I also have been really impressed with the fact that iPi Studio has been putting out continuous updates since I bought my license. The Intel HD4000 in my laptop does something like 16s/frame or more so it's only useful for checking if the calibration recording was ok. Integrated graphics cards take too long to be practical though. I've used ATI 4870, nvidia GTX 275, and GTX 560 ti and there's only around 0.1-0.3 seconds/frame difference between them. Processing does require a graphics card, but you don't need a state-of-the-art card. You only have to record the depth stream from each camera and you don't need any special lighting. Capture with 1 Kinect sensor requires even less. Any recent $500 i5 laptop with USB2 and USB3 ports and Windows 8 can run the iPi Recorder software and handle 2 Kinect sensors (I've had trouble with 2 sensors in Win7). I spent some time trying to combine the capture data myself from two Kinects and it is not a trivial task. With two kinect sensors 180 degrees apart this is not an issue. With just a single sensor there are plenty of free SDKs out there, but there's no way to track a body part that is not in the camera's visible range. One of the reasons I picked this software is because when I purchased it, it was the only software I could find that could combine the data from 2 Kinect sensors. The very quick calibration process and no need for any markers on the capture subjects makes it a real pleasure to use. I use motion capture in a research environment to study how people move and maintain balance. I've been working with this software with 2 Kinects (basic license) and a more expensive marker-based system that uses 8 cameras for the last 6+ months and while the marker-based system is more accurate, the iPi system is so so much more pleasant to use.
0 kommentar(er)
