>From time to time in response to one of my rants about
controlling time in Java 3D, people have recommended that I look
at the clocks in the Java Media Framework (JMF) API.

Today, I finally got around to getting that running, and have no
progress to report.  If any of you have waded through any of my
earlier rants and are still wading through this one, my
complaints center on 1.) the granularity of the clocks that Java
3D uses to schedule tasks and 2.) the OS-dependency in the VM's
implementations of those clocks.  I usually go on to whine that
the granularity and the OS-dependence make it very difficult to
design and deliver hi-fi simulations of physical processes built
on Java 3D.

Because my principal interest is delivering Java 3D models over
the web, I've restricted myself to looking only at the
platform-independent (all Java?) version of the JMF.  The JMF
API provides a set of interfaces for dealing with time and
clocks, but the only implemented TimeBase is SystemTimeBase, the
same clock that underlies such Java constructs as
System.currentTimeMillis(), Object.wait(), and the Java 3D
Behavior scheduler.  On WinNT and Win2k, its time grain is a
little over 15 ms, on Solaris 10 ms, and on Win9x, some odd
pattern that begins at 50 or so milliseconds.  This is far too
ragged to manage if we're trying to construct frames at say 50
per second (20 ms) down to 10 per second (100 ms).

My conclusion so far is that the JMF, as-delivered, offers no
help.

Question now is:  Given that there must be high precision
hardware clocks on the sound and video cards installed on Java
3D-grade computers, is there another TimeBase accessible to JMF?

TIA,

Fred Klingener
Brock Engineering

===========================================================================
To unsubscribe, send email to [EMAIL PROTECTED] and include in the body
of the message "signoff JAVA3D-INTEREST".  For general help, send email to
[EMAIL PROTECTED] and include in the body of the message "help".

Reply via email to