avfilter/vf_idet: Use frame_requested instead of prev

This is more robust if the delay is not constant

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

avformat/cdxl: Fix integer overflow of image_size

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

x86/swr: disable pack_8ch functions on msvc/icl x86_32

Until a proper fix is committed.

Signed-off-by: James Almer <jamrial@gmail.com>

x86/swr: add missing alignment check to pack_6ch functions

Reviewed-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: James Almer <jamrial@gmail.com>

avcodec/loco: check the init_get_bits8() return code

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

avcodec/loco: use enum for mode variable

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

doc/filters: document bit-depth independent cropdetect limit

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

avcodec/loco: Support LOCO reference encoder with odd width

Fixes Ticket1221

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

x86/swr: add SSE2/AVX pack_8ch functions

Reviewed-by: Michael Niedermayer <michaelni@gmx.at>
Reviewed-by: Ronald S. Bultje <rsbultje@gmail.com>
Signed-off-by: James Almer <jamrial@gmail.com>

avfilter/vf_cropdetect: adjust Default threshold based on bit depth

Also allow specifying thresholds that are bit depth independant

Previous version reviewed by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avfilter/vf_yadif: add >8bit planar rgb formats

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

ffprobe: Support extracting the number of reference frames

Fixes Ticket3141

Reviewed-by: Stefano Sabatini <stefasab@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avfilter/vf_uspp: clear AVPacket to not leave uninitialized memory

Fixes CID1260707

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

fate: add a fate-filter-framepack rule

avfilter/framepack: use FF_CEIL_RSHIFT()

avfilter/vf_cropdetect: extend limit to cover 16bit pixel formats

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

avfilter/vf_cropdetect: fix ; typos

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

avfilter/vf_cropdetect: support 9-16bit planar formats

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

avfilter/vf_cropdetect: Unroll 3 & 4 bytes per sample loop

19484 -> 10266 decicycles

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

avfilter/vf_cropdetect: Unroll 1byte per sample loop

12807 -> 6336 decicycles

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

avfilter/vf_cropdetect: add RGB & RGBA support

These where already supported in the code but not listed in the supported
pixel formats.

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

Merge commit '8d54bacb789c7d37ca3cf48d9ac13083ad0c1ba7'

* commit '8d54bacb789c7d37ca3cf48d9ac13083ad0c1ba7':
  dashenc: Remove some stray double spaces

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

Merge commit 'b91a5757fcbf723da99b05b298a6f820271dbc2b'

* commit 'b91a5757fcbf723da99b05b298a6f820271dbc2b':
  dashenc: Fix writing of timelines that don't start at t=0

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

Merge commit '50036c30df83b609bc5a95276f1287f8b9b8bdd6'

* commit '50036c30df83b609bc5a95276f1287f8b9b8bdd6':
  fate: Use bitexact conversions in the dpxparser test

Conflicts:
tests/fate/image.mak

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

Fix decoding for little endian Aja Kona 10-bit RGB.

The little endian fourcc is r10k, big endian uses R10k.
Reported and analyzed by Jason Freets, jasonslife hotmail

dashenc: Remove some stray double spaces

Signed-off-by: Martin Storsjö <martin@martin.st>

dashenc: Fix writing of timelines that don't start at t=0

When writing an explicit time, reset the cur_time variable to this
value as well. This avoids writing excessive time attributes for each
segment in the timeline, as long as the segments are continuous.

Signed-off-by: Martin Storsjö <martin@martin.st>

fate: Use bitexact conversions in the dpxparser test

This fixes fate on e.g. ppc.

Signed-off-by: Martin Storsjö <martin@martin.st>

libavformat/cdxl: fix duration in case of overflow

Fixes integer overflow
Fixes CID1260706

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

avutil/rational: Check that av_reduce() returns values within the requested max

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

avcodec/ituh263enc: Check den==0 in ff_h263_aspect_to_info()

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

avcodec/ffwavesynth: Use av_freep() to avoid leaving stale pointers in memory

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

avformat/mux: 2 subtitle packets could have the same DTS

Fixes Ticket3514

See: ETSI EN 300 743 V1.3.1 (2006-11)
"In summary, all of the segments of a single display set shall be carried in one (or more) PES packets that have the same
 PTS value."
with PTS = DTS and remuxing of such a stream it is to be expected that sometimes
multiple packets would have the same DTS

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

libavformat/segment.c: Add strftime expansion for segment filename templates

Allows expansion of the filename template with strftime() with the option
-strftime 1 (disabled by default). This allows segments to be named by time of
creation, adding some flexibility.

Fixes Ticket 4104 (add strftime to segment muxer)

Signed-off-by: Pedro E. M. Brito <pedroembrito@gmail.com>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

Fix R10k blue channel output.

Makes R10k -> gbrp10 ffv1 -> R10k roundtrip bit-identical.

doc/ffmpeg.texi: document the new -sdp_file option

Signed-off-by: Simon Thelen <ffmpeg-dev@c-14.de>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avdevice/dshow: Remove unneeded NULL checks

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

avdevice/dshow: Use av_freep() to avoid leaving stale pointers in memory

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

avdevice/dv1394: Use av_freep() to avoid leaving stale pointers in memory

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

avfilter/vf_cropdetect: add max_outliers parameter

Fixes Ticket3030

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

ffmpeg: Use av_freep(), avoid leaving stale pointers in memory

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

ffmpeg: add sdp_file option

Allow printing of sdp information to a file specified by -sdp_file
This allows users to print sdp information when at least one of the
outputs isn't an rtp stream.

Signed-off-by: Simon Thelen <ffmpeg-dev@c-14.de>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avcodec/vdpau: fix assertion failure and < vs > error

Fixes Ticket4211

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

Merge remote-tracking branch 'rbultje/vp9-32bit-lpf'

* rbultje/vp9-32bit-lpf:
  vp9/x86: add myself to copyright holders for loopfilter assembly.
  vp9/x86: make filter_16_h work on 32-bit.
  vp9/x86: make filter_48/84/88_h work on 32-bit.
  vp9/x86: make filter_44_h work on 32-bit.
  vp9/x86: make filter_16_v work on 32-bit.
  vp9/x86: make filter_48/84_v work on 32-bit.
  vp9/x86: make filter_88_v work on 32-bit.
  vp9/x86: make filter_44_v work on 32-bit.
  vp8/x86: save one register in SIGN_ADD/SUB.
  vp9/x86: store unpacked intermediates for filter6/14 on stack.
  vp8/x86: move variable assigned inside macro branch.
  vp9/x86: simplify ABSSUM_CMP by inverting the comparison meaning.
  vp8/x86: remove unused register from ABSSUB_CMP macro.
  vp9/x86: slightly simplify 44/48/84/88 h stores.
  vp9/x86: make cglobal statement more conservative in register allocation.
  vp9/x86: save one register in loopfilter surface coverage.

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

r210enc.c: Simplify and never store more than 10 bits.

The r10k and avrp decoders would previously store 12 bit precision
for the blue channel, which is inconsistent and probably not a
desirable behaviour.
Now the 2 unused extra bits are set to 0.
This is possibly not ideal either as RGBA1010102 format has the same
layout but stores alpha in these bits, thus explicitly setting them
to 1 might be preferable.

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

vp9/x86: add myself to copyright holders for loopfilter assembly.

vp9/x86: make filter_16_h work on 32-bit.

vp9/x86: make filter_48/84/88_h work on 32-bit.

vp9/x86: make filter_44_h work on 32-bit.

vp9/x86: make filter_16_v work on 32-bit.

vp9/x86: make filter_48/84_v work on 32-bit.

vp9/x86: make filter_88_v work on 32-bit.

vp9/x86: make filter_44_v work on 32-bit.

vp8/x86: save one register in SIGN_ADD/SUB.

vp9/x86: store unpacked intermediates for filter6/14 on stack.

filter16 goes from 508 to 482 (h) or 346 to 314 (v) cycles; filter88
goes from 240 to 238 (h) or 174 to 165 (v) cycles, measured on TOS.

vp8/x86: move variable assigned inside macro branch.

The value is not used outside the branch.

vp9/x86: simplify ABSSUM_CMP by inverting the comparison meaning.

vp8/x86: remove unused register from ABSSUB_CMP macro.

vp9/x86: slightly simplify 44/48/84/88 h stores.

vp9/x86: make cglobal statement more conservative in register allocation.

vp9/x86: save one register in loopfilter surface coverage.

avutil/atomic: reuse ret to avoid dereferencing twice the same value.

avfilter/vf_cropdetect: Factorize duplicated code using a macro

This simplifies subsequent changes

Reviewed-by: Benoit Fouet <benoit.fouet@free.fr>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

avfilter/vf_blend: Fix AVClass

Fixes infinite loop

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

lavfi: add tblend filter

avutil/hmac: use av_freep() to avoid leaving stale pointers in memory

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

avutil/audio_fifo: use av_freep() to avoid leaving stale pointers in memory

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

avutil/dict: Use av_freep() to avoid leaving stale pointers in memory

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

Merge commit '60d4c6ff76467d4d8f55c1cc61ab6c618e8ea2f3'

* commit '60d4c6ff76467d4d8f55c1cc61ab6c618e8ea2f3':
  h264: restore a block mistakenly removed in e10fd08a

See: 0ecb3075c1c243cfb656d1cbe0b19cdb91bde659
Merged-by: Michael Niedermayer <michaelni@gmx.at>

Merge commit '737d35e33408263c04d7730f5487eed0d04938ba'

* commit '737d35e33408263c04d7730f5487eed0d04938ba':
  vdpau: add support for the H.264 High 4:4:4 Predictive profile

Conflicts:
libavcodec/vdpau_internal.h

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

Merge commit '1f9237f2ac46dfbed1bfa1f4f0f1314c2a1d62ec'

* commit '1f9237f2ac46dfbed1bfa1f4f0f1314c2a1d62ec':
  avconv_vdpau: allocate video surface of VDPAU-specified size

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

Merge commit 'ebd5320afd42d4315851f3e0ca7f5d4a6300eb68'

* commit 'ebd5320afd42d4315851f3e0ca7f5d4a6300eb68':
  vdpau: add support for 4:2:2 and 4:4:4 chroma sampling

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

Merge commit 'c220a60f92dde9c7c118fc4deddff5c1f617cda9'

* commit 'c220a60f92dde9c7c118fc4deddff5c1f617cda9':
  vdpau: add helper for surface chroma type and size

Conflicts:
libavcodec/vdpau.c
libavcodec/version.h

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

Merge commit '6c99c92a42add7f6a462114d5a4a53c93c551058'

* commit '6c99c92a42add7f6a462114d5a4a53c93c551058':
  libavcodec: add AV_HWACCEL_ALLOW_HIGH_DEPTH flag

Conflicts:
libavcodec/version.h

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

Merge commit '57b6704ecd0f56d6a3092e448687cfd837bb0ac1'

* commit '57b6704ecd0f56d6a3092e448687cfd837bb0ac1':
  avcodec: add AVCodecContext.sw_pix_fmt

Conflicts:
doc/APIchanges
libavcodec/avcodec.h
libavcodec/version.h

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

Merge commit '4cfbeef31d4e6096c0596359d212f5d99a7ba4b5'

* commit '4cfbeef31d4e6096c0596359d212f5d99a7ba4b5':
  h264: factor hwaccel pixel formats list

Conflicts:
libavcodec/h264_slice.c

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

h264: restore a block mistakenly removed in e10fd08a

CC: libav-stable@libav.org
Bug-ID: 781

avfilter/vf_cropdetect: add yuv440p and yuv410p support

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

avfilter/vf_fspp: Add GBRP and gray8 support

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

ffserver: check for EOF|error at loop condition

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: reflow close_connection()

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: minor formatting improvement to http_server()

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: reindent start_multicast()

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: reflow start_multicast()

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: reindent start_children()

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: reflow start_children()

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

ffserver: break lines at 80 in func prototypes

Signed-off-by: Reynaldo H. Verdejo Pinochet <reynaldo@osg.samsung.com>

x86/vf_fspp: move pxor in store slice functions out of the loop

m7 is not overwritten, so we only need to clear it once.
Found by Christophe Gisquet.

Signed-off-by: James Almer <jamrial@gmail.com>

x86/vf_fspp: port inline asm to yasm

Reviewed-by: Michael Niedermayer <michaelni@gmx.at>
Signed-off-by: James Almer <jamrial@gmail.com>

avfilter/vf_fspp: clarify comment

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

swscale: Pass through chroma positions in sws_getCachedContext

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

avfilter/vf_boxblur: generate supported pixfmt list instead of hardcoding

This adds support for several more >8bit planar formats

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

avformat/rtsp: Use av_freep() to avoid leaving stale pointers in memory

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

avformat/cache: remove ftruncate usage, its not always available

In case of errors the cache file will be slightly larger than needed,
this should have no practical relevance though

Should fix build on VS201*

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

lavfi/blend: add difference128 mode

avformat/cache: Extend cache entries if possible instead of creating new ones

This reduces the number of cache entries and should significantly
reduce memory requirements

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

avformat/cache: avoid lseek() on reading from the cache if possible

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

avformat/cache: keep cache_pos updated

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

avformat/cache: cleanup cache file on cache write failure

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

Changelog: Mention non continuous cache protocol

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