Commit Graph

61 Commits

Author SHA1 Message Date
Claudio Freire
01ecb7172b AAC encoder: Extensive improvements
This finalizes merging of the work in the patches in ticket #2686.

Improvements to twoloop and RC logic are extensive.

The non-exhaustive list of twoloop improvments includes:
 - Tweaks to distortion limits on the RD optimization phase of twoloop
 - Deeper search in twoloop
 - PNS information marking to let twoloop decide when to use it
   (turned out having the decision made separately wasn't working)
 - Tonal band detection and priorization
 - Better band energy conservation rules
 - Strict hole avoidance

For rate control:
 - Use psymodel's bit allocation to allow proper use of the bit
   reservoir. Don't work against the bit reservoir by moving lambda
   in the opposite direction when psymodel decides to allocate more/less
   bits to a frame.
 - Retry the encode if the effective rate lies outside a reasonable
   margin of psymodel's allocation or the selected ABR.
 - Log average lambda at the end. Useful info for everyone, but especially
   for tuning of the various encoder constants that relate to lambda
   feedback.

Psy:
 - Do not apply lowpass with a FIR filter, instead just let the coder
   zero bands above the cutoff. The FIR filter induces group delay,
   and while zeroing bands causes ripple, it's lost in the quantization
   noise.
 - Experimental VBR bit allocation code
 - Tweak automatic lowpass filter threshold to maximize audio bandwidth
   at all bitrates while still providing acceptable, stable quality.

I/S:
 - Phase decision fixes. Unrelated to #2686, but the bugs only surfaced
   when the merge was finalized. Measure I/S band energy accounting for
   phase, and prevent I/S and M/S from being applied both.

PNS:
 - Avoid marking short bands with PNS when they're part of a window
   group in which there's a large variation of energy from one window
   to the next. PNS can't preserve those and the effect is extremely
   noticeable.

M/S:
 - Implement BMLD protection similar to the specified in
   ISO-IEC/13818:7-2003, Appendix C Section 6.1. Since M/S decision
   doesn't conform to section 6.1, a different method had to be
   implemented, but should provide equivalent protection.
 - Move the decision logic closer to the method specified in
   ISO-IEC/13818:7-2003, Appendix C Section 6.1. Specifically,
   make sure M/S needs less bits than dual stereo.
 - Don't apply M/S in bands that are using I/S

Now, this of course needed adjustments in the compare targets and
fuzz factors of the AAC encoder's fate tests, but if wondering why
the targets go up (more distortion), consider the previous coder
was using too many bits on LF content (far more than required by
psy), and thus those signals will now be more distorted, not less.

The extra distortion isn't audible though, I carried extensive
ABX testing to make sure.

A very similar patch was also extensively tested by Kamendo2 in
the context of #2686.
2015-10-11 17:29:50 -03:00
Andreas Cadhalpun
107e54c5bf avutil: add ff_reverse as av_reverse replacement
The table is used in libavutil/eval.c.

Reviewed-by: Michael Niedermayer <michael@niedermayer.cc>
Signed-off-by: Andreas Cadhalpun <Andreas.Cadhalpun@googlemail.com>
2015-08-12 00:14:14 +02:00
ZhouXiaoyong
0ace686ae8 avcodec: optimize mathops for Loongson-3 v1
HAVE_LOONGSON is replaced by HAVE_LOONGSON3. Even Loongson-2E and 2F support
Loongson SIMD instructs but have low performance for decoding. We plan to focus
on optimizing Loongson-3A1000, 3B1500 and 3A1500, and modify the configure file
to support Loongson-2 series later by adding HAVE_LOONGSON2.

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-04-27 02:25:12 +02:00
zhaoxiu.zeng
0f83415569 avcodec/vc1_mc: move median4() to mathops.h
Needed for architecture specific optimizations

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2015-02-16 11:41:00 +01:00
Bernd Kuhls
bf6c84d7eb Fix compile error on bfin.
After the removal of all Blackfin architecture optimizations in
http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=b55d3bbeed375f7b74442c4dd274d116a3e3d2e1
some includes were left behind leading to a compile error:

CC      libavformat/adtsenc.o
In file included from ./libavcodec/get_bits.h:35,
                 from ./libavcodec/ac3_parser.h:27,
                 from libavformat/ac3dec.c:23:
./libavcodec/mathops.h:43:29: error: bfin/mathops.h: No such file or directory

This compile error was found by buildroot autobuild system:
http://autobuild.buildroot.net/results/ae0/ae056f267e907091d09d2a1546d6f1ae02fa23b9/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
2014-08-05 01:54:47 +02:00
Michael Niedermayer
672c610d84 Merge commit 'f9279ee74936b4990fc46287c27816ef828cc06c'
* commit 'f9279ee74936b4990fc46287c27816ef828cc06c':
  dsputil: Move ff_zigzag_direct and ff_crop_tab declarations to mathops.h

Conflicts:
	libavcodec/imgconvert.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-03-22 22:23:04 +01:00
Diego Biurrun
f9279ee749 dsputil: Move ff_zigzag_direct and ff_crop_tab declarations to mathops.h 2014-03-22 06:17:29 -07:00
Michael Niedermayer
8cc581fbf7 Merge commit '5397386effba2e53e4ff82852a86f6be4d59e9c1'
* commit '5397386effba2e53e4ff82852a86f6be4d59e9c1':
  mathops: move macro to the only place it is used

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2014-03-04 13:38:04 +01:00
Vittorio Giovara
5397386eff mathops: move macro to the only place it is used
This helps in disentangling lavf and lavc too.
2014-03-04 11:45:31 +01:00
Michael Niedermayer
ac2fc63ac6 Merge commit '9ae53c5860e17d087642d27b00dade071970a2e7'
* commit '9ae53c5860e17d087642d27b00dade071970a2e7':
  lavc: add ff_u8_to_s8() for safe type punning of uint8_t to int8_t

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2013-09-19 11:47:40 +02:00
Justin Ruggles
9ae53c5860 lavc: add ff_u8_to_s8() for safe type punning of uint8_t to int8_t 2013-09-18 14:10:05 -04:00
Michael Niedermayer
0df55e1ba8 Merge commit 'a22ae9f0c579793f411e2bd7a8db557091a3a4ae'
* commit 'a22ae9f0c579793f411e2bd7a8db557091a3a4ae':
  mpegts: Remove one 64-bit integer modulus operation per packet

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2013-08-06 16:23:32 +02:00
Ben Avison
a22ae9f0c5 mpegts: Remove one 64-bit integer modulus operation per packet
The common case of the pointer having increased by one packet (which results
in no change to the modulus) can be detected with a 64-bit subtraction,
which is far cheaper than a division on many platforms.

           Before          After
           Mean   StdDev   Mean   StdDev  Change
Divisions   248.3  8.8      51.5   7.4    +381.7%
Overall    2773.2 25.6     2372.5 43.1     +16.9%

Signed-off-by: Martin Storsjö <martin@martin.st>
2013-08-05 19:25:55 +03:00
Michael Niedermayer
d6c342fdc0 Merge commit 'd5c62122a7b26704bf867a1262df358623bf5edf'
* commit 'd5c62122a7b26704bf867a1262df358623bf5edf':
  Move av_reverse table to libavcodec

Conflicts:
	libavcodec/asvenc.c
	libavcodec/vble.c
	libavutil/common.h
	libavutil/mathematics.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-10-13 14:35:42 +02:00
Diego Biurrun
d5c62122a7 Move av_reverse table to libavcodec
It is only used in that library.
2012-10-12 20:39:18 +02:00
Michael Niedermayer
e335658370 Merge commit '9734b8ba56d05e970c353dfd5baafa43fdb08024'
* commit '9734b8ba56d05e970c353dfd5baafa43fdb08024':
  Move avutil tables only used in libavcodec to libavcodec.

Conflicts:
	libavcodec/mathtables.c
	libavutil/intmath.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-10-12 14:26:46 +02:00
Diego Biurrun
9734b8ba56 Move avutil tables only used in libavcodec to libavcodec. 2012-10-11 18:29:36 +02:00
Michael Niedermayer
9f088a1ed4 Merge remote-tracking branch 'qatar/master'
* qatar/master:
  mpegvideo: reduce excessive inlining of mpeg_motion()
  mpegvideo: convert mpegvideo_common.h to a .c file
  build: factor out mpegvideo.o dependencies to CONFIG_MPEGVIDEO
  Move MASK_ABS macro to libavcodec/mathops.h
  x86: move MANGLE() and related macros to libavutil/x86/asm.h
  x86: rename libavutil/x86_cpu.h to libavutil/x86/asm.h
  aacdec: Don't fall back to the old output configuration when no old configuration is present.
  rtmp: Add message tracking
  rtsp: Support mpegts in raw udp packets
  rtsp: Support receiving plain data over UDP without any RTP encapsulation
  rtpdec: Remove an unused include
  rtpenc: Remove an av_abort() that depends on user-supplied data
  vsrc_movie: discourage its use with avconv.
  avconv: allow no input files.
  avconv: prevent invalid reads in transcode_init()
  avconv: rename OutputStream.is_past_recording_time to finished.

Conflicts:
	configure
	doc/filters.texi
	ffmpeg.c
	ffmpeg.h
	libavcodec/Makefile
	libavcodec/aacdec.c
	libavcodec/mpegvideo.c
	libavformat/version.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-08-09 19:31:56 +02:00
Mans Rullgard
d7a4f8f8b9 Move MASK_ABS macro to libavcodec/mathops.h
This macro is only used in two places, both in libavcodec, so this
is a more sensible place for it.

Two small tweaks to the macro are made:

- removing the trailing semicolon
- dropping unnecessary 'volatile' from the x86 asm

Signed-off-by: Mans Rullgard <mans@mansr.com>
2012-08-09 00:58:20 +01:00
Michael Niedermayer
c065255bba Merge remote-tracking branch 'qatar/master'
* qatar/master:
  aacenc: Fix LONG_START windowing.
  aacenc: Fix a bug where deinterleaved samples were stored in the wrong place.
  avplay: use the correct array size for stride.
  lavc: extend doxy for avcodec_alloc_context3().
  APIchanges: mention avcodec_alloc_context()/2/3
  avcodec_align_dimensions2: set only 4 linesizes, not AV_NUM_DATA_POINTERS.
  aacsbr: ARM NEON optimised sbrdsp functions
  aacsbr: align some arrays
  aacsbr: move some simdable loops to function pointers
  cosmetics: Remove extra newlines at EOF

Conflicts:
	libavcodec/utils.c
	libavfilter/formats.c
	libavutil/mem.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2012-01-29 00:34:59 +01:00
Alex Converse
7181c4edee cosmetics: Remove extra newlines at EOF 2012-01-27 17:19:09 -08:00
Michael Niedermayer
91eb1b1525 Merge remote-tracking branch 'qatar/master'
* qatar/master: (22 commits)
  prores: add FATE tests
  id3v2: reduce the scope of some non-globally-used symbols/structures
  id3v2: cosmetics: move some declarations before the places they are used
  shorten: remove the flush function.
  shn: do not allow seeking in the raw shn demuxer.
  avformat: add AVInputFormat flag AVFMT_NO_BYTE_SEEK.
  avformat: update AVInputFormat allowed flags
  avformat: don't unconditionally call ff_read_frame_flush() when trying to seek.
  truespeech: use sizeof() instead of hardcoded sizes
  truespeech: remove unneeded variable, 'consumed'
  truespeech: simplify truespeech_read_frame() by using get_bits()
  truespeech: decode directly to output buffer instead of a temp buffer
  truespeech: check to make sure channels == 1
  truespeech: check for large enough output buffer rather than truncating output
  truespeech: remove unneeded zero-size packet check.
  mlpdec: return meaningful error codes instead of -1
  mlpdec: remove unnecessary wrapper function
  mlpdec: only calculate output size once
  mlpdec: validate that the reported channel count matches the actual output channel count
  pcm: reduce pointer type casting
  ...

Conflicts:
	libavformat/avformat.h
	libavformat/id3v2.c
	libavformat/id3v2.h
	libavformat/utils.c
	libavformat/version.h

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2011-10-14 03:48:22 +02:00
Mans Rullgard
f59bb3d8f3 mathops: remove undefined behaviour from sign_extend()
This function intentionally overflows the signed range on
the left shift.  Using this type-punning avoids errors from
the overflow checker without disabling this test globally.

Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-10-13 15:40:16 +01:00
Michael Niedermayer
46eb300d01 Merge remote-tracking branch 'qatar/master'
* qatar/master:
  ARM: remove MULL inline asm
  mathops: use MUL64 macro where it forms part of other ops
  tty: factorise returning error codes.
  rawdec: add framerate private option.
  x11grab: add framerate private option.
  fbdev,v4l2: remove some forgotten uses of AVFormatParameters.time_base.
  bktr: don't error when AVFormatParameters.time_base isn't set.
  cmdutils: add missing const qualifier
  Skip headers not designed to work standalone during 'make checkheaders'.
  Add missing #includes to make headers self-contained.
  musepack: remove unnecessary #include from mpcdata.h
  musepack: remove extraneous mpcdata.h inclusions
  Fix error check in av_file_map()

Conflicts:
	cmdutils.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2011-06-05 03:00:31 +02:00
Mans Rullgard
95912731c2 mathops: use MUL64 macro where it forms part of other ops
Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-06-04 20:10:20 +01:00
Diego Biurrun
8d459acc10 Add missing #includes to make headers self-contained.
This fixes 'make checkheaders'.
2011-06-04 18:15:50 +02:00
Michael Niedermayer
f8ae3a2108 Merge remote branch 'qatar/master'
12 files changed, 36 insertions(+), 81 deletions(-)
yes thats 36 new lines in 14 commits

* qatar/master:
  ffmpeg: fix -aspect cli option
  Restructure video filter implementation in ffmpeg.c.
  ffplay: remove audio_write_get_buf_size() forward declaration
  lavfi: print key-frame and picture type information in ff_dlog_ref()
  mathops: remove ancient confusing comment
  cws2fws: Improve error message wording.
  tools: Check the return value of write().
  mpegaudio: move OUT_FMT macro to mpegaudiodec.c
  mpegaudio: remove OUT_MIN/MAX macros
  Add missing #includes to mp3_header_(de)compress bsf
  dct: fix indentation
  dct: bypass table allocation for DCT_II of size 32
  h264dsp_mmx: Add #ifdefs around some mmxext functions on x86_64.
  Remove unused header mpegaudio3.h.

Conflicts:
	ffmpeg.c
	libavcodec/mpegaudio.h
	libavcodec/mpegaudio3.h
	libavfilter/avfilter.c

Merged-by: Michael Niedermayer <michaelni@gmx.at>
2011-05-17 04:51:33 +02:00
Mans Rullgard
005db47011 mathops: remove ancient confusing comment
Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-05-16 21:34:39 +01:00
Mans Rullgard
2912e87a6c Replace FFmpeg with Libav in licence headers
Signed-off-by: Mans Rullgard <mans@mansr.com>
2011-03-19 13:33:20 +00:00
Måns Rullgård
cf60d669a6 Add more int packing macros, name them consistently
Originally committed as revision 23916 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-06-30 22:33:38 +00:00
Ronald S. Bultje
157cb0694a Rename PACK4x8() to PACK4UINT8().
Originally committed as revision 23721 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-06-22 20:57:00 +00:00
Ronald S. Bultje
b39a2d19d4 Add a macro to pack 4 bytes into native byte-order so they can be written
at once using a single 32-bit store.

Originally committed as revision 23715 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-06-22 19:15:27 +00:00
Michael Niedermayer
e969e261a5 Use standard C for implementing sign_extend() and zero_extend().
This fixes compilation of probetest

Originally committed as revision 23116 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-13 16:32:21 +00:00
Måns Rullgård
48960b8f24 Add zero_extend() function
Originally committed as revision 21947 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-02-21 21:48:37 +00:00
Måns Rullgård
5e46be96f8 Move NEG_[US]SR32 macros to mathops.h
Originally committed as revision 21873 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-02-17 23:58:59 +00:00
Måns Rullgård
f2c694af28 Add UMULH()
This function/macro returns the high half of the unsigned 64-bit
product of two unsigned 32-bit operands.

Originally committed as revision 21463 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-26 18:05:55 +00:00
Måns Rullgård
5e7dfb7de1 Move COPY3_IF_LT to lavc/mathops.h
This obscure macro is only used in motion_est.c so having it in lavc
makes more sense.  See discussion here:
http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2008-November/056561.html

Originally committed as revision 21346 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-01-20 06:01:54 +00:00
Måns Rullgård
b7904f78c1 AVR32: optimised mathops.h
44% faster MP3 decoding

Originally committed as revision 19667 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-08-17 02:23:53 +00:00
Måns Rullgård
9d48e28f00 MIPS: add MIPS64 and Loongson asm for MAC64/MLS64
Originally committed as revision 18979 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-28 23:19:32 +00:00
Måns Rullgård
e20a4f53a0 mathops: sort arch-specific #include lines
Originally committed as revision 18886 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-20 18:49:55 +00:00
Måns Rullgård
26f6a57467 mathops: whitespace cosmetics
Originally committed as revision 18885 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-20 18:49:52 +00:00
Måns Rullgård
101dfa7d0a Add sign_extend() function to mathops.h
Originally committed as revision 17738 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-03-03 00:03:55 +00:00
Loren Merritt
6166516d1f re-enable mid_pred asm on x86_64. (broke in r16681)
Originally committed as revision 17058 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-02-08 17:44:04 +00:00
Diego Biurrun
406792e7b0 cosmetics: Remove pointless period after copyright statement non-sentences.
Originally committed as revision 16684 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-01-19 15:46:40 +00:00
Aurelien Jacobs
199436b952 moves mid_pred() into mathops.h (with arch specific code split by directory)
Originally committed as revision 16681 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-01-18 22:57:40 +00:00
Aurelien Jacobs
b250f9c66d Change semantic of CONFIG_*, HAVE_* and ARCH_*.
They are now always defined to either 0 or 1.

Originally committed as revision 16590 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-01-13 23:44:16 +00:00
Diego Biurrun
79cb09b2f7 consistency cosmetics: Rename POWERPC identifiers to PPC.
Originally committed as revision 16359 to svn://svn.ffmpeg.org/ffmpeg/trunk
2008-12-27 11:33:26 +00:00
Diego Biurrun
a6493a8fbd Rename libavcodec/i386/ --> libavcodec/x86/.
It contains optimizations that are not specific to i386 and
libavutil uses this naming scheme already.

Originally committed as revision 16270 to svn://svn.ffmpeg.org/ffmpeg/trunk
2008-12-22 09:12:42 +00:00
Måns Rullgård
a2fc0f6a6d ARM: replace "armv4l" with "arm"
Originally committed as revision 16179 to svn://svn.ffmpeg.org/ffmpeg/trunk
2008-12-17 00:54:54 +00:00
Måns Rullgård
4deaa94639 Add shift argument to MULL() macro
This replaces use of FRAC_BITS in the MULL() definition with a third
argument specifying the shift amount.  All uses of this macro are
updated to pass FRAC_BITS as third argument.

Originally committed as revision 15921 to svn://svn.ffmpeg.org/ffmpeg/trunk
2008-11-24 09:28:55 +00:00