3 ms·
Tested have an interview/hands-on where they discuss why Oculus have used an optical approach with the external camera to track head movements. They note that i
by prawn 13y ago
Tested have an interview/hands-on where they discuss why Oculus have used an optical approach with the external camera to track head movements. They note that it was superior to various other methods they tried. Wonder if eventually it could work with typical laptop web cams?
http://www.tested.com/tech/459692-ces-2014-hands-oculus-vrs-crystal-cove-prototype/ http://www.tested.com/tech/459692-ces-2014-hands-oculus-vrs-...
Also worth watching that video to hear about the switch to an OLED screen and the latency advantage there. Employees at that company must have so much fun trying all the alternatives to build the best product.
(My dev kit in Australia is for sale if anyone is interested. Email in profile. It's amazing, but I'm curious more than I am a software developer so it might be more useful to someone else.)
- YZF 13y agoA similar approach could be used with a typical webcam. In their case it's probably a camera with a filter so they can isolate their IR LEDs. That's kind of a nice to have but it's not absolutely necessary. It's neat how they can get by with a single camera. Note the pattern of LEDs, as you move around they can use the pattern to locate you in 3d space. IMO the downside of using a camera is that you automatically get some latency as part of the deal. A full frame has to be read out, digitized and the 3d position derived. Some other systems could have better real time characteristics. The cost of this system is presumably a factor.
- knowaveragejoe 13y agoWith an embedded camera system you could be transmitting just the positional data to be fed into the computer after doing most/all of the work in hardware, no?
- YZF 13y agoIf the image you get out of the sensor is processing friendly (such as just the image of the IR leds) then you can do this pretty quickly in software. Some preprocessing on the camera can help but the source of most of the latency is going to be the image sensor readout. I've actually worked on a real time tracking system like this about 20 years ago.
- cma 13y agoYou can use sensor fusion to combine it with acceleromters, which have terrible drift double integrating from acceleration to position, but can actually manage it in the few-milliseconds time scale between frames/signaling/processing of the camera.
- YZF 13y agoYes. That should work... I'll add it to the patent ;-)
- prawn 13y agoIs it likely that they'll have to drastically change their methods to handle standing use? The interviewer noted that leaning forward a lot or turning around would hamper tracking. I had hoped for a wireless unit by the time of the consumer release but I guess driving the screen might chew through battery? Not to mention that a battery would add complexity and size to the unit.
- YZF 13y agoPresumably if they have a wide enough angle and a high enough resolution they can cover a much larger space. It's symmetric situation, you need just as much resolution at the distance as the minimum amount you want something in the VR world the same distance from you to respond to. A well designed system can even resolve sub-pixel motion of those LEDs in the field of view of the camera.
- skeletonjelly 13y agoHow much are you asking for it? With regards to IR, I suppose it would work with a regular webcam.
- prawn 13y agoDrop me an email. Thanks!