1
0
mirror of https://github.com/FFmpeg/FFmpeg.git synced 2024-11-21 10:55:51 +02:00
Commit Graph

19966 Commits

Author SHA1 Message Date
Reimar Döffinger
10f9ff9bf6 Set AV_PKT_FLAG_KEY in avcodec_decode_video.
This fixes decoding of MPNG, which currently is incorrectly decoded as
CorePNG delta frames.
CorePNG can not be correctly decoded via this old API either way.

Originally committed as revision 19079 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-06-03 11:23:16 +00:00
Reimar Döffinger
8f3f2e0676 Add AV_PKT_FLAG_KEY and deprecate PKT_FLAG_KEY.
Originally committed as revision 19078 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-06-03 11:20:35 +00:00
Michael Niedermayer
da45d5b9c6 Clarify the input AVPacket for avcodec_decode*()
Originally committed as revision 19077 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-06-03 10:25:21 +00:00
Kostya Shishkov
5d5a8813fc Testing RGB48 variants requires bigger stride in swscale-example.c
Originally committed as revision 29343 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale
2009-06-02 15:35:58 +00:00
Kostya Shishkov
9706d1c766 Check combined depth and number of components in TIFF decoder, thus eliminating
additional checks for each depth (like 48-bit gray vs. 48-bit RGB)

Originally committed as revision 19076 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-06-02 15:31:51 +00:00
Kostya Shishkov
68e7f4828d YUV into RGB48 BE/LE conversion support
Originally committed as revision 29342 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale
2009-06-02 12:30:11 +00:00
Kostya Shishkov
e8417235d0 Partial (low bits ignored, no direct transcoding into other RGB formats) support
for inputting RGB48BE/LE.

Originally committed as revision 29341 to svn://svn.mplayerhq.hu/mplayer/trunk/libswscale
2009-06-02 12:28:49 +00:00
Baptiste Coudurier
b2984add80 do not parse full header for private streams
Originally committed as revision 19075 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-06-02 07:53:44 +00:00
Baptiste Coudurier
617c461625 correctly handle maximum streams limit
Originally committed as revision 19074 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-06-02 07:30:19 +00:00
Baptiste Coudurier
1c4df2ab24 export data from private streams
Originally committed as revision 19073 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-06-02 07:26:58 +00:00
Stefano Sabatini
0802356cfc Remove the const qualifier for the name field of AVFilterInOut, since
it is meant to be freed.
Fix warnings.

Originally committed as revision 19072 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-06-01 16:07:48 +00:00
Ramiro Polla
9e1cc5984c indent
Originally committed as revision 19071 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 20:19:16 +00:00
Ramiro Polla
851218584e Remove useless if(), leftover from the vhook removal.
Originally committed as revision 19070 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 20:17:58 +00:00
Reimar Döffinger
24dd21fa41 add #if CONFIG_ZLIB_DECODER around zlib_decomp function.
Fixes compilation when zlib is not available.

Originally committed as revision 19069 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 19:51:21 +00:00
Reimar Döffinger
79183d3c3e mszh decompression: add a special case for an all-0 mask, i.e. 32 uncompressed
bytes in a row.
About 15% faster mszh_decomp on an Atom N270 for
http://samples.mplayerhq.hu/V-codecs/mszh-zlib/avimzsh_sample.avi

Originally committed as revision 19068 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 18:17:33 +00:00
Reimar Döffinger
a7bfbe4e13 lcldec: ensure that the offset for av_memcpy_backptr is valid.
Originally committed as revision 19067 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 17:44:35 +00:00
Stefano Sabatini
4ec42240ca Implement avfilter_graph_config_links().
Originally committed as revision 19066 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 16:30:12 +00:00
Reimar Döffinger
c70ac8e7f5 I will take care of the lcl de-/encoder while Roberto is unavailable.
Originally committed as revision 19065 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 11:50:30 +00:00
Reimar Döffinger
bbf306742c Fix memleak due to c->decomp_buf never being freed.
Originally committed as revision 19064 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 11:47:52 +00:00
Reimar Döffinger
908425c758 Only call inflateEnd when we were actually using the zlib code.
Originally committed as revision 19063 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 11:42:50 +00:00
Reimar Döffinger
4ad3df9da2 avctx->priv_data is initialized to 0, get rid of useless extra initialization.
Originally committed as revision 19062 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 11:41:49 +00:00
Reimar Döffinger
7ffc844742 Merge variable declaration and initialization.
Originally committed as revision 19061 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 11:35:20 +00:00
Reimar Döffinger
7c20421661 Use int-size types instead of char where it makes no difference.
Originally committed as revision 19060 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 11:34:19 +00:00
Reimar Döffinger
7a15b25827 Take advantage of available input padding to optimize mszh_decomp
Originally committed as revision 19059 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 11:32:03 +00:00
Reimar Döffinger
a5d2745d8e Change maskbit variable to contain (1 << maskbit)
Originally committed as revision 19058 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:58:20 +00:00
Reimar Döffinger
e0b3af9059 Remove check that thanks to padding is no longer necessary.
Originally committed as revision 19057 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:53:47 +00:00
Reimar Döffinger
0b99764459 100l, the compression field in lcl extradata must be interpreted as
int8_t, not uint8_t to allow -1 for "no compression".
The original code worked, but only when char was signed.

Originally committed as revision 19056 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:50:54 +00:00
Reimar Döffinger
e4f141ad60 One more use for FFMIN.
Originally committed as revision 19055 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:44:42 +00:00
Reimar Döffinger
0d48e6ec7f Pad the decompression buffer and use av_memcpy_backptr for the mszh decompression.
Originally committed as revision 19054 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:42:26 +00:00
Reimar Döffinger
974ce78544 Mark srcptr as const in mszh_decomp
Originally committed as revision 19053 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:38:17 +00:00
Reimar Döffinger
6065dcea37 Remove unnecessary put_bits/get_bits includes.
Originally committed as revision 19052 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:37:05 +00:00
Reimar Döffinger
2bee59083d Use bytestream_get_le16 to simplify offset/count calculation for mszh decompression.
Originally committed as revision 19051 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:35:39 +00:00
Reimar Döffinger
5a18221ba8 Remove useless & 0x1f
Originally committed as revision 19050 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:31:37 +00:00
Reimar Döffinger
7b151a98b0 Document padding requirements of mszh_decomp srcptr buffer
Originally committed as revision 19049 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:28:40 +00:00
Reimar Döffinger
5e5fe7dab5 Use srcptr_end variable to avoid having to update both srcptr and srclen.
Originally committed as revision 19048 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:26:58 +00:00
Reimar Döffinger
dbc53ffc7c Change buffer size checks to avoid the undefined overflow case.
Originally committed as revision 19047 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:23:38 +00:00
Reimar Döffinger
2fddb91e94 Simply use memcpy instead of AV_RN32/AV_WN32 combination.
Originally committed as revision 19046 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:16:28 +00:00
Reimar Döffinger
661cb0d4a9 Fix decoding of multithread-encoded lcl files on big-endian.
Originally committed as revision 19045 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:14:27 +00:00
Reimar Döffinger
ce22c7d075 Remove another useless ()
Originally committed as revision 19044 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:05:21 +00:00
Reimar Döffinger
21bf2f51d7 Correct calculation of compressed input length.
Originally committed as revision 19043 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 10:01:26 +00:00
Reimar Döffinger
46adcc27a2 Add sanity check for mthread_inlen, avoids crashes due to invalid reads.
Originally committed as revision 19042 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:59:46 +00:00
Reimar Döffinger
1207c84adb Use FFMIN
Originally committed as revision 19041 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:57:42 +00:00
Reimar Döffinger
c2c0c1c023 Factor out zlib decompression code to avoid massive code duplication,
particularly due to error checks.

Originally committed as revision 19040 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:51:46 +00:00
Reimar Döffinger
f005808bf2 Use FFALIGN
Originally committed as revision 19039 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:28:29 +00:00
Reimar Döffinger
46c02b1b50 Move variable into block where it is used, avoiding a unused variable
warning if the zlib decoder is disabled.

Originally committed as revision 19038 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:27:07 +00:00
Reimar Döffinger
2b550f816a Make lcldec less annoyingly verbose, move messages from AV_LOG_INFO to AV_LOG_DEBUG.
Originally committed as revision 19037 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:23:54 +00:00
Reimar Döffinger
75f6d6c7e6 Get rid of unreachable code: avctx->codec_id == CODEC_ID_ZLIB is not possible
here when the zlib decoder is disabled and libavcodec is used correctly.

Originally committed as revision 19036 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:19:21 +00:00
Reimar Döffinger
56b44694bd lcldec.c: change #if CONFIG_ZLIB to #if CONFIG_ZLIB_DECODER.
The zlib related code should not be compiled in when the decoder is disabled
and it thus will never be used, even if we have zlib available.

Originally committed as revision 19035 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:16:06 +00:00
Reimar Döffinger
445e0ab539 Get rid of extradata casts, it already has the right uint8_t * type
Originally committed as revision 19034 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:12:17 +00:00
Reimar Döffinger
8c18e490c4 Make lcldec produce YUV output when the input file is coded like that, instead
of having it do its own inefficient fixed-point YUV to RGB conversion.

Originally committed as revision 19033 to svn://svn.ffmpeg.org/ffmpeg/trunk
2009-05-31 09:09:32 +00:00