Forum Discussion
Multiple Cameras & Increased Accuracy/Range of Motion?
Just thought about multiple cameras with the DK2. After using the Demo Scene "Camera Bounds" for a while I realized that having multiple, opposing cameras would increase the accuracy and range of motion. There is the issue of multiple sync cables, though....
And sure, it might not be the optimal way to go, versus Valve's "VR room", where I'm assuming the camera is integrated in to the HMD and senses the unique patterns on the walls and ceiling, etc).
Which also makes me wonder something else.I didn't see any mention in the documentation for what the USB + 2.5mm jack on the HMD is for.... Some way to mount + connect the camera to the HMD? Or simply hardware debugging features? Serial firmware access?
And sure, it might not be the optimal way to go, versus Valve's "VR room", where I'm assuming the camera is integrated in to the HMD and senses the unique patterns on the walls and ceiling, etc).
Which also makes me wonder something else.I didn't see any mention in the documentation for what the USB + 2.5mm jack on the HMD is for.... Some way to mount + connect the camera to the HMD? Or simply hardware debugging features? Serial firmware access?
6 Replies
- It has been mentioned that the usb port on the HMD is for developers to plug in other USB devices (eg leap motion), and that the 2.5mm jack is a dedicated power for that usb jack. That is if i remember correctly.
- With all of the recent hires the past year I suspect the separate camera is a just convenient and cost effective step towards a hybrid solution.
I'm wondering if a 360* IR texture/grid projector, and a Rift mounted camera, would achieve a more consumer friendly solution (with more generalised/cheaper processing) of Valve's room above? It could be a nice reference for optical assisted peripherals too... - Personally I think cheap integrated magnetic tracking is the way forward. No occlusion, full 360 and highly accurate these days when combined with an IMU as in the STEM.
This worked superbly in the Virtuality machines in the '90s but was too expensive for home use. Although there was a cheap(ish) internal PC card solution introduced in the later systems which I'm sure with today's microprocessors could be produced in large quantities at an affordable price point. "V8Griff" wrote:
Personally I think cheap integrated magnetic tracking is the way forward. No occlusion, full 360 and highly accurate these days when combined with an IMU as in the STEM.
This worked superbly in the Virtuality machines in the '90s but was too expensive for home use. Although there was a cheap(ish) internal PC card solution introduced in the later systems which I'm sure with today's microprocessors could be produced in large quantities at an affordable price point.
I agree with this.
Either that or some sort of inside out tracking which uses a 360 degree projected IR dot pattern (Palmer I think talked about this at some point on certain forums).
Personally I'd prefer moving away from optical systems as they have pretty well known robustness issues, but I think inside out tracking with a 360 degree IR dot projection would be alright. Not as good as magnetic tracking but at least good enough for CV1 (I don't think the single camera is, and multiple cameras are a big hassle)."V8Griff" wrote:
Personally I think cheap integrated magnetic tracking is the way forward. No occlusion, full 360 and highly accurate these days when combined with an IMU as in the STEM.
Absolutely, earlier I was trying to emphasise an optical assisted hybrid solution - introducing some level of absolute positioning in addition to the relative.- Crescent Bay's tracking camera apparently has a larger tracking volume (people report standing *and* sitting down in the same demo!), and of course it now tracks your head for the full 360 degrees. Amazing feat given different head sizes and potential strap placements.
