Use ff_pw_8 in MMX/SSE VP3 IDCT

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15053 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Let ff_pw_8 be used as an SSE constant

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15052 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: update comment about dithering

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15051 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove the cos_mod table that has become unused after replacing the imdct
by ff_imdct_half().

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15050 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Replace obfuscated mdct in qmf_32_subbands() by ff_imdct_half().

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15049 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Get rid of the 512 sample memmove().

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15048 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add missing FFMPEG_ prefix to multiple inclusion guard.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15047 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove 'currently unused' warning about SAMPLE_FMT_xxx.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15046 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

timestamps in flv are 32bits _signed_ according to flash player, specs are wrong

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15045 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

remove now useless casts

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15044 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

make has_codec_parameters not returning true when sample_fmt is not set

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15043 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

skip empty flv data packets, fix issue #602

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15042 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

factorize packet skip code

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15041 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

simplify size handling code

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15040 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Okayed parts of nellymoserenc.c

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15039 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove useless zeroing of a buffer

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15038 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix typo in element type

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15037 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove unnecessary portion of a condition. This fixes issue607 and avoids
another unintended bug where channel coupling could have been run on an SCE.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15036 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

remove vorbis dec hack, decoding will be done anyway to get frame size

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15035 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix typo.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15034 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Half the size of subband_fir_noidea and get rid of memmove & memset of it.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15033 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetics: vertical align in print_all_lib_versions().

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15032 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Make show_version()/show_banner() print the libswscale versions if
enabled.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15031 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add to the ffmpeg manual an example showing how to add streams to the
output.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15030 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Make X264_init() do not set anymore the rc_eq from avctx to the
libx264 context. Fix a crash due to X264_init() setting the libx264
rc_eq field to NULL when rc_eq was not defined.
This patch drops the possibility to set the rc_eq for libx264,
feature which is going to be removed from libx264 either.

See the thread: "Fix crash in libx264 when rc_eq is not set".

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15029 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

speex in flv demuxing

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15028 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics, typos, case

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15027 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15026 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

fix version string ul

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15025 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Move t1 and t2 declaration from the QMF code closer to where they are used.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15024 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Simplify (drop useless intermediate sum and diff arrays).

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15023 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

5,1l: samples_chanptr[] should be (const float *)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15022 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

dca: Set data_size outside block loop.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15021 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

dca: Do float -> int16 interleaving in-place using s->dsp.float_to_int16_interleave()

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15020 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

While APE changed container format in 3.98, frequency tables for range coding
were changed only in 3.99.
This fixes decoding of v3.98 APE file I have.

Eternal curses to the creator of such unstable format.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15019 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Bump libavcodec minor version number for r15017.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15018 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add AVCodecContext.bits_per_raw_sample field.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15017 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Change AVCodecContext.bits_per_sample to bits_per_coded_sample. Nb: This change will become active on the next libavcodec major version bump.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15016 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Do not truncate timestamps before the muxer as it makes simple
things like last_pts - pts rather tricky and is not good for anything.
Timestamps should be truncated just before storing when needed.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15015 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: change another "get" to "decode"

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15014 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: change "get" to "decode"

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15013 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: indentation

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15012 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

remove unneeded variable, dither_all

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15011 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Intial implementation of the DV100 (AKA DVCPRO HD) decoder and demuxer as
specified in SMPTE 370M

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15010 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix a typo.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15009 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Reword a sentence in a doxy making its interpretation clearer.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15008 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Introducing video_stype for holding the value of VAUX source pack
and simplifying dv_frame_profile

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15007 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Change implementation for ff_data_to_hex(), this is faster. See discussion on
mailinglist in "Realmedia patch" thread.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15006 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add generic IIR filter interface with Butterworth lowpass filter implementation
and remove obsoleted old lowpass filter.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15005 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix mid/side stereo buggy output zeroing, fixes issue264 part 2.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15004 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: rename a table

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15003 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Export data_to_hex() as private API in lavf, rename to ff_data_to_hex() and
move it from sdp.c into utils.c. Also add new header internal.h specifically
for lavf-specific internal API. See discussion in "Realmedia patch" thread
on mailinglist.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15002 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: indent after last commit

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15001 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

do not read exponent strategies from each block for E-AC-3 since they are in the header

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15000 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Two cosmetic changes for improved readability and ease of HDVCPRO HD integration

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14999 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetic change aimed at making it easier to see how bits are consumed

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14998 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Factoring out chan_slice / 27 into seq variable

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14997 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Simplification of how dct_mode affects decoding.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14996 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: cosmetics: indentation

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14995 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: seek to the last position known in the index before resorting to
full cluster parsing when seeking after the end of index

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14994 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetics (format ff_nelly_dequantization_table nicer)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14993 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Simplify the indexing logic in the av_encode() streams mapping code.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14992 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

add definitions and documentation for shared functions for E-AC-3 decoding

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14991 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

copy default coupling band structure from table to decoding context (used for E-AC-3)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14990 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

add more OKed parts of the E-AC-3 decoder: structural changes to support AHT mode mantissa decoding

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14989 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: rename some functions from *get_* to *decode_*

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14988 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Use RTPDynamicProtocol parse_sdp_a_line() handlers in case of unknown SDP
lines. This allows "private" SDP tags to be forwarded to the specific handler,
allowing protocol-specific handling of SDP data. See mailinglist discussion
in the "Realmedia patch" thread.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14987 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

simplify mpeg12 avctx->pix_fmt set

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14986 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

remove useless assignment (st is never read after this line)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14985 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: index timestamps are supposed to be in the timebase of the stream

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14984 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

mxfdec: set sample_aspect_ratio

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14983 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add a temp variable to ff_sine_window_init() to match the way wma.c computed
window before r14757. This fixes a regression introduced by r14757 on amd64.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14982 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add explicit (int) cast to i386 optimized MUL* macros.
Wrong result is returned when 16-bit value is passed as value.
Also fixes "Warning: using `%edx' instead of `%dx' due to `l' suffix".

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14981 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

(cosmetics) Remove incorrect comment.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14980 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

(cosmetics) Describe for which tests clipping is required.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14979 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

I'll maintain mxf muxer

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14978 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

(cosmetics) Remove wrong comments.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14977 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

MXF is a valid output format now.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14976 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remaining parts of GSoC MXF muxer by Zhentan Feng.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14975 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

(cosmetics) Align comments consistently.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14974 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Another set of definitions for G.729

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14973 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

vertical align

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14972 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Noone seems to want to maintain MPEG-PS so i will.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14971 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Replace cliping in ff_acelp_interpolate() by a check&av_log, this should be the
quickest way to find out if it is needed.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14970 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix the seek regressions to match changes in r14956, r14959 and r14960.
The changes are verified and desired.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14969 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Import more MXF muxer code from the SoC tree

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14968 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Missing static in float_to_int16_altivec declaration

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14967 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

When setting codec_id during codec probe we must also set codec_type.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14966 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Little-endian bitstream writer

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14965 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

fir_32bands_perfect[] contents were mixed, restore original order
Patch by Alexander Patrakov ($lastname <whirlpool> gmail.com)
Thread: dcadata.h: fir_32bands_perfect[] is wrong

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14964 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix udp.c compilation with recent glibc (>= 2.8)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14963 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

ARM: add prefetch function using ARMv5 PLD instruction

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14962 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

ARM: set STRIDE_ALIGN and DECLARE_ALIGNED_8 to 16 for NEON

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14961 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: unset matroska->done when seeking
just in case someone try to seek back after reaching the end of file

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14960 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: full seeking support in index-less files
when trying to seek past the last index entry, just parse more clusters to add them to the index, until an index entry match the desired position

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14959 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

right shift the values by 8 in GAQ remap "B" table. data now fits in int8_t.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14958 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: on the fly index construction for index-less files
each cluster that is read and that contains a video key frame is added
to the index

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14957 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: don't try to seek to negative timestamp
matroska timestamps are unsigned

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14956 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: move setting of matroska->done inside matroska_parse_cluster()

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14955 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

restore accidently deleted file

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@14954 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b