Motion Control Camera

I'm considering getting in to microcontroller programming and have a project I want to work toward.
Any suggestions on how to build a decent motion control camera rig? I
want to play around a bit with greenscreen model shots, but mostly I want something to perform smooth repeatable camera moves. I'd start out with two axis and move up from there.
My concerns are the following.
-Learning (as opposed to programming the move each time as you would a CNC lathe) -how to get home position or position feedback -Memory (Do I store only angle information or angle AND acceleration?) -physical layout -Microcontroller power and ease-of-use
Ease of use dictates that the rig be operated by joystics operating the axis (pan/tilt/up/down/forward travel/reverse travel). I could probably figure out how to use a Basic Stamp reading pots and operating some geared-down stepper motors to get this functionality, but how do you store the information? How fast do you need to poll the joystics and set the steppers for smooth movement? How much memory is required for, say, 2 minutes of data?
I think the layout should follow the camera usage as much as possible. So, no boom and no arm. The camera should probably be on a track with counterweighted movements up and down a central shaft, or series of shafts.
Finally, what's a good platform to use for someone who knows a very little bit about a lot?
Basic Stamps are easy, but I'm concerned about cost and the power of the Basic programming language. I REALLY like Basic Stamp's all-inclusive starter kits, though.
Ooopics look better, but cost is a problem (I'd rather fry a $5 pic than a $70 daughter board) and no nifty starter kits.
Proton Basic looks promising, but is expensive and no all-inclusive starter kids with books to hand-hold you through the learning process.
And that's just the PIC side. There's Atmel and some Rabbit thing too. Along with Lego's Mindstorm.
Thanks for your help!
P.S. I HAVE done a motion control scene already! I put my camera on the rotating platform of a stretch wrapper. The movement slows and stops in roughly the same place so I videotaped myself three times at three different locations and composited them together. The result is the camera panning over one of me, then stopping on two of me sitting at a table. That was fun!
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
UGH..
1) MoCo Two Modes. Record/edit/playback Moves are taught, editable on a per axis basis Moves can be played back slowly Moves can be smoothed
Keyframe Keyframes are selected 'tweens are calculated lead in/lead out slope is calculated
2) Microcontroller There is a lot to it. maybe you can use the basic stamp as an intelligent LCD controller.
3) Repeatability If you truly want to do mattes, you need frame accurate motion control. You need to play the scene back exactly every time in relationship to the beginning of a captured frame.
4) Memory Lots...
If I were building it, I would use a Gumstix doing user IO on a nice LCD, with an IsoPodX doing motor control and timing.

Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload
Wow! Since I'm looking to learn by doing I think the Gumstix and IsoPodX are beyond my reach (financially and knowledge) just yet.
I hadn't thought about editing. That's a good point. Ideally, it would record the moves to a file on a USB or flash drive which could then be edited on a PC. (or, for real pie-in-the-sky it could output coordinates usable by 3D animation programs for use in virtual sets. But, I realize matching the lens makes this problematic).
Repeatable almost frame-accurate motion is a must. I realize with a home video camera I'm never going to get it exact (although, model shots could be done frame accurately with my EOS digital and long exposure times with some sort of shutter attachement). Ultimately I'm looking to be able to do twin effects with a moving/panning camera.
Add pictures here
<% if( /^image/.test(type) ){ %>
<% } %>
<%-name%>
Add image file
Upload

Polytechforum.com is a website by engineers for engineers. It is not affiliated with any of manufacturers or vendors discussed here. All logos and trade names are the property of their respective owners.