From: "Brandon Van Every" <vanevery@blarg.net>
Subject: HOME-BREW: cheap pseudo-motion tracking?
Date: 4 Feb 1997 09:25:55 GMT
Message-ID: <01bc127a$90949d20$299072ce@hammurabi.blarg.net>
Organization: Blarg! Online Services - 206/441-9109


I'm looking to do an "artsy fartsy" project involving some number
crunching visualizations and some form of motion capture.  The goals
are aesthetic rather than scientific, so for now the motion capture
does not have to be terribly accurate.

I've started to read the FAQ about commercial motion trackers, but the
first thing that grabbed my eye was the typically $1000 price tag.
While not out of the bounds of possibility, the technology is
expensive enough to make me investigate alternatives.  Especially when
I consider that I'm a relatively "well-endowed" artist, i.e. not a
professional artist at all but a programmer, and I probably want to do
the project in collaboration with the NorthWest Cyberartists (in
Seattle.)  In non-profit organizations like the Cyberartists, $1000
price tags per person do not "scale" very well.

To that end, I've also started up the learning curve on
Microcontrollers, and things that one can attach to Microcontrollers.
For my project, it would be highly desireable to capture a lot of
motion from a lot of cheap components.  I can think of no upper bound
on what "a lot" would mean.  If I could afford 1000 components that do
some form of motion detection cheaply, I'm sure I could find a way to
number crunch them into an interesting visualization.

One device I'd like to have is a "virtual paintbrush."  Using a
commercial solution, this is just a tracker on the end of a stick.
But the positional data is not nearly as important to me as the
"gestural" data.  If I'm painting on a canvas, I do care about where
I'm putting down the stroke.  But I also care about the twist of the
bristles, the pressure of the bristles upon the surface, the drag of
the paint on the surface, i.e. all the fine nuances that make up the
"gesture" of the brushstroke.  In traditional painting, this is mainly
a matter of physical resistance between brush and canvas.  In 3d space
I suppose you could use air resistance - I imagine a sort of "wire
whisk" that bends according to how fast you move it through the air.
You could put sensors of some kind (I don't know what) on the ends of
the wires, and they could detect air resistance or the bend in the
metal or something.

This bring up several questions on feasibility:

1) What kind of sensors do I need, to make something like a "wire whisk?" 
I don't know beans about electronics.  Can someone give me a clue?

2) What kind of microcontrollers?  (Yes, I'm really clueless.  :-)  I've
read some specs on different ones, the specs don't mean a lot to me right
now, but there are FAQ's on them and I can look up the technical jargon.

2) I was hoping to just blow off the XYZ positional motion tracking. 
Imagine a big wall-size projection screen in front of the user.  The screen
would have a cursor, and the whisk could just apply a delta to the cursor's
current position, without any absolute spatial reference.  Much like a
mouse rolling on a desktop.  But from a user interface standpoint - is this
going to prove to be a royal pain in the ass?  Or is the user's hand-eye
coordination likely to be enough to tackle the problem?

3) Several companies make microcontroller gizmos for use by artists.
However, they are typically MIDI-driven, and they are typically
oriented towards on/off switch closures.  What I'm describing seems
much more heavily an "analog" application.  I'm doubting that MIDI can
handle the input bandwidth.  Am I correct?

4) Am I really, really kidding myself about the level of expense and
expertise it would take to build such a device?  Or is this a good
"homebrew" project?


Any help appreciated,
-- 
Brandon J. Van Every       |  Free3d: old code never dies!  :-)
                           |  Starter code for GNU Copyleft projects.
DEC Graphics & Multimedia  |
Windows NT Alpha   OpenGL  |  vanevery@blarg.net   www.blarg.net/~vanevery