That’s a lot of movies!! 
Ok so I have also been considering resolution and here is what I came up with. A typical DVD in its case is about 3/8" thick? It might be thicker but if the DVDs were in CD cases that would be close so I’ll just assume it for a worst case. you need to be able to pick it up with “something” so let’s choose the gripper having a parallel pair of thin plates with something textured on the inside. if you can keep these plates less than 1/8" thick and have a tapered end then you could possibly stack DVDs on 1/2" centers and be able to pick them up. So in 6 feet you could stack maybe 140 or so DVDs on a row. 11-bits is probaby the best you will get for PWM accuracy and I’m not certain the SSC-32 resolves to that. Anyway, 72" / 2048 is about 0.035" per bit. So we are pretty much right at the limits of a normal servo signals resolution for this application, and we are assuming the SSC-32 can even do 11 bits just as much as we are assuming we can come up with a servo to resolve to that accuracy. Your (zoomkat) experimentation suggests this isn’t the case on one end or the other. If nobody knows the SSC-32 answer off hand I can wire mine up to a scope at home and see what its resolution really is.
As for wheels vs. slides… most of the X-Y tables I’ve seen use slide rails so I’m not thinking there will be a problem there in the horizontal but I am still betting you will need to do something specific to the vertical system to keep the horizontral track level. I think if you can solve the horizontal resolution issue then the vertical will be trivial to manage.
So here is the trick… “assuming” the SSC-32 resolves to 11 bits you need a way to measure linear position to at least 11 bits resolution so that a servo controller can drive a motor to the commanded position. Going back to the 10-turn pot idea (for lack of anything simpler so far) you could gear the pot so 1 turn = 7.2", roughly a 2.3" diameter wheel or gear on a rack. Servocity has both gear rack segments and gears/hubs that would allow this fairly easily.
You might want to check the thread below where some aspects of servo resolution are discussed. From what I’ve found, the limiting factor is the servo itself and not the output of the servo controller. Strangely, I seem to be the only person to actually test the servo resolution capability instead of just depending on the published servo controller output specs.
i was thinking the same thing atleast for the vertical move but thios could work well for the horz move too cheap and simple ,but instead of reading a bump u could use a dent to release the switch in front of the DVD,s and then just use a lookup table and count the button releases ,sweetttt
PS a bar code reader on this would be awsome
Yeah, sure, if we have the lattitude to just throw out the request that the solution uses the two unused outputs on RIOS + SSC-32 there’s a whole bunch of ideas… but…