Sony rumoured to be developing VR headset for PS4 *spawn

[pjbliverpool]

I read somewhere that for VR to work perfectly you need something like 1080p at least per eye and 90fps (I guess it was Valve presentation where they talked about the right VR experience or so).

They were Valves minimum requirements for "presence" if I recall correctly. i.e. fooling your mind into thinking you're really in the game world.

Incidentally, the guy that lead that team at Valve has just moved to Occulus RIft as a lead architect.

 

[Shifty Geezer]

A single 1080p screen could be a little lowres. Higher resolution is possible - a 2560x1440 phones and tablets are coming. A 2560x1440 7" screen would be desirable. It can run upscaled graphics (720p as dual 1080p is beyond PS4's capabilities, unless it's rendering last-gen visuals), but the higher resolution will help eliminated subpixel artefacts.

 

[DieH@rd]

Will they be going with an OR clone, a single large screen and lenses? Or dual smaller (3.5"?) screens? I doubt we'll see their tiny OLED display as its too expensive.

According to the 3rd party dev source who has talked to RoadtoTV and is supposedly making a game for this VR system, Sony solution has FOV simmilar to DK1 [and much higher rez]. That heavily hints that they are using big screens and not some microdisplays with complicated lensing.
http://www.neogaf.com/forum/showthread.php?t=763061

 

[Cyan]

I'm actually really looking forward to trying it out, and I think I would get a VR headset that was affordable, but I can't imagine using it as my primary way of gaming. You're too cut off form the world, especially if you're also using headphones. You would have no idea if your phone was ringing. You wouldn't hear the doorbell. Living in a house with other people, you'd be totally unaware of what they were doing around you. Same with pets. What time of day is it? Who knows. Did you miss dinner? Who knows. Just doesn't seem like a healthy way to do all your gaming. That said, would it be fucking awesome to play a shooter on PS4 with a VR headset for an hour? Yes.

That sums up my concerns about your typical VR device -I am also following Shifty's logic to some extent-, which I expressed months ago when talking about using Kinect and a VR device, or similar.

Personally, it would also be important to me if the VR device could be calibrated. I get frequent migraines in too bright environments, which leads me to not be as articulate as I'd like to -like trying to say dog and saying lob, for instance- and my vision gets blurry.

I think VR might well be the future of videogames, although I expect to be realised not only with the classical VR display that looks like a mask, but regular glasses.

 

[jonabbey]

Re: the Dual Shock 4 connector, they can't possibly have the PS4->Dual Shock 4 bandwidth high enough that it's going to be streaming video through it, can they? I always assumed the connector was for keypads and the like.

 

[Brad Grenz]

I read somewhere that for VR to work perfectly you need something like 1080p at least per eye and 90fps (I guess it was Valve presentation where they talked about the right VR experience or so).

Can GGs interlace interpolation trick help out the PS4...as it seems to me that PS4 is missing the computational power to proper support VR with shiny graphics (heck, even Dead Nation is 'only' 1080p/30fps).

I would think the artifacting possible with the Killzone MP reprojection technique might be exponentially more distracting in a VR context. But I think it's a mistake to assume VR games on PS4 would be targeting photo-realism in that way. A game doesn't need to be photo-realistic to be immersive. There will surely be tons of Tron-style cyberspace game environments, or other abstract visual designs. Or imagine exploring a Ghibli animation recreated in 3D like Ni No Kuni only in VR.

And still, think of all the resources that are spent on camera effects, DOF, motion blur, chromatic aberration, etc. You wouldn't bother in a VR game because you're not simulating a camera anymore. You should be able to do realistic graphics well above PS3 level on the PS4 in VR.

 

[Deleted member 11852]

Re: the Dual Shock 4 connector, they can't possibly have the PS4->Dual Shock 4 bandwidth high enough that it's going to be streaming video through it, can they? I always assumed the connector was for keypads and the like.

Surprisingly Apple's AirPlay works over Bluetooth so that may not be completely impossible but I'd be surprised. Surely it would be better to let the headset itself be the communicator, rather than the DualShock4.

 

[DieH@rd]

I read somewhere that for VR to work perfectly you need something like 1080p at least per eye and 90fps (I guess it was Valve presentation where they talked about the right VR experience or so).

It was actually 1Kx1K per eye, and the 90hz strobing that could possibly go as low as 72hz [not fully tested]. That was needed for full "presence", it was not needed for awesome VR immersion.

Sony is limited with resolution and image bandwidth because of HDMI on PS4. They will use 1080p screen and maybe 72Hz strobing [Crystal Cove used that kind of refresh rate apparently].

 

[loekf]

Maybe we'll finally learn what the DS4 front connector was planned for.
And hopefully a new Move controller, with their weird wire-mesh tactile feedback.

Front connector => charging ???

Please no silly useless keyboards anymore.

I'm praying for a big firmware update:

- DS4 light can be switched off
- Better BD playback incl. 3D
- CD audio / MP3 / AAC / FLAC audio
- Support for DLNA / SMB
- More voice commands and available in-game
- Opening up their BT stack for third party controllers, e.g. Logitech Harmony
- More SmartTV-like suppliers, I don't understand why in e.g. Holland there's no NPO-app to watch shows already broadcasted (is available on SmartTVs, Android, iOS etc)

 

[loekf]

Surprisingly Apple's AirPlay works over Bluetooth so that may not be completely impossible but I'd be surprised. Surely it would be better to let the headset itself be the communicator, rather than the DualShock4.

AFAIK, Airplay uses Wifi. There's no way you can send 1080p30 video over Bluetooth.. with its bitrate of ~2 Mbps.

 

[JPT]

Latest update adds supports for Bluetooth, but that might just be the "controll signals" and the moves are streamed over Wifi.

 

[RedVi]

Regarding the Sony device, I'd say 1080p screens is almost a given considering that's the target resolution of the console (higher would be useless) and lower would limit some games. That said, dual 1080p feeds at decent framerates (which in VR terms should be 60+) shouldn't be expected with anything more than the most basic graphics, i.e. last gen level at best.

Why would higher than 1080p be useless? If the problem is the pixel size and gap between the pixels, then the more the better. It's not the image quality that is the major driver for higher resolution in VR, it's how easily discernible the sub pixels are. Up-scaling from 1080 or even 720p to 1440p would look fine IMO. Put your face within 5 inches of a 20-27" 1080p screen and you will see the problem, the pixel quality degradation that up-scaling would bring is far less of an issue.

 

[Billy Idol]

I called it in the first PS4 gameplay vid we saw:

The DS4 light is really annoying as hell...why in hell did nobody at Sony notice this? Cerny...?

 

[DieH@rd]

IMHO, Sony should nudge [or push] developers to adopt cropped rendering of VR. So much space is wasted today on rendering pixels that will never be seen by users. This cropped rendering can even be optional, leaving devs to choose whether they need more performance or not [perfect for indies who are not targeting HQ visuals]. PS4 VR quality could benefit much from this.

However, is anything preventing Sony to just take one 1080p disply, put it farther away from the lenses, and force those lenses to cover much more of the screen area??? This would give users much higher pixel density than for example CrystalCove that is also using 1080p screen [and different approach on setting up rendering, camera settings, and fishweye warping].

#3 instead of moving the screen back, just place smaller 1080p screen in front of lenses, and force engine to render "active" area across entire screen. Result will be the same as #2.

 

[loekf]

So how these VR displays work ? Does it include stereoscopic 3D or is it just a shielded single small high-res LCD in front of your eyes, with a bunch of accelerometers so that the VOF moves with your head ?

 

[Shifty Geezer]

It uses optics to split the single screen into two half-res screens, one for each eye, providing stereoscopic 3D vision. Oculus Rift's optics clever use more of the screen for the centre part of each eye, and stretch the peripheral visuals over the outside edge where detail is less important. The game renderer produces a FOV warp that matches these optics.

The near round viewports are stretched into rectangular screens. I assume that approach is patented and Sony would have to go with something more straightforward? Or perhaps Oculus VR would be happy to license the tech?

 

[MrFox]

I'd prefer something that accept a rectilinear projection in 1080p from the HDMI, and use an internal ASIC to do all the corrections. It would be easier to port/adapt games, and the glasses would be compatible with bluray 3D or phones/tablets micro-hdmi out.

They'd need a chip that can correct everything: vignetting, strong geometric distortion, chromatic aberrations, and a full 3D LUT for the usually horrible LCD response. Sony already have some interesting asics with all these blocks integrated on one chip. They are used in their projectors and they are "in-line" processing, lag is only a fraction of a frame. They correct everything, including a 56 zones geometry correction per channel.

On glass panels (not the fancy oled on silicon from Sony), the max resolution possible is 560dpi from samsung. A pair of 3.5"x2" inch would do 1080p per eye. Even better, 3.5"x2.6" for 1440p.

With enough scene data, and using the Move-like tracking of the head, maybe they can do frame interpolation from a 30fps game, and then shift and distort the image to track the head without latency. The price to pay is a small amount interpolation artifacts, but unlike the current TVs interpolations which need to calculate motion vectors in a crude way from multiple frames, this time they have the exact scene data, and they have the camera's position associated with that frame. Heck, even a simple "camera anti-shake" algorithm would be useful, shift/rotate the image to match the head position to the current frame scene data.

All that for 999$

 

[Deleted member 11852]

AFAIK, Airplay uses Wifi. There's no way you can send 1080p30 video over Bluetooth.. with its bitrate of ~2 Mbps.

Bluetooth v3 is ~24Mbps and that's five years old (April 2009).

 

[MrFox]

Bluetooth v3 is ~24Mbps and that's five years old (April 2009).

From what I understand V3.0 is a weird hack and it basically doesn't exist in the real world. They piggy back on wifi for bandwidth, so that bandwidth doesn't even go through the bluetooth antenna, and it requires a working wifi antenna.

 

[Deleted member 11852]

From what I understand V3.0 is a weird hack and it basically doesn't exist in the real world. They piggy back on wifi for bandwidth, so that bandwidth doesn't even go through the bluetooth antenna, and it requires a working wifi antenna.

You could be right. I've not had to write code to support a bluetooth connection since about 1997. Is the same true for Bluetooth 4.0?