John Carmack's VR head set

Oculus Rift sucessfully kicstartered itslef ~6 hours ago, and it managed to achieve its goal in just 3.5 hours!
http://www.kickstarter.com/projects/1523379957/oculus-rift-step-into-the-game?ref=users
0c7ab132719971110b7c6ff2a833d7dd_large.jpg



Price for SDK model is excelent!
$275 for DIY kit, $300 for assembled version. They have announced software support from Epic, Valve and Unity, and their SDK will continue to grow with coming months. DIY kits [all are sold] will be shipped in November, and assembled versions in December.

Packaging is much better, you get motion tracker, but internal screen is still the same [1280/800, half for each eye, 90/110 FOV].
 
What you gain in immersion by moving your head to look around you lose because you're still pushing buttons to walk around.
How about using pressure sensing mat for input:
- rate of stomping in place = speed of movement
- assymetry of pressure pattern = direction
(more pressure at toes - go forward, more pressure at heels - go backwards)
- kneel touchdown = crouch
 
Oculus Rift sucessfully kicstartered itslef ~6 hours ago, and it managed to achieve its goal in just 3.5 hours!

Price for SDK model is excelent!
$275 for DIY kit, $300 for assembled version. They have announced software support from Epic, Valve and Unity, and their SDK will continue to grow with coming months. DIY kits [all are sold] will be shipped in November, and assembled versions in December.

Packaging is much better, you get motion tracker, but internal screen is still the same [1280/800, half for each eye, 90/110 FOV].

Sounds very good. Hope it will be out soon!
 
How about using pressure sensing mat for input:
- rate of stomping in place = speed of movement
- assymetry of pressure pattern = direction
(more pressure at toes - go forward, more pressure at heels - go backwards)
- kneel touchdown = crouch

That would help me loose so much weight lol. It sounds like that Micheal Douglus movie where demi moore rapes him. In that one of the plot points was that he had to use a vr set up his company developed to get information on what they wre doing. The vr set up allowed him to walk along on this treadmill type device
 
They have ball like treadmills that can go any direction. It would work great. It would also cost a bajillion dollars :)
 
Looking around on youtube the most functional systems consist of lots of looped strips with powered rollers for the loops on either side themselves being strung together in a loop. This shows the concept well, and this is a little more refined implementation.

That said, I seriously doubt it's going to feel natural.
 
how about a series of kinects with higher res sensors and their own cpu's arranged in a circle around you
 
His enthusiasm during the keynote really got me to the point where I want VR gaming myself...but it seems that the tec is still far away. I think he is right, that one of his major concerns is simulation sickness. The smallest movements of your head need a consequence in the simulated world.

But what about eye movement? What about eye movement and focus? Seems to be the most difficult thing...

Up to now, the only thing close to a VR gaming experience I had was in the context of scientific simulation/visualisation, where I could watch and go around a three dimensional static(!) snap shot of a turbulent flow past a car, with smallest interaction like changing some values...for this, not only head sets are needed but also a 'cave' full of projectors and 'other stuff'...
 
Last edited by a moderator:
His enthusiasm during the keynote really got me to the point where I want VR gaming myself...but it seems that the tec is still far away. I think he is right, that one of his major concerns is simulation sickness. The smallest movements of your head need a consequence in the simulated world.

it seemed all his major concerns have near term solutions as explained by john himself.

1. screen latency (OLED)
2. screen rez (OLED)
3. tracking/positioning ( he talked about multiple solutions).
 
Latency is mostly in the rendering ... not the screen.

Ideally you would render with a significant safety margin around the edges of the screen and use image processing to handle small rotations of the head ... this way you could respond at near 1/refresh seconds to any change in head orientation.
 
He claims he's got 3 frames of input+rendering delay max in the Rage engine. Running at 60FPS that's 50ms. Screens add that much or more. Also there's a question whether we're talking about input-to-screen delay or delay of the entire simulation (physics, AI, animation) - the latter is usually more significant than the former, isn't it?
 
Screens add that much or more.
In general they don't, most modern LCDs have 1 or less than 1 frame latency. For every frame of latency you need a frame of storage ... and that costs money. There is no good reason to have more latency than that for a monitor (for a TV with motion compensated up conversion maybe).
 
Carmack's numbers are different. He said it's because of the pipelining nature of modern "interactive" TVs: you get the data, latch it, do the down/upscaling, latch it, put on-screen display on it, latch it,... He measured and saw latencies as big as 80ms. Do I believe him? Sure. Are there TVs that do better? There's no doubt about it too. I would go and find him saying that in the 3h-long presentation, but I've got better things to do, really. ;)

//edit

VR panel from QuakeCon 2012 is online if anyone's interested.

http://www.youtube.com/watch?v=8gaqQdyfAz8
 
Last edited by a moderator:
Latency is mostly in the rendering ... not the screen.

Ideally you would render with a significant safety margin around the edges of the screen and use image processing to handle small rotations of the head ... this way you could respond at near 1/refresh seconds to any change in head orientation.

you should watch the 3 and 1/2 hour speech :LOL:

he said that if your walking forward in tech5 there are 2 frames of latency if you shoot or do some other more complex things there are 3.

he was talking about latency created by "smart" tv's, they tend to buffer frames etc
he was talking about real world latency on LCD with stereoscopic creating ghosting (he said oled basically doesn't have this problem)
he was talking about latency in tracking head position.
he was talking about latency on 120hz vs 60hz comparing console controller to mouse to headset.

Latency of tech 5 didn't even get mentioned as an issue.
 
As I said tv's can buffer a few more frames because they do more complex conversions, but this has little to do with LCDs ... even overdrive can (and is by some manufacturers) be implemented with less than 1 frame latency. OLED changes nothing about latency ... and suddenly calling ghosting latency is stupid.
 
If I remember correctly, he said that Rage has a 1 frame latency (so everything happens in that 16.7ms of 1 frame at 60fps) to your movement response. However, when you shoot, the animation for shooting has to be started up and that adds 1 frame of latency.

The bit about OLED was about the speed at which OLED draws and clears pixels, which is if I understand it correctly in the order of 1000x faster for OLED vs LCD. This does in fact still reduce lag as a result (time from action to pixel change), but also improves the quality of HMD stereoscopic images (less artifacts, ghosting, etc.).

At least, that's what I understand (among others from listening to the whole speech).

Apart from that he discussed (in a user question?) that it would be great if you could somehow get a signal in the HDMI CEC spec that tells the TV not to do any post-processing whatsoever, that games could use to tell the TV to leave the image alone and focus on getting the best response time.
 
Screens can have severe latency, up to 50ms and sometimes more, depending on the electronics.
On serious screen reviews you'll see delay compared to CRT, screens with 0 delay are *very* rare.
 
are modern LCD shutter glasses much improved, compared to those from the late 90s?
there was quite a lot of ghosting when playing stereoscopic on a CRT monitor. one of the fun and not bugged games was quake 1, which gives giant ghosting any time you're looking at torches.

in this context, ghosting is each eye seeing what is meant for the other eye. which is something even an LCD HMD would eliminate.
 
Back
Top