December 2009 - ffmpeg-devel - lists.ffmpeg.org

[PATCH] Add writing of vorbis comments to flac files
by james.darnley＠gmail.com 20 Mar '10

20 Mar '10

Hello, I have attached a patch which adds the creation of vorbis comments from AVMetadata and the writing of this as a METADATA_BLOCK_VORBIS_COMMENT in a flac file. Could someone please double check libavformat/makefile and the #includes on the files to eliminate any redundant options or to add any I have missed. I do have a question or two. Is it possible to determine the length of the audio from the AVFormatContext struct? I tried a couple of "duration" values but they always seemed to be 0. If that is not possible, is there a value in AVFormatContext which could be interpreted as a header padding value, a value which might request N bytes of padding? I ask because there is also the possibility of writing a METADATA_BLOCK_PADDING to allow extra metadata or possibly a seek table to be added later without the need to rewrite the whole file. Thanks for any comments.

9 46

[PATCH] Fix SDP demuxer dependencies
by martin＠martin.st 15 Mar '10

15 Mar '10

Hi, The SDP demuxer includes rtp_asf.o and rdt.o, which depend on the asf and rm demuxers. If compiling with e.g. these options, --disable-demuxers --disable-muxers --disable-protocols --disable-encoders --disable-decoders --disable-parsers --disable-bsfs --enable-protocol=udp --enable-demuxer=mpegts --enable-protocol=rtp --enable-demuxer=sdp one gets a linker failure. The attached patch fixes this issue. // Martin

6 12

[PATCH] Export large crop values via AVCodecContext
by lessen42＠gmail.com 13 Mar '10

13 Mar '10

Hi, As far as I can tell, it's impractical to crop large values internally, since buffers returned from decode_frame must have the same dimensions as AVCodecContext, which also must be the coded dimensions (or up to 15 less) in order for get_buffer to allocate buffers of the needed size. Thus, the only solution I see is to export these values and leave it to individual applications that care about such offsets to deal with it themselves.

3 6

[PATCH] Move error codes definitions from lavc to lavu
by stefano.sabatini-lala＠poste.it 13 Mar '10

13 Mar '10

Hi, as in subject, this is required if we want for example move the error handling code to lavu, as it uses it. Also reduces the clutter in avcodec.h. Regards. -- FFmpeg = Free & Funny Mean Portable Elaborated Generator

3 22

[PATCH] Is gray8 a paletted format?
by stefano.sabatini-lala＠poste.it 04 Mar '10

04 Mar '10

Hi all, I see that in imgconvert.c gray8 is treated like a paletted format: check for example ff_set_systematic_pal() and ff_fill_pointer(), where it is allocated space for its palette. If gray8 can be considered a paletted format, then I suggest to apply the attached patch, updating both the pixdesc array and the lsws macro. Regards. -- FFmpeg = Fundamental and Fantastic Most Porno Erudite Gadget

4 19

[PATCH] graph2dot tool
by stefano.sabatini-lala＠poste.it 25 Feb '10

25 Feb '10

Hi all, this neat tool should help to debug/understand avfilter graphs, if you find it useful we could include this in the tools dir. Usage example: $ echo "scale=100:100, crop=100:20:30:30, vflip, hflip" | graph2dot digraph G { "Parsed filter 0 (scale)" -> "Parsed filter 1 (crop)"; "Parsed filter 1 (crop)" -> "Parsed filter 2 (vflip)"; "Parsed filter 2 (vflip)" -> "Parsed filter 3 (hflip)"; } To directly display the graph: $ echo "scale=100:100, crop=100:20:30:30, vflip, hflip" | graph2dot | dot -Tpng -o graph.png; and display graph.png Regards. -- FFmpeg = Fundamental Fantastic MultiPurpose Extended Game

4 13

[PATCH] add timeout to udp_read
by kragen＠gmx.de 25 Feb '10

25 Feb '10

Hello, to fix issue 539 the attached patch introduces an timeout [seconds] which returns while waiting for an udp stream: ffplay udp://235.2.2.2:5000?recv_timeout=5 Hagen

6 28

[PATCH] '-longest' command-line argument realization
by pritula＠yandex-team.ru 22 Feb '10

22 Feb '10

Hello. I suggest patch to realize '-longest' key with following semantics: All video and audio streams with needed encoding in output file will be same time length and this length will be equal to longest stream in output file. For every shorter audio stream it will be continued with silence, and for every shorter video stream it will be continued with last picture, encoded to this stream. Waiting for your comments. The patch (to revision 16967) (doubled in attachment): --- ffmpeg.c.mplayerhq 2009-02-03 10:17:49.000000000 +0300 +++ ffmpeg.c.mplayerhq.mod 2009-02-03 10:29:12.000000000 +0300 @@ -192,6 +192,7 @@ static float audio_drift_threshold= 0.1; static int copy_ts= 0; static int opt_shortest = 0; +static int opt_longest = 0; static int video_global_header = 0; static char *vstats_filename; static FILE *vstats_file; @@ -286,6 +287,7 @@ is not defined */ int64_t pts; /* current pts */ int is_start; /* is 1 at the start and after a discontinuity */ + AVFrame last_picture; } AVInputStream; typedef struct AVInputFile { @@ -1260,6 +1262,7 @@ /* no picture yet */ goto discard_packet; } + ist->last_picture = picture; if (ist->st->codec->time_base.num != 0) { ist->next_pts += ((int64_t)AV_TIME_BASE * ist->st->codec->time_base.num) / @@ -2177,6 +2180,115 @@ } } + if (opt_longest) { + double opts_max = -1e100; + int istreams_needed[MAX_STREAMS]; + int nb_istreams_needed = 0; + for(i=0;i<nb_ostreams;i++) { + double opts; + ost = ost_table[i]; + if(ost->st->codec->codec_type == CODEC_TYPE_VIDEO) + opts = ost->sync_opts * av_q2d(ost->st->codec->time_base); + else + opts = ost->st->pts.val * av_q2d(ost->st->time_base); + if (opts > opts_max) + opts_max = opts; + } + for(i=0;i<nb_ostreams;i++) { + if ((ost_table[i]->st->codec->codec_type == CODEC_TYPE_VIDEO || ost_table[i]->st->codec->codec_type == CODEC_TYPE_AUDIO) && ost_table[i]->encoding_needed && ist_table[ost_table[i]->source_index]->decoding_needed) { + double opts, granularity; + ost = ost_table[i]; + os = output_files[ost->file_index]; + ist = ist_table[ost->source_index]; + if(ost->st->codec->codec_type == CODEC_TYPE_VIDEO) { + opts = ost->sync_opts * av_q2d(ost->st->codec->time_base); + granularity = av_q2d(ost->st->codec->time_base); + } else { + opts = ost->st->pts.val * av_q2d(ost->st->time_base); + granularity = av_q2d(ost->st->time_base); + } + if (granularity < 0) + granularity = 0; + if (opts + granularity / 2.0 < opts_max) { + int j; + for(j=0;j<nb_istreams_needed;j++) + if (ost->source_index == istreams_needed[j]) + break; + istreams_needed[j] = ost->source_index; + if (j >= nb_istreams_needed) + nb_istreams_needed = j + 1; + } + } + } + for(i=0;i<nb_istreams_needed;i++) { + int continue_flag = 0; + double input_granularity = 0; + ist = ist_table[istreams_needed[i]]; + switch (ist->st->codec->codec_type) { + case CODEC_TYPE_VIDEO: + input_granularity = av_q2d(ist->st->codec->time_base); + break; + default: + input_granularity = av_q2d(ist->st->time_base); + break; + } + do { + int j; + continue_flag = 0; + for(j=0;j<nb_ostreams;j++) { + if (ost_table[j]->encoding_needed && ost_table[j]->source_index == istreams_needed[i]) { + double opts = 0; + double granularity = 0; + ost = ost_table[j]; + os = output_files[ost->file_index]; + switch (ost->st->codec->codec_type) { + case CODEC_TYPE_AUDIO: { + uint8_t *samples = NULL; + int samples_size = 0; + opts = ost->st->pts.val * av_q2d(ost->st->time_base); + granularity = av_q2d(ost->st->time_base); + while (opts + granularity / 2.0 < opts_max) { + int count_to_write = (int)(av_get_bits_per_sample_format(ist->st->codec->sample_fmt) / 8 * ist->st->codec->channels * FFMAX(1, ist->st->codec->sample_rate * (opts_max-opts))); + while (count_to_write > 0) { + int data_size = FFMIN(AVCODEC_MAX_AUDIO_FRAME_SIZE, count_to_write); + if (samples_size < data_size) { + if (samples) + av_free(samples); + samples_size = data_size; + samples = av_malloc(samples_size); + memset(samples, 0, samples_size); + } + do_audio_out(os, ost, ist, samples, data_size); + count_to_write -= data_size; + } + opts = ost->st->pts.val * av_q2d(ost->st->time_base); + } + if (samples) + av_free(samples); + break; + } + case CODEC_TYPE_VIDEO: + opts = ost->sync_opts * av_q2d(ost->st->codec->time_base); + granularity = av_q2d(ost->st->codec->time_base); + if (opts < opts_max) { + int frame_size = 0; + do_video_out(os, ost, ist, &ist->last_picture, &frame_size); + if (vstats_filename && frame_size) + do_video_stats(os, ost, frame_size); + opts = ost->sync_opts * av_q2d(ost->st->codec->time_base); + continue_flag = opts < opts_max; + } + break; + } + } + } + if (continue_flag) { + ist->pts += input_granularity * AV_TIME_BASE; + } + } while (continue_flag); + } + } + term_exit(); /* write the trailer if needed and close file */ @@ -3794,6 +3906,7 @@ { "vglobal", HAS_ARG | OPT_INT | OPT_EXPERT, {(void*)&video_global_header}, "video global header storage type", "" }, { "copyts", OPT_BOOL | OPT_EXPERT, {(void*)&copy_ts}, "copy timestamps" }, { "shortest", OPT_BOOL | OPT_EXPERT, {(void*)&opt_shortest}, "finish encoding within shortest input" }, // + { "longest", OPT_BOOL | OPT_EXPERT, {(void*)&opt_longest}, "append shorter streams to longest duration" }, { "dts_delta_threshold", HAS_ARG | OPT_FLOAT | OPT_EXPERT, {(void*)&dts_delta_threshold}, "timestamp discontinuity delta threshold", "threshold" }, { "programid", HAS_ARG | OPT_INT | OPT_EXPERT, {(void*)&opt_programid}, "desired program number", "" }, { "xerror", OPT_BOOL, {(void*)&exit_on_error}, "exit on error", "error" }, ------------------------------------------------------------------------ --- ffmpeg.c.mplayerhq 2009-02-03 10:17:49.000000000 +0300 +++ ffmpeg.c.mplayerhq.mod 2009-02-03 10:29:12.000000000 +0300 @@ -192,6 +192,7 @@ static float audio_drift_threshold= 0.1; static int copy_ts= 0; static int opt_shortest = 0; +static int opt_longest = 0; static int video_global_header = 0; static char *vstats_filename; static FILE *vstats_file; @@ -286,6 +287,7 @@ is not defined */ int64_t pts; /* current pts */ int is_start; /* is 1 at the start and after a discontinuity */ + AVFrame last_picture; } AVInputStream; typedef struct AVInputFile { @@ -1260,6 +1262,7 @@ /* no picture yet */ goto discard_packet; } + ist->last_picture = picture; if (ist->st->codec->time_base.num != 0) { ist->next_pts += ((int64_t)AV_TIME_BASE * ist->st->codec->time_base.num) / @@ -2177,6 +2180,115 @@ } } + if (opt_longest) { + double opts_max = -1e100; + int istreams_needed[MAX_STREAMS]; + int nb_istreams_needed = 0; + for(i=0;i<nb_ostreams;i++) { + double opts; + ost = ost_table[i]; + if(ost->st->codec->codec_type == CODEC_TYPE_VIDEO) + opts = ost->sync_opts * av_q2d(ost->st->codec->time_base); + else + opts = ost->st->pts.val * av_q2d(ost->st->time_base); + if (opts > opts_max) + opts_max = opts; + } + for(i=0;i<nb_ostreams;i++) { + if ((ost_table[i]->st->codec->codec_type == CODEC_TYPE_VIDEO || ost_table[i]->st->codec->codec_type == CODEC_TYPE_AUDIO) && ost_table[i]->encoding_needed && ist_table[ost_table[i]->source_index]->decoding_needed) { + double opts, granularity; + ost = ost_table[i]; + os = output_files[ost->file_index]; + ist = ist_table[ost->source_index]; + if(ost->st->codec->codec_type == CODEC_TYPE_VIDEO) { + opts = ost->sync_opts * av_q2d(ost->st->codec->time_base); + granularity = av_q2d(ost->st->codec->time_base); + } else { + opts = ost->st->pts.val * av_q2d(ost->st->time_base); + granularity = av_q2d(ost->st->time_base); + } + if (granularity < 0) + granularity = 0; + if (opts + granularity / 2.0 < opts_max) { + int j; + for(j=0;j<nb_istreams_needed;j++) + if (ost->source_index == istreams_needed[j]) + break; + istreams_needed[j] = ost->source_index; + if (j >= nb_istreams_needed) + nb_istreams_needed = j + 1; + } + } + } + for(i=0;i<nb_istreams_needed;i++) { + int continue_flag = 0; + double input_granularity = 0; + ist = ist_table[istreams_needed[i]]; + switch (ist->st->codec->codec_type) { + case CODEC_TYPE_VIDEO: + input_granularity = av_q2d(ist->st->codec->time_base); + break; + default: + input_granularity = av_q2d(ist->st->time_base); + break; + } + do { + int j; + continue_flag = 0; + for(j=0;j<nb_ostreams;j++) { + if (ost_table[j]->encoding_needed && ost_table[j]->source_index == istreams_needed[i]) { + double opts = 0; + double granularity = 0; + ost = ost_table[j]; + os = output_files[ost->file_index]; + switch (ost->st->codec->codec_type) { + case CODEC_TYPE_AUDIO: { + uint8_t *samples = NULL; + int samples_size = 0; + opts = ost->st->pts.val * av_q2d(ost->st->time_base); + granularity = av_q2d(ost->st->time_base); + while (opts + granularity / 2.0 < opts_max) { + int count_to_write = (int)(av_get_bits_per_sample_format(ist->st->codec->sample_fmt) / 8 * ist->st->codec->channels * FFMAX(1, ist->st->codec->sample_rate * (opts_max-opts))); + while (count_to_write > 0) { + int data_size = FFMIN(AVCODEC_MAX_AUDIO_FRAME_SIZE, count_to_write); + if (samples_size < data_size) { + if (samples) + av_free(samples); + samples_size = data_size; + samples = av_malloc(samples_size); + memset(samples, 0, samples_size); + } + do_audio_out(os, ost, ist, samples, data_size); + count_to_write -= data_size; + } + opts = ost->st->pts.val * av_q2d(ost->st->time_base); + } + if (samples) + av_free(samples); + break; + } + case CODEC_TYPE_VIDEO: + opts = ost->sync_opts * av_q2d(ost->st->codec->time_base); + granularity = av_q2d(ost->st->codec->time_base); + if (opts < opts_max) { + int frame_size = 0; + do_video_out(os, ost, ist, &ist->last_picture, &frame_size); + if (vstats_filename && frame_size) + do_video_stats(os, ost, frame_size); + opts = ost->sync_opts * av_q2d(ost->st->codec->time_base); + continue_flag = opts < opts_max; + } + break; + } + } + } + if (continue_flag) { + ist->pts += input_granularity * AV_TIME_BASE; + } + } while (continue_flag); + } + } + term_exit(); /* write the trailer if needed and close file */ @@ -3794,6 +3906,7 @@ { "vglobal", HAS_ARG | OPT_INT | OPT_EXPERT, {(void*)&video_global_header}, "video global header storage type", "" }, { "copyts", OPT_BOOL | OPT_EXPERT, {(void*)&copy_ts}, "copy timestamps" }, { "shortest", OPT_BOOL | OPT_EXPERT, {(void*)&opt_shortest}, "finish encoding within shortest input" }, // + { "longest", OPT_BOOL | OPT_EXPERT, {(void*)&opt_longest}, "append shorter streams to longest duration" }, { "dts_delta_threshold", HAS_ARG | OPT_FLOAT | OPT_EXPERT, {(void*)&dts_delta_threshold}, "timestamp discontinuity delta threshold", "threshold" }, { "programid", HAS_ARG | OPT_INT | OPT_EXPERT, {(void*)&opt_programid}, "desired program number", "" }, { "xerror", OPT_BOOL, {(void*)&exit_on_error}, "exit on error", "error" },

4 7

[PATCH] libavformat/avc.c annexb parsing corrupts packets
by matteo.elli＠m-elli.it 22 Feb '10

22 Feb '10

Hi, code in libavformat/avc.c is seldom executed but it surely does when you transmux (vcodec=copy) from h264 in a ts container to h264 in an mp4/mov/mkv/avi container. The code there fails to handle annexb 00 00 01 NAL three byets prefixes properly and cuts the last byte in the previous NAL corrupting the whole frame. Here is my proposed fix. Index: libavformat/avc.c =================================================================== --- libavformat/avc.c (revision 19515) +++ libavformat/avc.c (working copy) @@ -39,15 +39,15 @@ if( (x - 0x01010101) & (~x) & 0x80808080 ) { // generic if( p[1] == 0 ) { if( p[0] == 0 && p[2] == 1 ) - return p-1; + return p; if( p[2] == 0 && p[3] == 1 ) - return p; + return p+1; } if( p[3] == 0 ) { if( p[2] == 0 && p[4] == 1 ) - return p+1; + return p+2; if( p[4] == 0 && p[5] == 1 ) - return p+2; + return p+3; } } } @@ -69,8 +69,9 @@ size = 0; nal_start = ff_avc_find_startcode(p, end); while (nal_start < end) { - while(!*(nal_start++)); + while(nal_start < end && !*(nal_start++)); if(nal_start == end) break; nal_end = ff_avc_find_startcode(nal_start, end); + while(nal_end > nal_start && !*(nal_end - 1)) nal_end--; put_be32(pb, nal_end - nal_start); put_buffer(pb, nal_start, nal_end - nal_start); size += 4 + nal_end - nal_start; ---------- Matteo Elli

4 7

[PATCH] Fix timestamp retrieval for MPEG-2 non-frame picture encoding
by bbrice＠gmail.com 22 Feb '10

22 Feb '10

Hello. I have found a bug with timestamp information being lost with MPEG-2 files that were encoded not as a frame picture. You'll essentially have 2 fields encoded one after another, with 2 PICTURE_START_CODE's occurring in a single frame. The current implementation of the MPEG video parser overwrites the PTS/DTS that was stored at the start of the frame with AV_NOPTS_VALUE when the second PICTURE_START_CODE comes around. I've seen this with I-frame only MPEG-2 files along with select DVD/VOBs. I have uploaded a sample I-frame only clip for anyone who would like to test also: http://heapify.org/dynamic/intra-only-hd.mpg -- Only about 20 frames (0.667 sec). The patch is attached. Thanks! -- Brian Brice http://heapify.org/

2 1