Monthly Archive for July, 2008

Home Brew Rigs

This is one of the first tests on our home made motion control system.

Contrast this with another test we made on a professional motion control unit last year:

The home made rig is as of yet pretty unstable, but the project is in progress (sponsored no less by LEGO).  Following a suggestion from Saschka Unseld over at Studio Soi, we discovered that by stabilizing just one point in the dolly shot and cropping in, you can fake the illusion of a track/pan.

Stability test 2

Ok, so this is our second test – using a more manageable gear factor of 576 (24 ^ 2), and with some increased speed on the motor.  Unfortunately, we’ve got some issues here:

  1. Our software is not resetting the rig back to the same position
  2. The motion is more inconsistent between passes (than the last test)
  3. The camera looks like it’s shifted a bit (could be focal length, could be focus) in the first frame – we’ll need to lock it down a bit more for the next tests.

Below you see two passes that have edge detection applied screened together.  The jitter you see (where the red and green jump around each other) represents the motion instability.


First stability test

Here are two separate passes we shot with our stability rig.  Our gear factor is an incredible 1:13824 and the motors take a couple of minutes to move .1 of a centimeter.  While that’s great, it’s a bit unwieldy, unless we get some higher power motors.  So we’ll be working with finding the sweet spot, where we can hit the same precision, but get our rig to move a bit quicker (click on the images for the movies).

This move is driven by processing and Maya.  We’re mostly just using Maya to set some simple moves and use it’s curve editors, until we get our own flavor of 3d interface up and running in processing.  The moves are exported as .move files, which then I parse through in processing and transfer to translation for the rig.

This test is to find out whether we can accurately repeat the same motion.  Below is the combination clip, where blue represents one layer, and red represents the other layer.  For the next few tests, I’ll shoot each pass with the same lighting conditions and do some difference composites (instead of this red/blue).

While the test is interesting, I’ve had to slightly hand transform (a few pixels) and then animate over the length of the shot one pass in order to match the other.  Our task will be to narrow down where this descrepancy is happening (either our rig, or the motors, or the software).

This test is only 25 frames, but we’ll get some longer length tests also up soon.

Servo Motors

Regarding the Prototype Test, while the movement can be stablized in post, we can’t yet recreate the same motion over multiple passes.  So that will be our first task – testing the useability of the Mindstorms servos.

Our software is based on processing and the NXT java library.  Processing is a great environment to sketch in, and we’ll be writing our interfaces through it.

Prototype Test

Here’s a test we made with our prototype system a couple of months ago.  The movement was stabilized in post, and that will be one of our challenges with the actual build – if we can create stable motion with the actual rig.

The lighting for this shot was a simple office style lamp and some ambient fill from the windows (which I then deflickered in Shake).  One of the nice things about shooting with a stills camera (rather than a moving image one) is that our exposure can be extremely long, and our lighting can be relatively low.

Legos have arrived!

These are some of the photos we took when the LEGO’s first arrived.

Warning: require_once(/home/sunit_parekh/ failed to open stream: Permission denied in /home/sunit_parekh/ on line 572

Fatal error: require_once(): Failed opening required '/home/sunit_parekh/' (include_path='.:/usr/local/lib/php:/usr/local/php5/lib/pear') in /home/sunit_parekh/ on line 572