hmm... this might not answer the question, but it might help the thought
When thinking about how to make a certain kind of image, I always find
it useful to think about if something is (for want of better terms)
an *artifact* or an *effect*.
For example, I spent a signficant portion of time (3 + 10 months)
working on Mike McKenna & Bob Sabiston's animation "Grinning Evil Death",
and I did all the motion blur *effects* and caused the depth of field
*artifacts*, using my recursive camera compositor.
Here's what I mean: Dave Sturman, for his qualifying exam in the Ph.D.
program here at the Media Lab, did some interesting depth of field
experiments. When I built my recursive camera code (a "useful fiction"
for building a digital multi-plane camera), I used Dave's code to let
me specify normal 35mm camera parameters to my software so I could use
my still photographer's intuition to relatively position plates and
do rack focus. When it came time to do motion blur, though, I designed
a bunch of filters that I thought "looked right" and tweaked various
things to generate image sequences that I thought worked. The depth
of field effects were artifacts, the motion blur was an effect.
On the other hand, I recently designed and executed two shots for a really
nice short film called "Dizzy Horse", by Gary Cohen. In that piece, there's a
ring flying at, through, and away from the camera, where the camera is doing
a follow focus on the ring as it moves through a forest. In that piece,
the depth of field is an effect, but the motion blur is an artifact.
What I mean is this: in my animation system, the ring is moving continuously,
over the course of 7 seconds. I set up a camera to shoot the ring for
two shots, each 3 seconds, where the camera was sampling 24 times per second,
and its shutter was open 1/48 of a second. This spit out the appropriate
RIB file which prman rendered (with the appropriate motion blur). The
motion blur here was an artifact. On the other hand, the depth of field
(i.e. the follow focus) was done for me by Greg Zinn at DuArt using some images
I had shot on 35mm (scanned to PhotoCD) by blurring them using Alias
Eclipse on an SGI using start and end points I gave him. I then took
those images off tape and composited them with the motion blurred rendered
In this case, the depth of field was an effect, but the motion blur was
This is all a long-winded way of saying that you might want to think
about how to do your SW streaky stars; and think about if you have
to do it as an effect, or if you can figure out some process whereby
it's an artifact (which is always easier, I think). Of course, one
man's artifact is another's effect, but getting the right perspective on that
can be really useful...
For example, I would think (off the top of
my head, probably wrong) is that the streaky stars were originally done
as a slit scan shot with an old analog optical printer. One easy way
to do them digitally is by compositing two shots of the same scene; one
rendered strobe and one rendered motion blurred. That's a popular
party-shooting strategy (check out the rave pics off my home page),
where you shoot with a strobe but leave the shutter open for a second
In all cases, I doubt that this is a custom shader issue, I would think it
would be a matter of rendering with a variety of shutter speeds and
--> MIT Media Lab, Computer Graphics & Animation Group
--> 20 Ames St. E15-023G, Cambridge, MA 02139