For some reason, this did not appear in the mailman management interface to
allow me to approve it.
---------- Forwarded message ---------
From: <mlt-devel-ow...@lists.sourceforge.net>
Date: Sat, Apr 16, 2016 at 12:16 PM
Subject: Mlt-devel post from j...@kdenlive.org requires approval
To: <mlt-devel-ow...@lists.sourceforge.net>
As list administrator, your authorization is requested for the
following mailing list posting:
List: Mlt-devel@lists.sourceforge.net
From: j...@kdenlive.org
Subject: Problem with mlt_filter_get_length2
Reason: The message headers matched a filter rule
At your convenience, visit:
https://lists.sourceforge.net/lists/admindb/mlt-devel
to approve or deny the request.
---------- Forwarded message ----------
From: Jean-Baptiste Mardelle <j...@kdenlive.org>
To: <mlt-devel@lists.sourceforge.net>
Cc:
Date: Sat, 16 Apr 2016 21:15:46 +0200
Subject: Problem with mlt_filter_get_length2
Hi,
While working on filters with 2 pass analyse feature, I ran into a problem
with mlt_filter_get_length2.
Looking at the code, mlt_filter_get_length2 returns the length not from the
producer it is attached but from the parent of the producer it is attached
to:
mlt_filter.c, line 256
if ( producer )
{
producer = mlt_producer_cut_parent( producer ); <---- gets parent
of the
cut
in = mlt_producer_get_in( producer );
out = mlt_producer_get_out( producer );
}
But in Kdenlive the length of a clip in a playlist may not be the same at
the one of its parent.
For example we can have this xml:
<producer id="producer0" out="2000" in="0">
(...)
<playlist>
<entry producer="producer0" in="200" out="1000">
</playlist>
So if we apply an effect to the clip in playlist, the filter length is 800,
but mlt_filter_get_length2 will return 2000 (length of parent clip).
In mlt_filter_get_length2, removing the above line
"mlt_producer_cut_parent" fixes the issue and returns a correct length of
800. However, seeing how it is used in MLT, this fix is not really wanted,
but maybe the documentation should be updated...
Now my problem is that some filters, like loudness and vidstab perform
analysis on the frames, and save data when processing the last frame of the
filter.
Currently, last frame is checked like this:
if ( mlt_filter_get_position( filter, frame ) + 1 ==
mlt_filter_get_length2( filter, frame ) ) {}
This does not work in Kdenlive, and these filters never produce a result
because with my above example, last frame of the filter is 1000 and filter
length is 2000.
My proposal would be to introduce a new function, mlt_filter_get_out2 that
retturns the real last frame of the filter, like this:
mlt_position mlt_filter_get_out2( mlt_filter self, mlt_frame frame )
{
mlt_properties properties = MLT_SERVICE_PROPERTIES( &self->parent );
mlt_position out = mlt_properties_get_position( properties, "out" );
if ( out == 0 && frame )
{
// If always active, use the frame's producer
mlt_producer producer = mlt_frame_get_original_producer(
frame );
if ( producer )
{
out = mlt_producer_get_out( producer );
}
}
return out;
}
We can then compare this value to mlt_filter_get_position to know if the
filter reached last frame.
What do you think about it ? Do you have another idea / proposal on how to
solve this issue ?
Thanks
jb
------------------------------------------------------------------------------
Find and fix application performance issues faster with Applications Manager
Applications Manager provides deep performance insights into multiple tiers of
your business applications. It resolves application problems quickly and
reduces your MTTR. Get your free trial!
https://ad.doubleclick.net/ddm/clk/302982198;130105516;z
_______________________________________________
Mlt-devel mailing list
Mlt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mlt-devel
