Allow to show the supported options in the ffmpeg -h full output.

Fixes CID703674
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes CID703717
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fixes CID703849
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

We would crash a moment later anyway if this fails.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

The native filter was ported from the MPlayer project when the mp
wrapper didn't even exist, and when mp=hqdn3d was added, it was already
obsolete.

Fix possible crash occurring when libc does not support NULL printing.

All the rectangle features have been ported to the native drawbox filter,
which has an equivalent syntax.

Based on a libmpcodecs/vf_rectangle.c feature.

The filter has been ported to a native libavfilter filter.

The filter is a port of libmpcodecs/vf_field.c, since there is no common
code I relicensed it as LGPL, while keeping the original author
copyright.

This is more consistent with the way we handle C #includes and
it simplifies the build system.

This is necessary to allow refactoring some x86util macros with cpuflags.

That interface is deprecated, and apparently useless.

Same functionality is provided by libswscale.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

This fixes output with filtergraphs such as 'split[a][b];
[a]pad=iw*2[src]; [b]smartblur=1:1:1[filt]; [src][filt]overlay=w'

This is simpler and also should allow the code to function when the
channel number changes.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

The filter is useless, swscale should be used instead.

Signed-off-by: Paul B Mahol <onemda@gmail.com>

It can be useful for debugging.

Based on a patch by Stefano Sabatini <stefano.sabatini-lala@poste.it>

Update path handling against the latest Frei0r specification changes.

Signed-off-by: rogerdpack <rogerpack2005@gmail.com>
Signed-off-by: Stefano Sabatini <stefasab@gmail.com>

Fri0r project moved from piksel.org to frei0r.dyne.org.

If the first argument can be read as a video size, set that output size
in the scale. This allows to specify in a filtergraph scale=qcif or
scale=320x240.

This is not completely safe, for example in case of a typo in the video
size string the first argument will be read as the input width
expression, giving rise to a confusing comment.

Avoid an assert, since now the function supports error code return.

Also update documentation accordingly.

Fixes ticket 1782.

Fixes CID703716
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>

Signed-off-by: Mans Rullgard <mans@mansr.com>

Signed-off-by: Mans Rullgard <mans@mansr.com>

If the stats are printed when request_frame on the input
returns EOF, then they are never printed if the filter is
not flushed, in particular when using the -t ffmpeg option.

Fix trac ticket #1727.

By default, fps selects frames near PTS 0, 0.5, 1.5, 2.5, etc.
With round=down, it selects the frames near PTS 0, 1, 2, 3, etc.

Fixes CID700635, CID700636 and CID732274.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Allow to specify a filename where to put the filtergraph description.

This is useful to override limitations or glitches of particular shell
environments, and allows a level of indirection for specifying
filtergraphs.

Avoid hanging file.

This fixes ff* -h infinite looping

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>