Forum Discussion
Regarding the flamewar/reasoned debate on position tracking
Correct me if I'm wrong, but I have inferred that the IR camera for positional tracking has absolutely no bearing on detecting rotation, correct?
Because if so, there is absolutely no need for it to be able to detect the orientation of the rift, and only needs to be able to see where it is in space. this means that there are far fewer limitations on the positioning of the IR markers on the HMD. So, what's stopping me from affixing my own IR markers to the back of my head, so the camera can see them regardless of what direction I'm looking?
It seems to me that the whole question over whether to add another camera, or whether to add some new exotic self-contained positional tracker from space-DARPA is rather unnecessary. if the camera doesn't need to figure out the orientation of your head, then a sufficiently large constellation of IR markers should be entirely adequate.
Edit: Meant to post this in General Discussion rather than General Development. If one of you mods feels like it, go ahead and move it.
Because if so, there is absolutely no need for it to be able to detect the orientation of the rift, and only needs to be able to see where it is in space. this means that there are far fewer limitations on the positioning of the IR markers on the HMD. So, what's stopping me from affixing my own IR markers to the back of my head, so the camera can see them regardless of what direction I'm looking?
It seems to me that the whole question over whether to add another camera, or whether to add some new exotic self-contained positional tracker from space-DARPA is rather unnecessary. if the camera doesn't need to figure out the orientation of your head, then a sufficiently large constellation of IR markers should be entirely adequate.
Edit: Meant to post this in General Discussion rather than General Development. If one of you mods feels like it, go ahead and move it.
3 Replies
- It's more complex than that. As far as I understand it, the camera is not the primary data source for anything. The gyros and accelerometers are what provide the immediate low-latency high-frequency tracking (both for orientation and position), and the camera is used as a high(er)-latency low(er)-frequency source of calibration data (for both position and orientation) to correct for the drift inherent in MEMS tracking.
- Unless the camera has accelerometers built in (like the Kinect does), the camera needs to be able to detect the rift's orientation relative to the camera, in order to work out the camera's orientation. But it only needs to do that once, so it wouldn't need to track orientation when facing backwards. You would need only 2 LEDs on the back to track position (so you can also get distance), since the orientation would be known.
- It will serve as a good replacement for the still error-prone and not always usable magnetic calibration.
Better yet it should allow the magnetometer to automatically calibrate itself as you use the Rift, since the camera provides a stable ground truth that it can use as a reference. It could even account for magnetic distortion at different head positions. Then the magnetometer can take over in the rare case that the camera loses you for a significant period of time.
