Commit Graph

60 Commits

Author SHA1 Message Date
Nathan Caldwell 0bc01cc9fe psymodel: Add channels and channel groups to the psymodel. 2011-06-29 14:28:52 -07:00
Alex Converse 8dbaa5bd69 aacenc: Fix codebook trellising for zeroed bands.
Choose band type (codebook) zero, count its bits, and mark the other
states as unnavigable.
2011-06-27 21:47:45 -07:00
Mans Rullgard e65ab9d94f Remove unused variables 2011-06-02 20:06:00 +01:00
Reinhard Tartler 083e715f33 aac: workaround for compilation on cygwin
On cygwin, math.h needs to be included before float.h because of a bug
in the system headers. Including libavutil/libm.h first works around
this issue.

Longer discussion of the topic:
http://thread.gmane.org/gmane.comp.video.ffmpeg.devel/128582
2011-05-11 14:38:14 +02:00
Alex Converse d70fa4c423 Define POW_SF2_ZERO in aac.h and use for ff_aac_pow2sf_tabp[] offsets instead
of hardcoding 200 everywhere.
2011-04-27 12:39:37 -04:00
Michael Niedermayer 30fe971934 aac: add headers needed for log2f()
Signed-off-by: Anton Khirnov <anton@khirnov.net>
2011-04-26 08:33:21 +02:00
Young Han Lee 57cc1ad35f aacenc: reduce the number of loop index variables
Signed-off-by: Janne Grunau <janne-libav@jannau.net>
2011-04-04 09:12:37 +02:00
Young Han Lee 00f0b4b92e aacenc: indentation 2011-03-24 21:13:29 -07:00
Young Han Lee 92efa2bdd9 aacenc: fix the side calculation in search_for_ms 2011-03-24 21:12:23 -07:00
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
Nathan Caldwell 4afedfd8e5 aacenc: cosmetics, indentation, and comment clarification
Correct bad indentation in aaccoder
Clarify and correct comments in 3GPP psymodel, other cosmetics.
2011-03-16 23:35:39 -04:00
Young Han Lee f578854efc aaccoder: Change FFMAX for allzero flag to OR bit operation
Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com>
2011-03-16 23:12:43 -04:00
Alex Converse 63e1278d88 aacenc: TLS: Save maximum values for each swb in a table.
This gives an almost 20% speedup.

Originally committed as revision 24484 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-24 17:11:51 +00:00
Alex Converse 031d5cea04 10l: Remove some commented out code that slipped in.
Originally committed as revision 24483 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-24 17:01:50 +00:00
Alex Converse fe461767e6 aacenc: TLS: Try to preserve some energy in each non-zero band.
Reduce scalefactors in non-zero bands that underflow by twice as much as those
in bands that just fail to hit psy targets.

Originally committed as revision 24482 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-24 16:36:01 +00:00
Alex Converse c226fc5bfb aacenc: Prevent premature termination of the two loop search.
Originally committed as revision 24476 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-24 04:23:26 +00:00
Alex Converse 51ffd3a62f aacenc: Favor log2f() and sqrtf() over log2() and sqrt().
Originally committed as revision 24473 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-24 02:10:59 +00:00
Alex Converse 04d72abf17 aacenc: Factorize some scalefactor utilities.
Originally committed as revision 24472 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-23 22:37:42 +00:00
Alex Converse 581a96976b aacenc: Convert if () abort() to assert().
Originally committed as revision 24336 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-19 20:53:58 +00:00
Alex Converse 6d9f1b67e8 Cosmetics: Whitespace
Originally committed as revision 24333 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-19 18:37:45 +00:00
Alex Converse a2310d1e41 10l: Add a missing const.
Originally committed as revision 24274 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-16 20:28:55 +00:00
Alex Converse 1676b09994 aacenc: Template quantize_and_encode_band_cost().
Originally committed as revision 24273 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-07-16 20:02:46 +00:00
Alex Converse 804d489c4a aacenc: Fill in the estimated codebook for the final computed scalefactor in the TLS.
Originally committed as revision 23869 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-06-28 21:42:57 +00:00
Alex Converse c91dce994d aacenc: Remove an unnecessary division from the TLS.
Originally committed as revision 23368 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-28 21:29:56 +00:00
Alex Converse 911fbc452b Cosmetics: whitespace
Originally committed as revision 23367 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-28 20:45:35 +00:00
Alex Converse 04d6a54e13 aacenc: Remove unnecessary variables and scopes in the TLS.
Originally committed as revision 23366 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-28 20:42:01 +00:00
Alex Converse 9069b7d35f Fix declaration after statement
Originally committed as revision 23331 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-25 23:25:29 +00:00
Alex Converse 0ecfa7b7c7 aacenc: Split find_max_val() from find_min_book() to eliminate duplicate searches.
Originally committed as revision 23322 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-25 18:35:01 +00:00
Alex Converse af2ee6fc49 Remove useless costly inf checks from the trellis scalefactor search.
Originally committed as revision 23319 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-25 18:32:59 +00:00
Alex Converse acc9f51fe1 aacenc: Trellis over scalefactors using an estimated codebook rather than every codebook.
The minimal codebook to encode the band without clipping is used (as is done in the TLS).

Originally committed as revision 23318 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-25 18:31:55 +00:00
Alex Converse 144c5e3dd2 aacenc: Only trellis over a column of 61 scalefactors (reduced from 256).
This still provides plenty of dynamic range, makes every move legal, and greatly reduces the search space.

Originally committed as revision 23317 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-25 18:28:18 +00:00
Alex Converse 05e659ef83 aacenc: Factor out find_min_book so it can be used by multiple coefficient coders.
Originally committed as revision 23316 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-25 18:24:16 +00:00
Alex Converse 0175d7886f fmaxf -> FFMAX to fix pre-C99 systems
Originally committed as revision 23138 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-14 19:38:17 +00:00
Alex Converse 759510e6af aacenc: Add a rate only trellis for codebook selection for the TLS.
Originally committed as revision 23136 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-14 16:49:51 +00:00
Alex Converse 4d986b7172 aacenc: Use exact values when quantizing, not fuzzy values.
This requires us to code small escapes; we can't avoid it.

Originally committed as revision 23135 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-14 16:49:40 +00:00
Alex Converse 51cbabc792 aacenc: Use an estimated codebook for the TLS (two loop search).
Originally committed as revision 23134 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-14 16:49:30 +00:00
Alex Converse 46174079c9 10l: store the result of clipping added in r23035
Originally committed as revision 23037 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-06 20:38:42 +00:00
Alex Converse 32fa7725d2 Make the faac inspired quantizer search make sense for a slightly narrower definition of "make sense."
Originally committed as revision 23035 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-05-06 20:18:36 +00:00
Diego Biurrun ba87f0801d Remove explicit filename from Doxygen @file commands.
Passing an explicit filename to this command is only necessary if the
documentation in the @file block refers to a file different from the
one the block resides in.

Originally committed as revision 22921 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-04-20 14:45:34 +00:00
Alex Converse 508f092a78 aacenc: Merge quantize_band_cost() with quantize_and_encode_band().
If these two functions aren't matched results may be unexpected.

Originally committed as revision 22655 to svn://svn.ffmpeg.org/ffmpeg/trunk
2010-03-24 17:09:21 +00:00
Alex Converse f5e82fec3d aacenc: Use preprocessor constants for trellis states and stages.
Originally committed as revision 19948 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-09-21 04:32:51 +00:00
Alex Converse 9072c29e40 aacenc: Split paths in the scalefactor selection trellis into a 2-D array.
Originally committed as revision 19947 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-09-21 04:15:12 +00:00
Alex Converse 9d4f6f10ae aacenc: Replace cbrt() with cbrtf() when the result is destined for float
storage.

Originally committed as revision 19946 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-09-21 03:53:39 +00:00
Alex Converse 0bd9aa449a aacenc: Multiple distortion by lambda after it's summed and not each individual
term before.

Originally committed as revision 19945 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-09-21 03:50:59 +00:00
Alex Converse 06d6b962e5 Cosmetics: Reindent after r19943.
Originally committed as revision 19944 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-09-21 03:42:54 +00:00
Alex Converse a71e9b6254 aacenc: Don't make unnecessary compares to the escape value in tight loops.
Originally committed as revision 19943 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-09-21 03:41:08 +00:00
Alex Converse dd0e43e4bb Re-add search_for_ms to the coefficients encoders to silence warnings
Originally committed as revision 19733 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-08-27 17:46:13 +00:00
Alex Converse a62d6cfeff Be sure to increment our position in the coefficient array when skipping a zero
band in the twoloop scalefactor search.

Originally committed as revision 19480 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-07-22 03:53:30 +00:00
Alex Converse 5cf20d07ad Remove an unused field from the BandCodingPath struct.
Originally committed as revision 19473 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-07-20 21:52:03 +00:00
Alex Converse 0193992252 Actually use all the codebooks we are iterating over in the two-loop scalefactor search.
Originally committed as revision 19472 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-07-20 20:32:27 +00:00