Hello!

Thank you very much for the patch.

Sorry for not responding early, have had a few boring things that
needed doing ... :-)

I've done some experiments, and this is what I have found:

1: The code added by the patch does not execute, when I use openshot to render
my minimal test case.

Also, there are neither any white frames inserted, nor is the last
frame repeated.

I added some printouts to producer_get_image, and I can see that it seems to
be called once for every frame that is actually present in the source
clip. After
playback reaches the end of the source clip, produce_get_image is not
called at all.

2: If I use melt, as you suggested, giving an out parameter after the
end of the file,
the output is what would be expected - the last frame IS duplicated.

I'm not sure what to make of this.

I've attached an .mlt file generated by Openshot, which demonstrates
the problem.
It is a rather minimal example. It doesn't seem to make much difference what the
actual clips are, so I'm not including any clips. Just take some
random clip, and
change its name to "four_and_a_half_second_clip.MOV", and then render the
.mlt-file using melt, and I hope the problem can be reproduced. It is important
that the clip is ~4.5 seconds long, if the clip is longer, everything seems
to work.

The expected behaviour is that the clip plays for slightly more than 3 seconds,
then there is some "blank", and then at exactly 4 seconds (frame number 100),
the clip comes back in.

However, that is not what I get. Instead, the first clip only plays
until it runs out of
frames, then the 18-frame blank appears prematurely, which means that the second
showing of the clip is way early. (Which of course would make any
transitions/effects
be badly timed).

Something that may be related to all this is that I've done the a
modification to
consumer_avformat.c, to work-around a segfault that happened when
running my minimal
example. This patch is a wild shot in the dark from my part, so it may
be totally wrong. I've
included the patch as well to this mail.

I'm willing to spend some time troubleshooting this further. I'll post
to this mailing list again
if I make any progress!

cheers,
Anders Musikka


On Fri, Apr 29, 2011 at 9:56 AM, Dan Dennedy <[email protected]> wrote:
> On Fri, Apr 29, 2011 at 12:54 AM, Dan Dennedy <[email protected]> wrote:
>> Can you test this patch against mlt head that tries to fix that?
>
> Attached patch this time.
>

Attachment: mlt_segfault_fifo_patch.diff
Description: Binary data

Attachment: test_mlt_patch.mlt
Description: Binary data

------------------------------------------------------------------------------
WhatsUp Gold - Download Free Network Management Software
The most intuitive, comprehensive, and cost-effective network 
management toolset available today.  Delivers lowest initial 
acquisition cost and overall TCO of any competing solution.
http://p.sf.net/sfu/whatsupgold-sd
_______________________________________________
Mlt-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/mlt-devel

Reply via email to