Overlapping Grid Setup for a one computer system
--------------------------------------------------------------------------------
Video Jump Discussion Board: Video Jump Discussions: Overlapping Grid Setup for a one computer system
--------------------------------------------------------------------------------
By Dave Clark on Tuesday, July 9, 2002 - 06:46 am:
Bob, I wanted to explicitly share with you my experience and suggestions with regards to one specific point.
I've attached the layout of the Tianjin site the way we've planned it for VJ measurement. What we have planned here, are two camera views. The "Far Side" of each view is marked by two buoys out of a set of six buoys which have been laid out along a line 19m to the left of the ramp centerline in the skier landing area, at uniform 15m intervals (0-15-20-45-60-75). If you study this a little, you will see that we only really need four buoys over there instead of six, but when I was doing the planning I found the four I was getting were going to be at 0, 29, 45 and 74 downcourse, and I recognized that those were falling on almost perfectly regular 15m intervals, so we put six out there and made them exactly regular, for the crowd's edification. Actually, that will provide us two check buoys for each view, which can't hurt at all.
You can see that the way the grids have been laid out, the four reference buoys will be right near the edge of the screen in each view, and the camera views are each just about exactly 25 degrees wide, so there will be no possible issues about lens distortion at all with this setup.
In terms of jump distances, the short view will be able to see distances from minimum out to about 44m, and the longer view will be able to see distances from about 31 meters out to about 74 meters. At the planned camera elevation of 11 meters above water level, this layout completely satisfies the class "R" sensitivity requirements at all points beyond about 15 meters -- the very shortest jumps have a little bit of a cross-the-grain problem in the short view -- but then again nothing of great significance happens back there, and that sensitivity is still well below the "C" limits and hence far better than you'd get with johnson meters anyways.
So by deliberate design, there is a range from between 31 to 44 meters, which is contained within both of these two views. I've done some experimentation with such layouts, and once each view has been verified with it's respective check buoy(s), we find that measurements in the overlap zone between two separate computers typically come up with the exact same result in both meters and feet, and occasionally will differ by 0.1 meter or one foot -- rarely more (and then if you look we've found a systematic difference in the way the marking was done by the two operators).
I've also done some more detailed testing with static buoy images in the overlap zone of such a layout, and once the views are verified and proven to read true through their respective check buoys, the comparitive readings on those static buoys in the overlap zone are for all practical purposes identical, differing at most by one half the per-pixel horizontal movement amount -- or within the limits of error of the underlying process, in other words.
So within that overlap zone, I believe that a completely accurate -- and hence official -- result could be obtained from either viewpoint. What this leads me to, is the desire to run such a setup with only one computer and two VCR's and an A/B switch. We've actually been doing such a framework at one site here in our region for the past three years, where we have a 35 foot overlap zone between 100 and 135 feet. Our rule of thumb is if a skier has a seeding score or rating indicating they have jumped no more than 120 feet, then we watch them in the shorter view, and to this date I cannot recall a single instance where we had a skier outjump the far end of that short view. On the other hand, if they've jumped more than 120 in the past, we switch to the long grid and rarely have we found a jumper land short of the short end of that long grid -- and in that extremely rare instance, we can just throw the switch to the other grid and play back the other tape, which is not a painful remedy.
So when the grids have been laid out in such an overlapping fashion as this, I think that two cameras and two VCR's and a switch and one computer is a perfectly acceptable equipment configuration for such a setup. From a sponsor or organizers standpoint, that is lots less intimidating than three cameras and three VCRs and two computers.
