August 2009 - ffmpeg-devel - lists.ffmpeg.org

[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] 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

cavs.c DECLARE_ALIGNED in ff_cavs_filter
by Reimar.Doeffinger＠gmx.de 15 Feb '10

15 Feb '10

Hello, currently there is this code in cavs.c: DECLARE_ALIGNED_8(uint8_t, bs[8]); The reason for it is that kind of code: > *((uint64_t *)bs) = 0x0202020202020202ULL; > if( *((uint64_t *)bs) ) { I thought of "solving" this via a union, but is this code really this speed critical that we need such a mess? Can't we just use memset for the writes and e.g. AV_RN64 for the read?

4 6

[PATCH] Document url_fopen()
by stefano.sabatini-lala＠poste.it 09 Feb '10

09 Feb '10

Hi, as in subject. -- FFmpeg = Frenzy Fundamentalist Mournful Peaceless Elitist Guru

1 4

[PATCH] Document url_fdopen()
by stefano.sabatini-lala＠poste.it 05 Feb '10

05 Feb '10

As in subject. -- FFmpeg = Fast and Fundamentalist Minimalistic Philosophical Elastic Ghost

3 12

[PATCH] GSM-MS decoder and encoder
by mbardiaux＠mediaxim.be 01 Feb '10

01 Feb '10

-- Michel Bardiaux R&D Director T +32 [0] 2 790 29 41 F +32 [0] 2 790 29 02 E mailto:mbardiaux at mediaxim.be Mediaxim NV/SA Vorstlaan 191 Boulevard du Souverain Brussel 1160 Bruxelles http://www.mediaxim.com/

6 56