I thought I had ironed out all the issues with the ReacTIVision fiducial CV system last week when I discovered the new update of the software, but I've hit another snag. Maybe someone has an idea to solve or work around this problem. The problem is that the new computers all have Thunderbolt ports rather than native Firewire and ReacTIVision does not recognize the cameras once they are connected with an adapter. I didn't encounter this issue with my earlier testing because my computer is old enough to still have a Firewire port. Interestingly, Max has no problem seeing the Firewire camera when it is connected through the Thunderbolt adapter, so it is not just a power issue or complete incompatibility. The simplest solution I can think of is to purchase new cameras that are compatible with both the computer at a hardware level and with the ReacTIVision software. Tracking down cameras that we have a reasonable guarantee of working shouldn't be too difficult, but I don't know if we want to prioritize this as an investment or not. The key (at least for my camera-projector system) is to have a camera with a variable zoom so that the imaging sensing can be closely matched to the projected image. For an overhead tracking system, the key factors would be a wide angle lens, high resolution (to minimize the required size for the fiducials) and reasonably high speed (for stable image tracking).
Another solution to this is to simply used older computers that have firewire ports so that we can use existing cameras. This would certainly be feasible for the camera-projector system. We would have to test this out for the overhead tracking system. The best thing to do of the overhead system is to try out the existing camera arrangement to see how well it works with the fiducials. I can try the out this afternoon (assuming the space is open for testing).
In general I think a fiducial based computer vision system is a good way to reliably track individuals and to afford conscious interaction with the system.
Best,
Byron
On Sun, Nov 9, 2014 at 8:19 PM, Sha Xin Wei <Xinwei.Sha@asu.edu> wrote:
After the Monday tests of hardware and networking mapping to lights in iStage with Pete, Omar and I can code the actual logic the rest of the week with Garrett, and maybe with some scheduled consults w Evan and Julian for ideas. Omar’s got code already working in on the body lights. I want to see it working via the overhead lights beaming onto the floor. For this game we need to know who’s who.Byron, Ozzie anyone: What is the easiest way in the iStage for us to track individual identities? Not blob because they’ll confuse.Byron: IF indeed you’ve got the fiducial code working. Can you show this to Omar (and Garrett) so they can work on the Darkener and Lighteners ecology with me? Can we have people carry fiducials visible to the camera from above? Maybe they can place a fiducial next to them in order to draw light. Can the fiducuals be printed in a legible size that is not ludicrously large? How about if we print a bunch of fiducials as “LIGHT COINS” of three different denominations — RGB — each good for a certain amount of elapsed time. Exposing a fiducial to the camera charges it against your account (it can be used only once per session). Maybe waving it can increase the intensity but shorten the time (reciprocal constraint). People can then choose to band together or in the +/- version, subtract etc.