augmented reality

The issue like everything with oculus would be latency. The camera recording, sending the signal to a processor, sending it back out, all adds latency. Recording 60fps video at 1080 is not a common thing for cameras.

As well, just as spacing is different for each person's eyes, just like you have to set up the spacing in a game, you'd have to do the same to the cameras physical position.

It would be cool if it worked right, but there are problems, like what I mentioned.

You would feel like you were walking around drunk with the latency induced by any webcam.

Same concept with most car backup cameras. There is slight enough delay in them for someone like the *coughwifecough to hit cars in parking lots. But then again that could just be the driver ;)

IF you could reduce the latency to the point where it would be like looking through glass then its feasible, but then you are still contending with viewing angles and being able to see left and right. If you have seen the Battleship Shredder car that they built on Dream Machines they used monitors with outside mounted cameras to see outside and found it VERY hard to maneuver without hitting anything in the process lol.

@CaliberMengsk you would only need a 640×800 stream from the cameras

I'm sure it would be possible to reduce the latency to a level where you can get used to it with the right hardware.

I also thought about that. I really think, there is a need of a kind of combined VR and AR Headset which can switch modes, because I guess for the normal user, it would be rather expensive to buy a headset for VR and one for AR.

There are 120FPS/150FPS webcams available which should reduce the latency to make this do-able!

"peterept" wrote:
There are 120FPS/150FPS webcams available which should reduce the latency to make this do-able!

The FPS of the webcam is unrelated to the amount of latency in the system, and instead refers to how much information it can capture. Processing it for return to a display adds overhead, in addition to the overhead the camera already may have.

Other things that apply when looking for camera systems for an AR interface include FOV, which is very important, pupil depth, eye spacing, mounting angles, and all sorts of exciting things like that. The goal of AR through VR is to be able to as accurately as possible reproduce the vision of a person through VR. Right now, there are problems with it because with the equipment I've played with, you end up with an effect that appears like you're looking through a backwards telescope that moves half a second after you do. Natural, wearable computing it tends not to be.

Anyone interested in these problems should totally read Mike Abrash's blog. The amount of information he's releasing from his own research is amazing.

The processing bit is largely being addressed by this project: http://opencv.org/

They have very fast algorithms running on GPUs. I'm sure they also have recommendations for video input devices.

J Carmack's take on latency: http://www.altdevblogaday.com/2013/02/2 ... trategies/
M Abrash's blog on virtual reality: http://blogs.valvesoftware.com/author/mabrash/

Ok so the main problem seemst to be the latency.

The first thing i found was this framegrabber which uses "Ultra Low Latency technology with a latency below 40ms" but there is no information on the price and you would also need 2 analogue cameras and something like a PCI to PCI/104 Adapter.

What do you think of the PS3Eye? According to this thread you can get pretty good latency values out of it, the lens is replaceable, it uses USB and it only costs round 20$. Only the picture quality would not be very good.