Why not put the camera on the oculus itself?

"frankzappa" wrote:
Well the new thing here is not the camera on the oculus. That has been discussed. My question is really the device that would send out IR dots. Why wouldn't that work? Somthing that sends IR lasers all over the room. People here are assuming you would have to place IR leds all over the room.

If I'm reading your question correctly, that would mean making the Rift heavier by a significant amount, and also you would need sensors mounted in your room to get those IR signals and then relay them back to the computer.

TBH, what they have now is the probably the best option out there. Perhaps OR could make an "upgraded camera" option so that the camera would have 2 optics; one for Rift tracking, the other as a standard webcam.

The basic problem with IR dots on the walls is that ambient daylight (which also posesses huge amounts of IR light) is order of magnitude stronger than any reasonably powered LED, and while these laser dots are fairly bright against the unlit walls, they're very faint against daylight lit walls. Not to mention that in this case simple and fast threshold filtering wouldn't work, you will inevitably get overbrighted areas, as well as spots in too dark areas being clipped out. Also, something as simple as IR-color of your walls plays a great role - it will be extremely hard to pick up these dots if your walls are IR-black and don't refract much of the incoming light. The LED that directly emits light, on the other hand, yields much more luminance and can be powerful enough to outweight even the refracted sunlight, provided IR-black matte surface of the device.

"saviornt" wrote:
If I'm reading your question correctly, that would mean making the Rift heavier by a significant amount, and also you would need sensors mounted in your room to get those IR signals and then relay them back to the computer.
...

Nah, he did not mean to place the dot field projector (á la Kinect) on the HMD, but somewhere in the room. The camera would be on the HMD.

When it comes to the bit about weight, regarding tracking camera on HMD; That doesn't seem to be a problem for CastAR; A unit that (...in itself - we'll see about the control box) is significantly less bulky than a Rift.

If I were to set up a room for tracking today, and had the skills to pull it all together :P, I would probably try to get one of TI's tracking kits, place its wide angle camera on top of the Rift (...which is rarely obstructed by the user, or entirely upside down), and put most of the LED markers (...which have known, unchanging physical dimensions, unlike projected dots) (...also; assuming I bought more than one) in the ceiling. (The actual image analysis and tracking is, as far as I know, done autonomously by the unit, which only passes translation- and transformation matrices to the PC.)

"raidho36" wrote:
The basic problem with IR dots on the walls is that ambient daylight (which also posesses huge amounts of IR light) is order of magnitude stronger than any reasonably powered LED, and while these laser dots are fairly bright against the unlit walls, they're very faint against daylight lit walls. Not to mention that in this case simple and fast threshold filtering wouldn't work, you will inevitably get overbrighted areas, as well as spots in too dark areas being clipped out. Also, something as simple as IR-color of your walls plays a great role - it will be extremely hard to pick up these dots if your walls are IR-black and don't refract much of the incoming light. The LED that directly emits light, on the other hand, yields much more luminance and can be powerful enough to outweight even the refracted sunlight, provided IR-black matte surface of the device.

Aha, thanks for the explanation. :)

"saviornt" wrote:


If I'm reading your question correctly, that would mean making the Rift heavier by a significant amount, and also you would need sensors mounted in your room to get those IR signals and then relay them back to the computer.

No dude, my idea was an emitter on your desktop that sends out a bunch of tiny IR lasers and projects dots all over the room. That way you don't have to put LED's all over your room with wires etc. However the last post made me realize why that wouldn't work.

"frankzappa" wrote:
No dude, my idea was an emitter on your desktop that sends out a bunch of tiny IR lasers and projects dots all over the room. That way you don't have to put LED's all over your room with wires etc. However the last post made me realize why that wouldn't work.

Actually, that would be an interesting approach, but unfortunately there's a significant flaw to it. If you have a desktop widget projecting beams powerful enough that their reflections could be picked up by the Rift from across the room, then anything very close to the widget is going to be so bright as to overpower the sensor, making it seem like your entire desk area is as bright as the sun (to the sensor anyway). And god help you if the Rift base camera actually intersects one of those beams... it could easily burn out the sensor. Basically imagine the difference between shining a laser on a wall and being able to clearly see it, and then shining that laser in your eye.

Still, I personally think Crystal Cove seems like more engineering than is necessary, especially if all your want to achieve is workstation based position tracking. A static camera on a desk and a QR code (or several) printed on the Rift should allow for 90% of the use cases. You could deal with varying lighting conditions by making the the white parts of the QR code retro-reflective (the same stuff they use on street signs to make them appear super bright if you shine a light at them, even from a great distance) and put a small IR emitter on on the camera housing.

Daylight wouldn't be any more or less of a problem than it is with the Kinect... There's no technical reason why putting the camera on the Rift and using a dot field projection wouldn't work. The problem is that it would be a much less robust/reliable solution. The proposed solution would be far more dependent on ambient lighting conditions (daylight hurts!), room geometry (move that bookshelf!), surface reflection (don't use the wrong paint!), and be rather more processor intensive (need way more tracking points to ensure good coverage). It would make the setup far more finniky to get working, and in the end, it wouldn't provide any advantages or benefits to the typical seated use.

Basically, for the typical use case of seated VR (and let's face it, nearly everybody is going to want to remain seated, as standing in a rift is dangerous), under ideal conditions, the proposed solution would at best work as well as the current approach, and under less than ideal conditions it would be far worse.

Someone should tell Oculus that sitting experiences are just not immersive enough ... how can you seriously consider VR for sitting purposes? Just fire up any FPS with Oculus on and see the difference between when you sit and stand. In the first case you are just playing a full 3d game, whereas when standing you trick your brain in believing you are actually there!

I guess a lot of us will have to use wireless hydras to get a full 360 degree pos tracking.

"przecinek" wrote:
Someone should tell Oculus that sitting experiences are just not immersive enough ... how can you seriously consider VR for sitting purposes? Just fire up any FPS with Oculus on and see the difference between when you sit and stand. In the first case you are just playing a full 3d game, whereas when standing you trick your brain in believing you are actually there!

I guess a lot of us will have to use wireless hydras to get a full 360 degree pos tracking.

Give it time. Take the phone for example:

- First we got a wired phone (wired, seated HMD)
- Then we got a wireless phone (Wireless HMD w/ basestation)
- Afterwards, we got portable wireless phones, ie: cellphones (Portable HMD with onboard storage + video processing)

Even with the STEM and Omni, I'd still be a bit weary of using it for the simple fact that there is a cord coming out of the headset (cords coming from face + wailing arms = bad idea)

What about having a carpet with sensors in it? Like that dance game. You could have sensors that track your feet and also maybe magnetic tracking for the HMD. Add that thalmic labs MYO thingy for the arms and you have a complete system :)