byte position information, as stored in the pkt.pos, through the
filterchain.
Note that the pos field is added *non* at the end of the
AVFilterPicRef struct, thus breaking ABI compatibility, which is
allowed as the API is still considered non-stable.
Originally committed as revision 22506 to svn://svn.ffmpeg.org/ffmpeg/trunk
avcodec_get_chroma_sub_sample(), directly access
av_pix_fmt_descriptors instead.
Remove some of the dependancies of lavfi on lavc.
Originally committed as revision 21575 to svn://svn.ffmpeg.org/ffmpeg/trunk
auto-inserted scale filter, rather than the filter name, as this
provides more information.
Originally committed as revision 21123 to svn://svn.ffmpeg.org/ffmpeg/trunk
Avoid the need to implement slice direction detection code, thus
reducing code duplication.
See the thread:
"[FFmpeg-devel] [PATCH] Add a slice_dir parameter to avfilter_start_frame()".
Originally committed as revision 20734 to svn://svn.ffmpeg.org/ffmpeg/trunk
Create a new static array containing pointers to the AVFilter
definitions, so that the non-constant next filter in the AVFilter
struct is not anymore required and the AVFilter definitions may be
stored in shareable memory.
Also change the signature for avfilter_register(), make it return an
int since it may fail if there is not enough space in the static array
for the registered filters.
Originally committed as revision 20605 to svn://svn.ffmpeg.org/ffmpeg/trunk
avfilter_default_get_video_buffer() if the get_video_buffer() callback
is not defined in a filter.
Libavfilter filters author have to explicitely define the
get_video_buffer() callback if they want the buffer to be requested to
the filter following in the filterchain.
See the thread:
"[FFmpeg-devel] [PATCH] Make avfilter_get_video_buffer() use
default_get_video_buffer if callback not defined".
Originally committed as revision 20444 to svn://svn.ffmpeg.org/ffmpeg/trunk
for registering and finding filters, rather than use the struct
AVFilterList, which is removed.
Simplify the filter registration management code.
Originally committed as revision 20387 to svn://svn.ffmpeg.org/ffmpeg/trunk
min = FFMIN(min, avfilter_poll_frame(link->src->inputs[i]))
which, since FFMIN is a macro, was calling itself
twice for every input, causing an exponential cost in time.
Originally committed as revision 20295 to svn://svn.ffmpeg.org/ffmpeg/trunk
When called on a link with a filter whose destination pad has not a
get_video_buffer callback defined, it will call
avfilter_get_video_buffer() on the first output link of the
destination filer, rather than use avfilter_default_get_buffer(), so
the video buffer can be allocated forward in the filterchain.
Also add the w and h parameters to avfilter_get_video_buffer(), as the
minimum width and height requested by each filter in the filterchain
may change, this allows for example a memcpy-less pad filter.
This change breaks API / ABI backward compatibility.
See the thread:
"[PATCH] Implement recusive avfilter_get_video_buffer()".
Originally committed as revision 20272 to svn://svn.ffmpeg.org/ffmpeg/trunk
corresponding count is zero, rather than allocate a 16 bytes sized
block for them. Improve safety.
Originally committed as revision 16565 to svn://svn.ffmpeg.org/ffmpeg/trunk
avfilter_draw_slice() when the draw_slice callback is not defined in
the input pad.
Originally committed as revision 16554 to svn://svn.ffmpeg.org/ffmpeg/trunk
Suggested by Bobby Bingham.
Commited in SoC by Vitor Sessak on 2008-02-09 16:56:55
Originally committed as revision 12042 to svn://svn.ffmpeg.org/ffmpeg/trunk
