[MPlayer-users] Some questions about "exact rendering" and frame rate
Dean S. Messing
deanm at sharplabs.com
Fri Dec 20 22:49:02 CET 2002
I have a few technical questions which will reveal my ignorance of
X, rendering, video card operation, and mplayer. Oh well.
We are doing careful psycho-physical motion experiments in my laboratory.
We put up various moving patterns on either CRT or LCD and then viewers
evaluate them in various ways. The patterns are generated in Matlab
(or, in my case, in Scilab under Linux), converted to raw data .avi files
and then played.
In my case I'm trying to use mplayer for the playing (and later I need
to figure out how to use mencoder for encoding the raw .avi files from
.bmp files)
Our problem is that we are so far unable to render our test sequences
without either "tearing" or skipped/doubled frames. My meager understanding
is that in linux your can either "blit" the pixels into the framebuffer
(which can lead to "tearing") or you can "overlay" them which, as I understand
it, means you are writing into the framebuffer in synchrony with the
actual display framerate. This will prevent tearing but will still
lead to doubled/skipped frames if the .avi frame rate is not
exactly the display frame rate.
First: is my understanding more or less correct,
or am I hopelessly lost?
Second: If I do understand things, how do I go about getting "exact
rendering", namely no tearing, no frame doubling/skipping, and
having the frames actually displayed at the desired rate so
that our objects and patterns move at the desired
spatio-temporal velocities?
Third: In particular, is it sufficent to write, say, a 60Hz
(progressive) "mode" within the XF86Config file and then just
let `mplayer -noskip' play the .avi at the default 60fps embedded in the
.avi file?
As always, thanks for your help. Several in my group are waiting
to see if Linux can solve a longstanding problem that they cannot
in M$ Windoze.
Dean S. Messing
Display Algorithms & Visual Optimization Lab
Information Systems Technologies Dept.
Sharp Laboratories of America
More information about the MPlayer-users
mailing list