July 2020 - Cin - Cinelerra-GG Mailing Lists

UPD: Offtop: "Spp2Pgs - Converts general subtitles to HDMV PG streams"
by Andrew Randrianasulu 14 Jul '20

14 Jul '20

And of course I found something related to VCXPROJ->Cmake conversion minutes after I send this email out! https://pypi.org/project/cmake-converter/ written in Python3, tries to autoconvert c/c++/fortran projects ... Last version is 2.0.1 from 10 Dec, 2019 (so, not abandoned) Interesting, it may not work for big projects with GUI, but hopefully for small command line tools like this (lib)spp2pgs .... ---------- Пересланное сообщение ---------- Тема: Offtop: "Spp2Pgs - Converts general subtitles to HDMV PG streams" Дата: Вторник 14 июля 2020 Отправитель: Andrew Randrianasulu <randrianasulu(a)gmail.com> Получатель: "Cinelerra.GG" <cin(a)lists.cinelerra-gg.org>, Marton Balint <cus(a)passwd.hu>, "Cinelerra.GG" <cin(a)lists.cinelerra-gg.org> So, I was doing another pass at 'what exactly can encode those PGS streams?' and found this little project: https://github.com/subelf/Spp2Pgs --desc---- Spp2Pgs Converts general subtitles to HDMV PG streams. In other words, converts .ass files to .sup files. The project generates a command line application and a CLR dll file. Core of them is a static library, libspp2pgs. An external dll, xy-VSSppf, is introduced to deal with subtitle files. It's based on a forked version of xy-VSFilter, rendering and providing subtitles in a more easy way. Here is a breif description for all the four parts. Spp2Pgs*.exe An executable converts subtitles to .sup files. Usage: Spp2Pgs -i "X:\Saya1011con.ass" -s 1080 -r 23 "X:\Saya1011con.sup" Type Spp2Pgs -h for more help. libspp2pgs & Spp2PgsNet*.dll The core and its CLR library wrapper for .NET framework applications. Three classes play the main role: PgsEncoder, encoding and writing a pgs file. FrameStream, reading subtitles and rendering subtitles into image. Spp2Pgs, an entrance for all other things. Also reading images from FrameStream and sending them to PgsEncoder. --- It seems to be Windows-based (at least build files are *.vcxproj), and has some x86 asm (yasm) functions, but better than nothing, I guess ... This specific fork was not updated in four years (since 2016), so some subtle (!) bugs may exist .... For license it says: GPL-3.0 License Found this one via google and this forum post from 2019: http://forum.doom9.org/showthread.php?t=176899 Also found subtitler program written in C#/.Net (so it sorta run under Linux/Mono): https://nikse.dk/SubtitleEdit/Help#linux But I have no idea how (if at all) it can be compiled under Linux .. (Arch package basically repackages binary file). Found via https://unix.stackexchange.com/questions/278001/open-hdmv-pgs-subtitles-in-… Also there was KDE5 based subtitle editor with some speech recognition, but apparently it was archived by developer last year: https://github.com/maxrd2/SubtitleComposer Ah, no I'm blind , it was moved into KDE! https://invent.kde.org/multimedia/subtitlecomposer/-/commits/master So, some interesting findings, but making any of them to work (at least on my machine) will be ..non-trivial. -------------------------------------------------------

1 0

Offtop: "Spp2Pgs - Converts general subtitles to HDMV PG streams"
by Andrew Randrianasulu 14 Jul '20

14 Jul '20

So, I was doing another pass at 'what exactly can encode those PGS streams?' and found this little project: https://github.com/subelf/Spp2Pgs --desc---- Spp2Pgs Converts general subtitles to HDMV PG streams. In other words, converts .ass files to .sup files. The project generates a command line application and a CLR dll file. Core of them is a static library, libspp2pgs. An external dll, xy-VSSppf, is introduced to deal with subtitle files. It's based on a forked version of xy-VSFilter, rendering and providing subtitles in a more easy way. Here is a breif description for all the four parts. Spp2Pgs*.exe An executable converts subtitles to .sup files. Usage: Spp2Pgs -i "X:\Saya1011con.ass" -s 1080 -r 23 "X:\Saya1011con.sup" Type Spp2Pgs -h for more help. libspp2pgs & Spp2PgsNet*.dll The core and its CLR library wrapper for .NET framework applications. Three classes play the main role: PgsEncoder, encoding and writing a pgs file. FrameStream, reading subtitles and rendering subtitles into image. Spp2Pgs, an entrance for all other things. Also reading images from FrameStream and sending them to PgsEncoder. --- It seems to be Windows-based (at least build files are *.vcxproj), and has some x86 asm (yasm) functions, but better than nothing, I guess ... This specific fork was not updated in four years (since 2016), so some subtle (!) bugs may exist .... For license it says: GPL-3.0 License Found this one via google and this forum post from 2019: http://forum.doom9.org/showthread.php?t=176899 Also found subtitler program written in C#/.Net (so it sorta run under Linux/Mono): https://nikse.dk/SubtitleEdit/Help#linux But I have no idea how (if at all) it can be compiled under Linux .. (Arch package basically repackages binary file). Found via https://unix.stackexchange.com/questions/278001/open-hdmv-pgs-subtitles-in-… Also there was KDE5 based subtitle editor with some speech recognition, but apparently it was archived by developer last year: https://github.com/maxrd2/SubtitleComposer Ah, no I'm blind , it was moved into KDE! https://invent.kde.org/multimedia/subtitlecomposer/-/commits/master So, some interesting findings, but making any of them to work (at least on my machine) will be ..non-trivial.

1 0

Re: [Cin] FFmpeg 4.3 mpegts patch - questions ....
by Andrew Randrianasulu 14 Jul '20

14 Jul '20

В сообщении от Monday 13 July 2020 21:42:21 Marton Balint написал(а): > > On Mon, 13 Jul 2020, Andrew Randrianasulu wrote: > > > > > src on cgit: > > > > https://git.cinelerra-gg.org/git/?p=goodguy/cinelerra.git;a=blob;f=cinelerr… > > > > Well, I don't understand it, just worried about GROWING diff between > > original ffmpeg and Cin's version of it in this area. > > > > (added Marton as CC, as current maintainer of this area of code - sorry for my cont. ignorance, Marton!) > > This patch was a bit hard to go through even in its original form, but > some features got intergrated into ffmpeg master or my ffmpeg m2ts > branch. I'd at least consider dropping this and using the top 5 patches > from this branch: > > https://github.com/cus/ffmpeg/commits/mpegts2 Thanks for updating this patchest. Unfortunately, I don't have any hardware (BD burner, player) to play with those accurately enough (software players can be convince to play stream anyway). > > However, if somebody wants to mux compliant BluRay i'd probably say > tsMuxer is still slightly better for the job then ffmpeg, because some > codec descriptors are still not generated which should be needed for > BluRay, I don't remember which ones. At some point in the future I might try to burn (with tsMuxer's help) "AVC on DVD" type of disk just to see if some far away PS3 will really play it (but then I think I need to keep bitrate in typical DVD range - 8000 kbits/s or so ...) > > Regarding pgs muxing, there was an interesting patch series on > ffmpeg-devel which should be needed to convert between the pgs format of > mpegts and matroska, that might be of interest for you too: > > https://patchwork.ffmpeg.org/project/ffmpeg/list/?series=1163 Thanks, interesting (I wonder what kind of sw was creating those mkvs, but probably something proprietary/freeware from doom9 archives :} ) > > Regards, > Marton > > > > > > > @@ -76,11 +77,12 @@ > > MpegTSSection pat; /* MPEG-2 PAT table */ > > MpegTSSection sdt; /* MPEG-2 SDT table context */ > > MpegTSService **services; > > - int64_t sdt_period; /* SDT period in PCR time base */ > > - int64_t pat_period; /* PAT/PMT period in PCR time base */ > > + int64_t sdt_packet_timer, sdt_packet_period; > > + int64_t pat_packet_timer, pat_packet_period; > > int nb_services; > > - int64_t first_pcr; > > - int64_t next_pcr; > > + int onid; > > + int tsid; > > > > as far as I understand those (onid, tsid) used donwstream: > > > > @@ -259,7 +257,7 @@ > > put16(&q, service->sid); > > put16(&q, 0xe000 | service->pmt.pid); > > } > > - mpegts_write_section1(&ts->pat, PAT_TID, ts->transport_stream_id, ts->tables_version, 0, 0, > > + mpegts_write_section1(&ts->pat, PAT_TID, ts->tsid, ts->tables_version, 0, 0, > > > > > > was it just for shortening typing time? > > > > @@ -90,14 +92,14 @@ > > int service_type; > > > > int pmt_start_pid; > > + int pcr_start_pid; > > int start_pid; > > int m2ts_mode; > > - int m2ts_video_pid; > > - int m2ts_audio_pid; > > - int m2ts_pgssub_pid; > > - int m2ts_textsub_pid; > > + int64_t ts_offset; > > > > > > Hm, it think major point in 4.3 was exactly > > > > - Support for muxing pcm and pgs in m2ts > > > > ? (PGS are Blu-Ray subtitle graphics - no encoder in ffmpeg yet. But probably you can remux the from existing track?) > > > > > > @@ -217,14 +217,15 @@ > > /* mpegts writer */ > > > > #define DEFAULT_PROVIDER_NAME "FFmpeg" > > -#define DEFAULT_SERVICE_NAME "Service" > > +#define DEFAULT_SERVICE_NAME "Service01" > > > > > > Was it required from your Blu-Ray Disk player? May be it can be made > > conditional on new m2ts_b (say) mode? > > > > > > -/* we retransmit the SI info at this rate */ > > +/* we retransmit the SI info at this rate (ms) */ > > #define SDT_RETRANS_TIME 500 > > #define PAT_RETRANS_TIME 100 > > -#define PCR_RETRANS_TIME 20 > > +#define PCR_RETRANS_TIME 50 > > > > may be comment improvement can be made into standalone patch? > > > > @@ -281,148 +279,6 @@ > > *q_ptr = q; > > } > > > > -static int get_dvb_stream_type(AVFormatContext *s, AVStream *st) > > [..] > > -static int get_m2ts_stream_type(AVFormatContext *s, AVStream *st) > > > > here whole two functions are gone .. > > > > and probably replaced by homegrown one a bit downstream: > > > > @@ -472,8 +320,72 @@ > > err = 1; > > break; > > } > > - > > - stream_type = ts->m2ts_mode ? get_m2ts_stream_type(s, st) : get_dvb_stream_type(s, st); > > + switch (st->codecpar->codec_id) { > > > > > > any reason for this move? > > > > @@ -736,7 +648,7 @@ > > int i, running_status, free_ca_mode, val; > > > > q = data; > > - put16(&q, ts->original_network_id); > > + put16(&q, ts->onid); > > > > > > again, just shorter to type? > > > > @@ -802,49 +714,12 @@ > > return 0; > > } > > > > -static int64_t get_pcr(const MpegTSWrite *ts, AVIOContext *pb) > > > > -static void write_packet(AVFormatContext *s, const uint8_t *packet) > > > > -static void section_write_packet(MpegTSSection *s, const uint8_t *packet) > > > > all three gone .... can they stay ? :} > > (for reducing diff size) > > > > static MpegTSService *mpegts_add_service(AVFormatContext *s, int sid, > > - const AVDictionary *metadata, > > - AVProgram *program) > > + const char *provider_name, > > + const char *name) > > { > > MpegTSWrite *ts = s->priv_data; > > MpegTSService *service; > > - AVDictionaryEntry *title, *provider; > > - char default_service_name[32]; > > - const char *service_name; > > - const char *provider_name; > > - > > - title = av_dict_get(metadata, "service_name", NULL, 0); > > - if (!title) > > - title = av_dict_get(metadata, "title", NULL, 0); > > - snprintf(default_service_name, sizeof(default_service_name), "%s%02d", DEFAULT_SERVICE_NAME, ts->nb_services + 1); > > - service_name = title ? title->value : default_service_name; > > - provider = av_dict_get(metadata, "service_provider", NULL, 0); > > - provider_name = provider ? provider->value : DEFAULT_PROVIDER_NAME; > > > > > > I think this one allowed more data passing from ...upper libavcodec layer? > > > > -static void enable_pcr_generation_for_stream(AVFormatContext *s, AVStream *pcr_st) > > +static void mpegts_prefix_m2ts_header(AVFormatContext *s) > > { > > MpegTSWrite *ts = s->priv_data; > > - MpegTSWriteStream *ts_st = pcr_st->priv_data; > > - > > - if (ts->mux_rate > 1 || ts->pcr_period_ms >= 0) { > > - int pcr_period_ms = ts->pcr_period_ms == -1 ? PCR_RETRANS_TIME : ts->pcr_period_ms; > > - ts_st->pcr_period = av_rescale(pcr_period_ms, PCR_TIME_BASE, 1000); > > - } else { > > - /* By default, for VBR we select the highest multiple of frame duration which is less than 100 ms. */ > > - int64_t frame_period = 0; > > - if (pcr_st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) { > > - int frame_size = av_get_audio_frame_duration2(pcr_st->codecpar, 0); > > - if (!frame_size) { > > - av_log(s, AV_LOG_WARNING, "frame size not set\n"); > > - frame_size = 512; > > - } > > - frame_period = av_rescale_rnd(frame_size, PCR_TIME_BASE, pcr_st->codecpar->sample_rate, AV_ROUND_UP); > > - } else if (pcr_st->avg_frame_rate.num) { > > - frame_period = av_rescale_rnd(pcr_st->avg_frame_rate.den, PCR_TIME_BASE, pcr_st->avg_frame_rate.num, AV_ROUND_UP); > > - } > > - if (frame_period > 0 && frame_period <= PCR_TIME_BASE / 10) > > - ts_st->pcr_period = frame_period * (PCR_TIME_BASE / 10 / frame_period); > > - else > > - ts_st->pcr_period = 1; > > + if (ts->m2ts_mode) { > > + uint32_t tp_extra_header = ts->pcr % 0x3fffffff; > > + tp_extra_header = AV_RB32(&tp_extra_header); > > + avio_write(s->pb, (unsigned char *) &tp_extra_header, > > + sizeof(tp_extra_header)); > > } > > - > > - // output a PCR as soon as possible > > - ts_st->last_pcr = ts->first_pcr - ts_st->pcr_period; > > } > > > > May be just leave old function the same, just add short new one > > (static void mpegts_prefix_m2ts_header(AVFormatContext *s)) ? > > > > same for > > > > -static void select_pcr_streams(AVFormatContext *s) > > +static void section_write_packet(MpegTSSection *s, const uint8_t *packet) > > > > static int mpegts_init(AVFormatContext *s) > > > > ah, so av_dict_get and co moved here .... > > > > - /* MPEG pid values < 16 are reserved. Applications which set st->id in > > - * this range are assigned a calculated pid. */ > > - if (st->id < 16) { > > - if (ts->m2ts_mode) { > > - switch (st->codecpar->codec_type) { > > - case AVMEDIA_TYPE_VIDEO: > > - ts_st->pid = ts->m2ts_video_pid++; > > + program = av_find_program_from_stream(s, NULL, i); > > + if (program) { > > + for (j = 0; j < ts->nb_services; j++) { > > + if (ts->services[j]->program == program) { > > + service = ts->services[j]; > > break; > > - case AVMEDIA_TYPE_AUDIO: > > - ts_st->pid = ts->m2ts_audio_pid++; > > - break; > > - case AVMEDIA_TYPE_SUBTITLE: > > - switch (st->codecpar->codec_id) { > > - case AV_CODEC_ID_HDMV_PGS_SUBTITLE: > > - ts_st->pid = ts->m2ts_pgssub_pid++; > > - break; > > - case AV_CODEC_ID_HDMV_TEXT_SUBTITLE: > > - ts_st->pid = ts->m2ts_textsub_pid++; > > - break; > > - } > > - break; > > - } > > - if (ts->m2ts_video_pid > M2TS_VIDEO_PID + 1 || > > - ts->m2ts_audio_pid > M2TS_AUDIO_START_PID + 32 || > > - ts->m2ts_pgssub_pid > M2TS_PGSSUB_START_PID + 32 || > > - ts->m2ts_textsub_pid > M2TS_TEXTSUB_PID + 1 || > > - ts_st->pid < 16) { > > - av_log(s, AV_LOG_ERROR, "Cannot automatically assign PID for stream %d\n", st->index); > > - return AVERROR(EINVAL); > > } > > - } else { > > - ts_st->pid = ts->start_pid + i; > > } > > - } else { > > - ts_st->pid = st->id; > > } > > > > Oh, here our subtitle support go ....... > > > > a lot of skipped, I'm very unsure about all those changes ... > > > > > > @@ -1724,7 +1585,7 @@ > > > > ret = avio_open_dyn_buf(&ts_st->amux->pb); > > if (ret < 0) > > - return ret; > > + return AVERROR(ENOMEM); > > > > seems like possible standalone patch for ffmpeg upstream inclusion? > > > > @@ -1755,7 +1616,7 @@ > > } while (p < buf_end && (state & 0x7e) != 2*35 && > > (state & 0x7e) >= 2*32); > > > > - if ((state & 0x7e) < 2*16 || (state & 0x7e) >= 2*24) > > + if ((state & 0x7e) < 2*16 && (state & 0x7e) >= 2*24) > > > > ???? why it was changed? > > > > > > > > IIRC those changes come from fact ffmpeg's > > mpegts muxer mostly geared towards transmission vs standalone BD creation .... > > > > But at least potential for BD text subtitles look interestinge nough > > for keeping looking into this area ...I think. > > > > Also, ffmpeg 4.3.1 was already released, with small fixes: > > > > https://ffmpeg.org/download.html > > > > FFmpeg 4.3.1 "4:3" > > 4.3.1 was released on 2020-07-11. It is the latest stable FFmpeg release from the 4.3 release branch, > > which was cut from master on 2020-06-08. > > > > > > It was FAST ...... > > > > Also, there is one more patch on ML: > > http://ffmpeg.org/pipermail/ffmpeg-devel/2020-July/266181.html > > > > [FFmpeg-devel] [PATCH] x86/yuv2rgb: fix crashes when storing data on unaligned buffers > > > > Regression since fc6a5883d6af8cae0e96af84dda0ad74b360a084 on SSSE3 enabled > > CPUs. > > > > Fixes ticket #8747 > > > > ----- > > > > END! > > >

1 0

FFmpeg 4.3 mpegts patch - questions ....
by Andrew Randrianasulu 13 Jul '20

13 Jul '20

src on cgit: https://git.cinelerra-gg.org/git/?p=goodguy/cinelerra.git;a=blob;f=cinelerr… Well, I don't understand it, just worried about GROWING diff between original ffmpeg and Cin's version of it in this area. (added Marton as CC, as current maintainer of this area of code - sorry for my cont. ignorance, Marton!) @@ -76,11 +77,12 @@ MpegTSSection pat; /* MPEG-2 PAT table */ MpegTSSection sdt; /* MPEG-2 SDT table context */ MpegTSService **services; - int64_t sdt_period; /* SDT period in PCR time base */ - int64_t pat_period; /* PAT/PMT period in PCR time base */ + int64_t sdt_packet_timer, sdt_packet_period; + int64_t pat_packet_timer, pat_packet_period; int nb_services; - int64_t first_pcr; - int64_t next_pcr; + int onid; + int tsid; as far as I understand those (onid, tsid) used donwstream: @@ -259,7 +257,7 @@ put16(&q, service->sid); put16(&q, 0xe000 | service->pmt.pid); } - mpegts_write_section1(&ts->pat, PAT_TID, ts->transport_stream_id, ts->tables_version, 0, 0, + mpegts_write_section1(&ts->pat, PAT_TID, ts->tsid, ts->tables_version, 0, 0, was it just for shortening typing time? @@ -90,14 +92,14 @@ int service_type; int pmt_start_pid; + int pcr_start_pid; int start_pid; int m2ts_mode; - int m2ts_video_pid; - int m2ts_audio_pid; - int m2ts_pgssub_pid; - int m2ts_textsub_pid; + int64_t ts_offset; Hm, it think major point in 4.3 was exactly - Support for muxing pcm and pgs in m2ts ? (PGS are Blu-Ray subtitle graphics - no encoder in ffmpeg yet. But probably you can remux the from existing track?) @@ -217,14 +217,15 @@ /* mpegts writer */ #define DEFAULT_PROVIDER_NAME "FFmpeg" -#define DEFAULT_SERVICE_NAME "Service" +#define DEFAULT_SERVICE_NAME "Service01" Was it required from your Blu-Ray Disk player? May be it can be made conditional on new m2ts_b (say) mode? -/* we retransmit the SI info at this rate */ +/* we retransmit the SI info at this rate (ms) */ #define SDT_RETRANS_TIME 500 #define PAT_RETRANS_TIME 100 -#define PCR_RETRANS_TIME 20 +#define PCR_RETRANS_TIME 50 may be comment improvement can be made into standalone patch? @@ -281,148 +279,6 @@ *q_ptr = q; } -static int get_dvb_stream_type(AVFormatContext *s, AVStream *st) [..] -static int get_m2ts_stream_type(AVFormatContext *s, AVStream *st) here whole two functions are gone .. and probably replaced by homegrown one a bit downstream: @@ -472,8 +320,72 @@ err = 1; break; } - - stream_type = ts->m2ts_mode ? get_m2ts_stream_type(s, st) : get_dvb_stream_type(s, st); + switch (st->codecpar->codec_id) { any reason for this move? @@ -736,7 +648,7 @@ int i, running_status, free_ca_mode, val; q = data; - put16(&q, ts->original_network_id); + put16(&q, ts->onid); again, just shorter to type? @@ -802,49 +714,12 @@ return 0; } -static int64_t get_pcr(const MpegTSWrite *ts, AVIOContext *pb) -static void write_packet(AVFormatContext *s, const uint8_t *packet) -static void section_write_packet(MpegTSSection *s, const uint8_t *packet) all three gone .... can they stay ? :} (for reducing diff size) static MpegTSService *mpegts_add_service(AVFormatContext *s, int sid, - const AVDictionary *metadata, - AVProgram *program) + const char *provider_name, + const char *name) { MpegTSWrite *ts = s->priv_data; MpegTSService *service; - AVDictionaryEntry *title, *provider; - char default_service_name[32]; - const char *service_name; - const char *provider_name; - - title = av_dict_get(metadata, "service_name", NULL, 0); - if (!title) - title = av_dict_get(metadata, "title", NULL, 0); - snprintf(default_service_name, sizeof(default_service_name), "%s%02d", DEFAULT_SERVICE_NAME, ts->nb_services + 1); - service_name = title ? title->value : default_service_name; - provider = av_dict_get(metadata, "service_provider", NULL, 0); - provider_name = provider ? provider->value : DEFAULT_PROVIDER_NAME; I think this one allowed more data passing from ...upper libavcodec layer? -static void enable_pcr_generation_for_stream(AVFormatContext *s, AVStream *pcr_st) +static void mpegts_prefix_m2ts_header(AVFormatContext *s) { MpegTSWrite *ts = s->priv_data; - MpegTSWriteStream *ts_st = pcr_st->priv_data; - - if (ts->mux_rate > 1 || ts->pcr_period_ms >= 0) { - int pcr_period_ms = ts->pcr_period_ms == -1 ? PCR_RETRANS_TIME : ts->pcr_period_ms; - ts_st->pcr_period = av_rescale(pcr_period_ms, PCR_TIME_BASE, 1000); - } else { - /* By default, for VBR we select the highest multiple of frame duration which is less than 100 ms. */ - int64_t frame_period = 0; - if (pcr_st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) { - int frame_size = av_get_audio_frame_duration2(pcr_st->codecpar, 0); - if (!frame_size) { - av_log(s, AV_LOG_WARNING, "frame size not set\n"); - frame_size = 512; - } - frame_period = av_rescale_rnd(frame_size, PCR_TIME_BASE, pcr_st->codecpar->sample_rate, AV_ROUND_UP); - } else if (pcr_st->avg_frame_rate.num) { - frame_period = av_rescale_rnd(pcr_st->avg_frame_rate.den, PCR_TIME_BASE, pcr_st->avg_frame_rate.num, AV_ROUND_UP); - } - if (frame_period > 0 && frame_period <= PCR_TIME_BASE / 10) - ts_st->pcr_period = frame_period * (PCR_TIME_BASE / 10 / frame_period); - else - ts_st->pcr_period = 1; + if (ts->m2ts_mode) { + uint32_t tp_extra_header = ts->pcr % 0x3fffffff; + tp_extra_header = AV_RB32(&tp_extra_header); + avio_write(s->pb, (unsigned char *) &tp_extra_header, + sizeof(tp_extra_header)); } - - // output a PCR as soon as possible - ts_st->last_pcr = ts->first_pcr - ts_st->pcr_period; } May be just leave old function the same, just add short new one (static void mpegts_prefix_m2ts_header(AVFormatContext *s)) ? same for -static void select_pcr_streams(AVFormatContext *s) +static void section_write_packet(MpegTSSection *s, const uint8_t *packet) static int mpegts_init(AVFormatContext *s) ah, so av_dict_get and co moved here .... - /* MPEG pid values < 16 are reserved. Applications which set st->id in - * this range are assigned a calculated pid. */ - if (st->id < 16) { - if (ts->m2ts_mode) { - switch (st->codecpar->codec_type) { - case AVMEDIA_TYPE_VIDEO: - ts_st->pid = ts->m2ts_video_pid++; + program = av_find_program_from_stream(s, NULL, i); + if (program) { + for (j = 0; j < ts->nb_services; j++) { + if (ts->services[j]->program == program) { + service = ts->services[j]; break; - case AVMEDIA_TYPE_AUDIO: - ts_st->pid = ts->m2ts_audio_pid++; - break; - case AVMEDIA_TYPE_SUBTITLE: - switch (st->codecpar->codec_id) { - case AV_CODEC_ID_HDMV_PGS_SUBTITLE: - ts_st->pid = ts->m2ts_pgssub_pid++; - break; - case AV_CODEC_ID_HDMV_TEXT_SUBTITLE: - ts_st->pid = ts->m2ts_textsub_pid++; - break; - } - break; - } - if (ts->m2ts_video_pid > M2TS_VIDEO_PID + 1 || - ts->m2ts_audio_pid > M2TS_AUDIO_START_PID + 32 || - ts->m2ts_pgssub_pid > M2TS_PGSSUB_START_PID + 32 || - ts->m2ts_textsub_pid > M2TS_TEXTSUB_PID + 1 || - ts_st->pid < 16) { - av_log(s, AV_LOG_ERROR, "Cannot automatically assign PID for stream %d\n", st->index); - return AVERROR(EINVAL); } - } else { - ts_st->pid = ts->start_pid + i; } - } else { - ts_st->pid = st->id; } Oh, here our subtitle support go ....... a lot of skipped, I'm very unsure about all those changes ... @@ -1724,7 +1585,7 @@ ret = avio_open_dyn_buf(&ts_st->amux->pb); if (ret < 0) - return ret; + return AVERROR(ENOMEM); seems like possible standalone patch for ffmpeg upstream inclusion? @@ -1755,7 +1616,7 @@ } while (p < buf_end && (state & 0x7e) != 2*35 && (state & 0x7e) >= 2*32); - if ((state & 0x7e) < 2*16 || (state & 0x7e) >= 2*24) + if ((state & 0x7e) < 2*16 && (state & 0x7e) >= 2*24) ???? why it was changed? IIRC those changes come from fact ffmpeg's mpegts muxer mostly geared towards transmission vs standalone BD creation .... But at least potential for BD text subtitles look interestinge nough for keeping looking into this area ...I think. Also, ffmpeg 4.3.1 was already released, with small fixes: https://ffmpeg.org/download.html FFmpeg 4.3.1 "4:3" 4.3.1 was released on 2020-07-11. It is the latest stable FFmpeg release from the 4.3 release branch, which was cut from master on 2020-06-08. It was FAST ...... Also, there is one more patch on ML: http://ffmpeg.org/pipermail/ffmpeg-devel/2020-July/266181.html [FFmpeg-devel] [PATCH] x86/yuv2rgb: fix crashes when storing data on unaligned buffers Regression since fc6a5883d6af8cae0e96af84dda0ad74b360a084 on SSSE3 enabled CPUs. Fixes ticket #8747 ----- END!

3 4

Slightly updated russian translation
by Andrew Randrianasulu 13 Jul '20

13 Jul '20

I asked my russian-speaking friend to check it, but this may take some time... Please *do not merge* this version yet - new bugs may surface as we try to look at messages Attached modified .po file and compiled .mo file too. I used few first steps from HOW-TO.TXT 1) cd <path>/cinelerra-x.x.x 2) run xgettext: ./po/xlat.sh > po/cin.po then I moved cin.po and ru.po from /dev/shm/tmp/cinelerra-goodguy-20200711/cinelerra-5.1/po into my working directory /home/guest/Cin-translations after this I used msgmerge --output-file=ru-3.po ru-2.po ru.po msgmerge --output-file=ru-4.po ru-3.po cin.po msgfmt -c -o ru.mo.new-4 ru-4.po So many steps because I wanted to have backups in case I mismerge something ...

2 1

ffmpeg upgraded from 4.1 to 4.3 + Thirdparty libraries
by Phyllis Smith 11 Jul '20

11 Jul '20

CinelerraGG git checkin includes the following upgrades and a couple of fixes: 1) fixed ffmpeg scan remap 2) BT 473 configure.ac patch (noted by ferdnyc with patch) 3) thirdparty libraries updated to include: ffmpeg from 4.1 to 4.3 turbo-jpeg vorbis x265 vpx lv2 aom dav1d nv-codec-headers No thorough test has been done so just using it helps to make sure there are no problems. Thank you, GG/Phyllis

2 3

Mailing list & Gmail
by Sam 11 Jul '20

11 Jul '20

Hi Cinelerra friends, As some of you have already noticed, modifications to Mailman were necessary to get Gmail users back. To avoid spam, the major email providers use methods like SPF, DKIM and DMARC. We have adapted our mailing list so that the subject line and footer are no longer changed, i.e. the abbreviation [Cin] and the old familiar footer are missing. Thus DKIM is not broken and we are not classified as spam. We have also made other changes to meet modern requirements. Furthermore, only the mailing list address appears as the sender. We will gain more experience with the new procedures and if necessary we will make further optimizations. If you have any problems or other questions, please contact me. Sam

3 5

AOMedia AV1 2.0
by Andrew Randrianasulu 10 Jul '20

10 Jul '20

Hi all! Just was refreshing Phoronix page and this scrolls into view: https://www.phoronix.com/scan.php?page=news_item&px=AOMedia-libaom-2.0-Rele… --- AOMedia AV1 2.0 Codec Library Released With Many Improvements Written by Michael Larabel in Multimedia on 19 May 2020 at 06:53 AM EDT. 2 Comments MULTIMEDIA -- Version 2.0 of the libaom AOMedia AV1 video encoder / video codec SDK library is now available as the first major update in nearly two years. Libaom 2.0 is the first release since the original 1.0 release back in mid-2018 after the AOMedia codec working group approved the 1.0 release. The developers view this AOMedia AV1 2.0 release as now being their "first official release" for production. Libaom 2.0 adds a real-time encoding mode, SVC support, drops multi-resolution encoding, documentation updates, and other improvements to this open-source AV1 reference encoder. -- Hm, beaware about possible buildsystem changes, I guess? I have early snapshot running, from around Mar, 19 2020 but I have recent Cmake and stuff ....

2 5

Nest EDL
by Rafa Mar Multimedia en Gnu\Linux 10 Jul '20

10 Jul '20

Today I have been looking at nesting EDL, it seems like a wonderful function, but I have found a problem. Let me explain: When I open an .xml project in nested EDL mode, it loads in resources, very well, but if I open this EDL from there to make any changes, all the files of this project are loaded, which is very normal to occur this, so far so good, but when I close the EDL to return to the project, all the files get loaded, and this has puzzled me. I think the correct thing would be that when closing the EDL, only the files of this new project were left in the resources window. Or am I doing something wrong again? I have made a video so that what I quote is better understood. https://streamable.com/nn0lmf

3 4

Just a mailing list test 3
by Sam 09 Jul '20

09 Jul '20

You can ignore this email, we are doing some tests on Mailman and Gmail. Sam

2 1