El lun, 29-06-2009 a las 11:44 -0400, A. Craig West escribió:
> 2009/6/29 Aris Stathakis <aris@...>:
Hi Guys ,
i've been away from motion development, but i'm back already.
I took a look to this patch but it was not mature enough to be included
in svn yet , not sure what is the status now so let me take a look and
see how can it be integrated.
Did anyone create a Patch topic in Twiki about it ?
> > One way to test might be to output jpg frame files instead of creating a MOV - and see how many jpegs it writes per second with and without the patch.
> > I'm assuming your patch will increase the speed of writing frame captures?
> > --- On Mon, 6/29/09, Jeff Groves <jgroves@...> wrote:
> >> As far as the performance of the buffering, on my slow, old machine it
> >> appears that I'm getting about 20% better frame rate than without the
> >> patch -- I don't think I'm suffering from the placebo effect in my
> >> comparison, but just to be sure, how would I be able to measure the
> >> difference by looking at the properties of the .mov files that are
> >> created?
> It is worth checking your video player software to see if it can get
> the framerate from the file, but I don't know for sure if that would
> work, as I am at work at the moment. I have output_normal turned on as
> well as the ffmpeg, so I can use the timestamps of the jpegs to get
> the frame rate. Testing with this doesn't necessarily help tell how
> much of the improvement is in the jpeg writing, and how much is in the
> ffmpeg writing, though. When I am home, I can experiment with
> extracting framerate information from the .mov files. If all else
> fails, it should be possible to extract out a fixed time period to
> individual frames, and count them :-)
> As to the frame capture rate, the patch was originally written just
> for that, and was very effective, at least on the rather odd embedded
> platform I run on. I have the filesystem mounted in sync mode, so I
> suspect that I will see a larger difference than with a regular
> system, though. Because the ffmpeg code originally did all writes with
> totally non-buffered calls, I expect that the difference will be even
> larger than it was with the jpeg code.
ack ( at ) telefonica ( dot ) net
Key fingerprint = 3FD3 9C90 149E 7824 CECD 6BCF AC2C CA61 6EF1 B90D
"No basta saber, hay que aplicar lo que se sabe;
no basta querer hacerlas cosas, hay que hacerlas".
"Knowing is not enough; we must apply.
Willing is not enough; we must do"
Johann Wolfgang von Goethe