V4L2 support for OpenBSD, patch by Brad, brad comstyle com

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15376 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Make MBAFF temporal direct mode closer to the spec.
Fixes at least:
CAMANL2_TOSHIBA_B
CVMANL2_TOSHIBA_B
camp_mot_mbaff0_full

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15375 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Move direct_dist_scale_factor() and direct_ref_list_init() after fill_mbaff_ref_list()
as future changes will need the mbaff lists inited.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15374 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Set poc for field references in MBAFF.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15373 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Rename check_func2 to check_func_headers.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15372 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Log an error message in case of invalid number of channels.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15371 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove fixme that is based on the assumption of !direct_8x8_inference && MBAFF.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15370 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove check and error message for "MBAFF + !direct_8x8_inference is not implemented"
because this mode does not exist, H.264-2007 says "When frame_mbs_only_flag is
equal to 0, direct_8x8_inference_flag shall be equal to 1."

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15369 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Avoid undefined behavior for removing elements that where not in the tree.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15368 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Ensure that muxer and encoder aspect ratios match. So no
invalid files are generated.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15367 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Support comments in preset files.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15366 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Make pps/sps scaling_matrix initialization closer to the spec.
This does not affect any of the conformance streams.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15365 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Zero 'idx' for each iteration of the coupling gain loop and as it is only used
within this loop, we can move its declaration there too.

Based on a patch by Alex Converse (alex converse gmail com)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15364 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetics: missing brackets

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15363 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cast the return value of get_bits() to signed integer such that the subtraction
is conducted as a signed calculation.

Patch by Alex Converse (alex converse gmail com)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15362 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Ensure that extradata is not parsed twice.
Fixes FRExt1_Panasonic.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15361 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

add newline at the end of message

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15360 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

ZMBV encoder forgot to fill coded frame pointer

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15359 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

small wording fixes

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15358 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add fourcc for MS RLE4

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15357 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Factorize out code used for MS RLE format decoding in different decoders.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15356 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove duplication of TRANSPOSE8 macro, also removing some redundant load/stores

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15355 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Make a separate macro for writing an entire block from registers to memory

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15354 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Factorize SSE2_(Row|Column)_IDCT into one macro

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15353 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

revert r14983, value is not sample aspect ratio, it is display aspect ratio

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15352 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetics: remove excessive blank lines

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15351 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Ensure MMX/SSE2 VP3 IDCT selection isn't disabled when only Theora is enabled

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15350 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetics: reindent

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15349 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

libx264 ffpresets:
 * default - x264 CLI default values
 * fastfirstpass - disables options unnecessary for first pass
 * normal - Dark Shikari’s Q55 preset
 * hq - Dark Shikari’s Q75 preset
 * max - all options maxed out (at time of writing)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15348 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Rewrite SSE2 VP3 IDCT in inline asm

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15347 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Rewrite MMX VP3 IDCT in inline asm

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15346 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

MMX/SSE2 VP3 IDCT are bitexact now that the dequantization matrices are permutated correctly

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15345 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Take idct permutation into account when initializing dequantization matrices

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15344 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Use SAMPLE_FMT_FLT instead of doing the float->int convertion in the
decoder. This changes very little the output and any difference should
be inaudible.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15343 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Second-stage gain and Moving Average Predictor codebooks.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15342 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Another set of approved changes for G.729: includes, formats structure,
several parts of main decoding routine and initialization code.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15341 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Validate pulse position and error out if an invalid position is encountered.

Patch by Alex Converse (alex converse gmail com)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15340 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Only print "invalid new backstep" when its really invalid.
Fixes issue541.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15339 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix regression test failure caused by golomb limit not being considered
in r15334.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15338 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

sync to latest x264 git, field was renamed

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15337 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Make decoder more robust by default against broken encoders.
Fixes issue540.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15336 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix issue634.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15335 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix issue245.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15334 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

wording/spelling

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15333 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: fix indentation after r15321

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15332 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

write actual min and max frame size to FLAC header. update regression test checksum.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15331 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

cosmetics: Fix indentation.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15330 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove useless #ifdef around #include.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15329 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Only define x86 register names on x86.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15328 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Spelling...

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15327 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

write MD5 checksum to FLAC header. update regression tests.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15326 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

update FLAC regression test due to expected change after r15324

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15325 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

write number of samples in FLAC extradata.
based on a patch by Mathieu Velten (matmaul gmail com).

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15324 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add , to clarify sentence.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15323 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

add error message for unsupported codecs in WAVE.
Resolves issue 623.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15322 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Return error when trying to decode non-grayscale 16-bit PNM images.
Fixes issue 566.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15321 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix typos found by ispell.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15320 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Reword a few sentances to make them clearer.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15319 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Reindent after r15317.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15318 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Rename RTSP_SERVER_RDT to REAL, because RDT (the transport protocol) is not
strictly related to the server type (Real servers can stream both RDT and
RTP).

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15317 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

The official guide to swscale for confused developers.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15316 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Reindent after last commit

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15315 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Apply pulses to unscaled coefficients rather than scaled.

Slight reworking of a patch by Alex Converse (alex converse gmail com)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15314 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Operands of '+' are not necessarily evaluated in order so r15311 was pointless.
Splitting the assignment onto two lines should resolve the order issue.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15313 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Update link to FFmpeg Windows Help website.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15312 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Correct order of parsing for pulse scalefactor band and offset to match the
specification.

Patch by Alex Converse (alex converse gmail com)

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15311 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix setting default value of scan_offset option for target svcd.

Patch by Gert Vervoort gert D vervoort A hccnet D nl

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15310 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add sparc maintainer.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15309 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetics: line breaks

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15308 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Simplify: use a single history buffer for gain and a single one for
speech instead of having two for each in the context.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15307 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix compilation on Mac OS X 10.4: Defining _POSIX_C_SOURCE hides the u_char &
similar typedefs that sysctl.h needs. Since sysctl() itself isn't POSIX
undefining _POSIX_C_SOURCE for check_altivec.c seems the best way to fix this.
patch by David Conrad lessen42 at gmail com

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15306 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Switch regression tests to swscale.
Plain C, x86-32 and -64 have been tested and should work, other
archs that had asm optmizations in swscale likely will need some fixes
to either fallback to C if SWS_BITEXACT is set or make the asm match C.
This also disables the PAL8 test as neither swscale or the old scaler
really support PAL8 output, imgconvert supported a fixed 666 palette
as output and swscale supports fixed 884 and 422.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15305 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Reindent after last commit

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15304 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Only read some of TNS bitstream data in the case that the TNS filter order is
non-zero as per the specification.

Patch by Alex Converse (alex converse gmail com)
Fixes another issue pertaining to issue632

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15303 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: simplify, first_timecode is already in the index
no need to duplicate it

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15302 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: prevent seeking before the first keyframe of the reference stream

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15301 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetics: align spec references

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15300 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Rename function: s/colmult/apply_window/

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15299 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Reindent after r15927, see discussion in "[PATCH] rtsp cleanup part 1:
remove duplicate code" thread on ML.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15298 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Factorize out common code for opening of the RTP parsing context between
SDP and RTSP into a new function. See discussion on ML in "[PATCH] rtsp
cleanup part 1: remove duplicate code" thread.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15297 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

misc spelling/wording/grammar fixes

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15296 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Make it possible to disable loop filter in VC-1.
Unlike H.264 it does not affect picture quality on any files I've watched.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15295 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Correct pulse amplitude application - a negative or 0 coefficient implies the
pulse is subtracted, else it is added. Also avoid a divide by 0.

Based on a patch by Alex Converse (alex converse gmail com)
Fixes part of issue632

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15294 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Fix incorrect printing of brainfart cropping error in some cases in progressive mode.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15293 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Convert table to integer.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15292 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Clarify convergence_duration.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15291 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Convert two tables to signed decimal

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15290 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

set bits_per_coded_sample according to new lavc api

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15289 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Cosmetics: remove useless parenthesis

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15288 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Remove ChannelData.lev var and use ChannelData.level instead

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15287 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

Add support for creating Simple Profile (I-frame only, no arithmetic coding)
Dirac bytestreams.  patch by Anuradha Suraparaju, anuradha rd.bbc.co uk

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15286 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

vorbis_enc : set coded_frame->pts value to running sample count.
This is a temporary fix for issue 502 and will not be required when
we start using AVFrame for audio codecs.

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15285 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: demux relevant subtitle packets after a seek
only works for subtitle tracks which where properly indexed up to the seek point

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15284 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: ensure that overlapping subtitle are not added to the index

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15283 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: after seeking, skeep up to the desired timestamp instead of
the desired stream

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15282 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: move timecode calculation a little earlier
so as to be able to use it in other part of the code

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15281 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

matroskadec: sanitize track time_scale before using it
fix issue628

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15280 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

print stream index in decimal not in hex, easier for debugging

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15279 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

remove one more stuff under #if LIBAVCODEC_VERSION_INT
which was forgotten in r15263

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15278 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b

set block align in av_write_header if not set, fix issue #489

git-svn-id: file:///var/local/repositories/ffmpeg/trunk@15277 9553f0bf-9b14-0410-a0b8-cfaf0461ba5b