to reduce overhead like divested rendering and single pass stereo, I can see 4k per eye being doable by then, or 2k per eye being easy. I'd rather have 4K per eye even if that means up scaling, because it'd mean less visible pixels.
Blew my mind when i read about this.
One thing i'm curious about: surely eye tracking - and dynamically adjusting resolution to its position - will be heavy af on the CPU? Eyes move a lot, and quickly - surely we'll need to shift loads of clicks to track them?
qa_engineer
Member
Yep. The only way to get that type of performance in the near future (12-24 months) is for developers to fully embrace multi gpu setups. Then we can have CF and SLI setups that power each individual display
I read that the eye movements (saccades I think) are very rapid, but that might be within a very small angle of change. So you might be able to render a slightly larger circle and that is enough to mitigate for the smaller movements of the eye. in theory you might also want to decouple framerate and have a really high framerate for the centre area in focus, and a slightly lower framerate around it.
Haint
Member
Yes there is a shit ton of overhead. IIRC Carmack responded to this saying you'd have to be talking about an 8k HMD to see any performance improvements.
We could have fun with math. Current 1080 is doing lots of games around 4K at 60fps. It's a 300sqmm chip. I think it wouldn't be unreasonable to expect to see a 600sqmm chip in two years, so a doubling of performance compared to 1080 should be achievable. That'd give us enough to do 4K + 4K at 60fps in current game engines on ultra or very high settings. Current VR is already simplified graphically so I could see a '2x 1080 performance GPU' within 2 years being able to deliver VR optimised graphics at 4K + 4K 90fps. Single pass stereo means you only need one geometry pass so that helps in this case.
And that isn't even factoring in multires shading (rendering outer images at lower resolutions) which should improve performance an estimated 30-40% I think; or simultaneous multiprojection which warps the viewports to reduce the pixels you need to render - another 30% improvement.
So even without foveated rendering or VR SLI, I think a high end GPU could drive a VR HMD with 4K per eye in two years time. With foveated rendering, multires shading and simultaneous multiprojection, I think you're back into the ballpark of a 2018 GTX970 class GPU
Anyway this is all part of the fun of the journey
I think the overhead is more on the rendering side, since you are increasing the number of rendering passes even if one is at a lower resolution.
I assume the eye tracking would be done onboard the HMD with dedicated hardware. I don't think the image processing would be that difficult. The image will basically be just your eyes, at a known distance and angle, with no external crap to filter out.
If constellation can accurately track 30+ IR LEDs at 60hz from several feet away in a visually noisy environment with < 1% CPU load, tracking two eyeballs an inch away should be much easier even at 200hz.
Wowfunhappy
Member
Huh. There's nothing there, for me.
****
Decided on a whim to try playing without my glasses on... wow, the headset is MUCH more comfortable now. It doesn't look as good as it used to, of course, but on the whole it's worth the tradeoff. Especially since not having lasses allowed me to adjust the lens distance to increase my FOV.
Plus, now I can't see the screen door effect at all.
GuitarAtomik
Member
Yeah, people always want to jump straight to 4k during these resolution discussions, but there is PLENTY of room in between we can baby step to until we have the GPU power to actually do that. The screen resolution will (very likely) improve with each gen, but it will be a gradual improvement.
On a sidenote, I tried the sleep mask mod to replace the foam pad and wow. It dramatically improved the sweet spot for me to the point that I don't see any blur on edges at all. I used to have to have to fidget and adjust the hmd a lot to be able to look directly at the floor without blur but now it's all clear. Though I heard mileage may vary, it's cheap and easy enough that you should all try it just in case.
Just seems logical to double the res in each axis, especially as you can drive a car through the gaps between the pixels currently so I don't want half measures.
Would a regular RGB stripe layout improve perceived resolution btw?
Probably because 4K/120hz is the ideal
Guardian Monkey
Member
HDMI yes there could be issues. Displayport can run 4k @120
ShadowSoldier89
Member
strapped on 5 pound wrist weights and did the new longbow. (this was after playing 2 songs on audioshield and playing slingshot twice). Made it to wave 28 and my arms want to fall off
Zalusithix
Member
Given we're talking VR, that's per ~4k x 4k per eye. So that'd be an 8k traditional screen at 120hz. Not happening. Not even at 90Hz. Would need compression or some sort of non-standard transmission with foveated rendering.
Guardian Monkey
Member
Who's talking about 8k? I'm talking 4k ...1980 x 2160 per eye just like how the screens now are 1080*1200 (2160*1200 total)per eye with the CV rift and vive now. Yeah 8k is a ways off. But 4k is just around the corner in a couple years.
Edit: Sorry did see klaws response further up. Lol 8k is a ways off I am not sure if the next go around of GPUs will handle 4k max in a couple of years ... The 1080 ti and Titan aren't going to be able to run everything at 4k 60fps mos likely. So expecting in next gen to run 3x faster is a push. Yeah they are saying they are looking at 1 pass rendering but we don't know really how well that works in vr.
Zalusithix
Member
Because this is VR, when people are talking 4k they are typically meaning per eye. Not combined resolution.
Edit: But yes, 2k per eye is feasible and would be the sensible next step IMO.
Tumle
Member
I noticed that in games not made for VR.. The resolution is worse.. I think it's because of high detail textures in the background making things look more grainy.. In games like jobsimulator I hardly notice the Sde, and I think it's because of the plain bright colour surroundings.. Is it possible to change the texture lod on project cars and elite dangerous for far away objects?
d00d3n
Member
I just had my first power cord pull in the middle of a hectic session of Holopoint. It felt awful, but it is understandable that it happens considering how much you are rotating and stepping over the cables in that game. It is a fantastic game though, especially how the dodging mechanic elevates the experience compared with just shooting arrows (which would probably be quite mundane by itself).
Is Holopoint the most physically demanding game for the HTC Vive? I almost feel like I need a headband to make it viable to play for longer sessions.
Is Holopoint the most physically demanding game for the HTC Vive? I almost feel like I need a headband to make it viable to play for longer sessions.
AwesomeMeat
PossumMeat
I will occasionally unplug the HDMI, USB, and Power out of the back of the box if SteamVR has trouble getting everything connected. Works for me every time. Maybe give it a shot.
I don't know if order matters, but I always plug in my HMD and have it in proximity of the lighthouses before starting SteamVR. Then, after the lighthouses wake up, I turn on the controllers. Seems to be pretty reliable for me.
4k per eye doesn't mean 8k total. Two 4k screens combined have 1/2 the number of pixels as an 8k screen.
T
thepotatoman
Unconfirmed Member
The Solus Project exits early access tomorrow and seems like it could be really good. It's a story driven survival adventure game.
From what little I played so far in earily access, seems like the controls are a little confusing at first and the graphics requirements are really intense, but the concept is definitely cool, and it should be a very lengthy and meaty VR game which Vive definitely needs more of.
Movement is on the right track pad. Touching it walks you forward like a normal video game, clicking in teleports you a few feet in the direction you click, so you can basically get around by a thousand small teleports, which is a little closer to walking in real life than something like Vanishing Realms, but can also be a little more annoying for long distances.
Kinda amazing how so many games have such different ways of handling locomotion. I think this one fits decently well.
From what little I played so far in earily access, seems like the controls are a little confusing at first and the graphics requirements are really intense, but the concept is definitely cool, and it should be a very lengthy and meaty VR game which Vive definitely needs more of.
Movement is on the right track pad. Touching it walks you forward like a normal video game, clicking in teleports you a few feet in the direction you click, so you can basically get around by a thousand small teleports, which is a little closer to walking in real life than something like Vanishing Realms, but can also be a little more annoying for long distances.
Kinda amazing how so many games have such different ways of handling locomotion. I think this one fits decently well.
T
thepotatoman
Unconfirmed Member
4k is 3840x2160
8k is 7680x4320
4k per eye would be 7680x2160
So same as 8k in pixels per row, but still half of the number of rows. I don't see how changing the aspect ratio changes that fact. 8k is always going to be 4k squared, so no version of 4k times 2 will make it like 8k
8k is 7680x4320
4k per eye would be 7680x2160
So same as 8k in pixels per row, but still half of the number of rows. I don't see how changing the aspect ratio changes that fact. 8k is always going to be 4k squared, so no version of 4k times 2 will make it like 8k
DizzyCaffeine
Member
So much this. My girlfriend and I played Longbow with the new endless mode for almost 3 hours tonight. My arms are sore, but It was worth it.
Zalusithix
Member
The aspect ratio we're using in these VR sets is essentially 1:1. So "4k" in VR would mean 4k x 4k per eye which is 8k x 4k combined - which is effectively the "traditional" definition of 8k. As such, we're technically comparing 2x 4k @1:1 to 1x 8k @ 16:9. Apples and oranges, but the only reason we care about the traditional definition at all is because we're talking about the bandwidth available to DP/HDMI. The common limitations of those are specified in standard 16:9 terms. IE: DP 1.4 without DSC only supports 8k (16:9) @ 30Hz, so it thus cannot do 2x 4k (1:1) VR. Meanwhile 2k VR is close to 4k traditional, which is totally possible at even 120Hz.
Everything would be less ambiguous if we computed out the raw bandwidth used for the given explicit resolutions, but that'd likely just confuse people with all the bit depths, blanking methods, etc. Easier to just point to the nearest standard.
- Status