Ffmpeg-devel-irc

Download

ffmpeg-devel-irc@ffmpeg.org

September 2014

1 participants
60 discussions

ffmpeg-devel.log.20140924
by burek 25 Sep '14

25 Sep '14

[00:44] <cone-78> ffmpeg.git 03Michael Niedermayer 07master:00d3bb1a074b: tests/fate-run.sh: Cat .err file in case of error with V>0 [00:48] <cone-78> ffmpeg.git 03James Almer 07master:789274dca8e5: avcodec/libvpxdec: don't check for formats other than i420 when vp9 decoding is disabled [00:58] <cone-78> ffmpeg.git 03James Almer 07master:164d6c7f5b0e: x86/videodsp: fix warning about discarded 'const' qualifier [01:45] <cone-78> ffmpeg.git 03Christophe Gisquet 07master:de60ce391d38: hevc: remove unneeded buffer. [01:45] <cone-78> ffmpeg.git 03Christophe Gisquet 07master:3fe962940f1e: hevc: move intermediate bidir buffer [02:43] <cone-78> ffmpeg.git 03Michael Niedermayer 07master:a7ed01082f25: configure: support --disable-sdl [02:43] <cone-78> ffmpeg.git 03Michael Niedermayer 07master:fb01a1846244: configure: reindent the --disable-sdl related code [02:43] <cone-78> ffmpeg.git 03Michael Niedermayer 07master:3bc036171f2e: configure: fail if SDL is unavailable and --enable-sdl is specified [05:41] <cone-78> ffmpeg.git 03Michael Niedermayer 07master:170e2fd7c39c: configure: Check if "-D_POSIX_C_SOURCE=200112 -D_XOPEN_SOURCE=600" is needed for localtime_r() [09:17] <ubitux> "I still use gopher now and again, though not for multimedia. It's quick, simple and mostly unmonitored by the powers that be." [09:17] <ubitux> hehe [11:21] <ubitux> is it that much work to add the -skip_* functions to h264? [11:39] <pa> hello! [11:42] <alpa_astero> hi [11:43] <ubitux> heh, -skip_frame all actually skip them really completely :( [11:43] <ubitux> i wonder if i could still export the mvs with that [11:49] <sphoorti> Hello folks! I wish to apply for the OPW program. I am interested in the symmetric block cyphers and basic servers for network protocols projects. [11:52] <ubitux> michaelni: you can now add sphoorti i guess :p [11:54] <sphoorti> the mentors mentioned for the said projects do not seem to be on the channel at the moment I guess [11:54] <sphoorti> What could be the initial tasks which I could try ? [11:54] <sphoorti> ubitux: could you give me some pointers ? [11:54] <ubitux> indeed i don't see Holden, but michaelni seems to be here as a backup :p [11:54] <ubitux> mmh [11:55] <ubitux> "Qualification Task: Anything related to cryptography, like proving you can implement a cryptographic hash function or a HMAC function in C. The applicant can also propose her own qualification task." [11:56] <ubitux> can you tell us more about your knowledge on C, git & crypto? [11:56] <ubitux> i'm not familiar with the opw process though [11:57] <jul13t> michaelni, ubitux, i'm interested in working on the MPEG-4Audio lossles coding decoder [11:58] <jul13t> i'll like to know if it's still available [11:58] <sphoorti> I have used C, git and crypto as a part of my coursework [11:59] <ubitux> i haven't been briefed about what you are supposed to do yet [11:59] <michaelni> sphoorti, about "Basic servers for network protocols", best ask reynaldo (he is here in the channel) [12:01] <sphoorti> hey reynaldo ! [12:01] <michaelni> sphoorti, jul13t if you are interrested in some project / are working on a qualification task, please add yourself to https://trac.ffmpeg.org/wiki/SponsoringPrograms/OPW/2014-12 so that others know [12:01] <michaelni> the wiki can e edited by anyone with a valid email address [12:01] <jul13t> thanks very much [12:02] Action: jul13t gets right on it [12:03] <sphoorti> thanks michaelni :) [12:32] <jul13t> michaelni, need some assistance with editing the wiki [12:33] <jul13t> tried registering but i get this error thinking it's a spam [12:47] Action: michaelni sees "jul13t has quit (Read error: Connection reset by peer)", now how do i help with registering :/ [12:49] <ubitux> yeah, peer strikes again [12:49] <ubitux> always him [12:50] <alpa_astero> michaelni: hopefully she'll come back soon :/ [12:52] <michaelni> yes, btw, in case she comes back when iam afk, her IP is on several dnsbl lists, apprently it was used by some spammer months ago. i marked her registering attempt as ham so simply retrying to register might work, if not, i can add her manually or she could retry from a different IP if thats easy [12:53] <alpa_astero> Ok, cool [13:01] <Compn> i am convinced that windows has a fundamental flaw in its code, that flaw starts as soon as you change the time on your computer :\ its all down hill after you change that clock. system files misbehave, ger corrupted, etc. [13:03] <rcombs> that or Windows is just bad [13:51] <cone-443> ffmpeg.git 03Michael Niedermayer 07master:266b3d4fe48a: ffplay: use av_codec_get_pkt_timebase() [13:59] <pa> is it necessary to install mingw if i want to compile ffmpeg with visual studio 2013? [14:01] <pa> alternately, whats the recommended way to only open ffmpeg with MSVC (no compilation, only opening/editing) [14:01] <thardin> just import it? [14:02] <pa> hm.. like create new solution, and import from existing dir? [14:03] <thardin> should work [14:11] <J_Darnley> You need to run configure if you ever want to get ffmpeg to compile. [14:17] <pa> for now what i would like is understand how to add a command line switch, to prevent ffmpeg to write down the header and trailer of the file [14:17] <pa> but i got lost trying to figure out how these options are handled, so i wanted to use msvc that lets me jump to definitions etc [14:22] <pa> also, i dont fully get ffmpeg_parse_option: it returns an int, it doesn't get a context as parameter, and it does not modify global variables.. how does it communicate the output of the parsing? [14:41] <J_Darnley> You shouldn't need to understand how the command line parsing works [14:41] <J_Darnley> Add an option to the options options array and the relevent struct [14:41] <J_Darnley> ... [14:42] <J_Darnley> Add an option to the options array and the relevant struct [15:33] <ubitux> heh thilo isn't aware of opw? [15:35] <ubitux> btw, unrelated, do the mvs in h264 need the entropy decoding? (i suppose it does but just in case...) [15:42] <ubitux> someone should fix the picture links on that opw, that's not really serious [15:43] <mansi> pross, as a part of opw i'd like to work on apng demuxing. is the pre requisite task already taken up ? [15:57] <kierank> where did the opw money appear from? [15:57] <Ko_deZ> Hi. I am playing around with ffmpeg to see if I can make crude video analyzer. I have managed to get both average QP and standard deviation printout by extending some debug output code in h264.c. My only other wish was to print DTS together with this QP info, but I have been uable to figure out how to get that info. Is it awailable trough the H264Context struct somehow? [16:00] <ubitux> Ko_deZ: if you want to access QP, you can do it from filters, and then do something here [16:00] <ubitux> typically vf_spp and vf_pp fetch them (iirc) [16:01] <ubitux> another example is codeview which use the exported motion vectors [16:01] <ubitux> if you need to do analysis, you need to export the information in a similar way the motion vectors do [16:01] <ubitux> codecview* [16:07] <mansi> michaelni, as a part of opw i'd like to work on apng demuxing. is the pre requisite task already taken up ? [16:14] <michaelni> mansi, we have some limited format detection code for image2 / imagepipe stuff, maybe implementing APNG/PNG probing and a basic APNG demuxer could be a qualification task but best talk with pross and paul [16:16] <michaelni> mansi, the idea is that the qualificatio task would be part of the whole work, also if you start working on it then please add yourself to the wiki so others know and duplicate work / tasks are avoided [16:17] <wm4> so I've posted my .sup/pgs demuxer patch 4 days ago, and there were no further replies... could it be pushed? [16:19] <ubitux> wm4:i haven't followed the review, but can you add a ref in general.texi and bump minor? [16:19] <wm4> hm ok [16:19] <ubitux> (see around line 1015) [16:20] <ubitux> you might want to add it to the Changelog (and we need to reset RELEASE_NOTES but don't worry about that one) [16:24] <mansi> they don't seem to be around here, i will send a mail on the developers list .and i edited the wiki as you suggested. thanks michaelni [16:25] <michaelni> mansi, ok [16:33] <ubitux> why can't you link the images from another page? [16:39] <pa> J_Darnley, do you think this could be sufficient? http://paste.ubuntu.com/8418536/ (modified the second if condition). I'm asking because, in all honesty, it's never too clear what are the side effects of a function call in ffmpeg. It seems that all the functions do several things [16:43] <ubitux> pa: some muxers will break really badly if you don't call the header [16:43] <ubitux> same for the footer [16:43] <pa> i know, that's ok [16:44] <pa> i just want to make sure that all i'm skipping is the header, and also that i'm skipping it completely [16:44] <ubitux> add it as a private option to muxers that support it [16:44] <pa> ubitux, "call the header"? [16:44] <pa> that's actually what i was wondering [16:44] <ubitux> call the header callback [16:44] <pa> what i want to do is skipping header writeout [16:44] <nevcairiel> that seems like a terrible hack, hope you're not proposing an actual patch :p [16:45] <pa> no worries, no patch :) [16:45] <ubitux> his point was that we can't understand the paste if you're not posting a patch [16:45] <pa> ubitux, so i want to make sure that the initialization is completed [16:45] <ubitux> pass it a dummy AVIO context [16:45] <pa> ah [16:45] <ubitux> but that's evil [16:45] <pa> well i have a command line switch [16:45] <pa> that sets an int [16:46] <pa> and i can use it where the header is written to disk [16:46] <pa> but i don't know exactly where this is done [16:46] <wm4> pa: are you still pursuing into coercing matroska into being a streaming protocol? [16:46] <pa> wm4, yep , amost there :) [16:47] <pa> but i need a modified ffmpeg to produce the kind of stream i need [16:47] <pa> problem is that it looks like concerns are not really separated, in ffmpeg, so i get lost [16:49] <pa> ubitux, so you suggest to rather modify the write_header function in the actual container? [16:49] <ubitux> yes [16:50] <nevcairiel> you never know what it does in there, its not necessarily write_header, but commonly also initialize [16:51] <pa> hm.. but then where is the header actually written to disk? [16:52] <nevcairiel> there is no central place for that, every muxer decides when it needs to do that [16:53] <nevcairiel> some muxers can only write the header at the end, so they may reserve some space in the beginning and write it later [17:11] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:ac682955e90a: postproc: add basic deblock filter visualization support [17:11] <ubitux> michaelni: typo, one 's' to visualization [17:13] <michaelni> just in the commit message ? ... i cant fix it there ... [17:14] <ubitux> this one is correct, see doc/APIchanges [17:24] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:3cd252d29a15: doc/APIchanges: fix typo [17:24] <ubitux> thx [17:24] <ubitux> :) [17:31] <cone-600> ffmpeg.git 03Vittorio Giovara 07master:c463dfc7e499: rtpdec_hevc: Drop a duplicated, nonstandard entry [17:31] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:e27f781aeb74: Merge commit 'c463dfc7e49929a9891884312b23b27d14729c51' [17:36] <wm4> ubitux: bump minor or micro? [17:36] <ubitux> minor [17:39] <wm4> resent [17:40] <ubitux> huh, we still don't have the muxer? [17:41] <wm4> no [17:41] <wm4> unless I missed something [17:42] <wm4> the guy didn't resend it yet [17:43] <ubitux> sad [17:44] <ubitux> btw, not blocking, but the pgs sub decoder has no coverage [17:44] <ubitux> we could add a .sub to fate to cover both demuxer & decoder [18:16] <pa> jesus, it's a bloody mess.. i think i'll cut the header by hand rather. modifying the thousands write calls all over the matroska muxer will take me forever [18:42] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:d6ed67780bdb: libavformat/avidec: Make unchanged function arguments const [18:42] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:af19d2ed1efb: libavformat/avienc: Make unchanged function arguments const [19:16] <ubitux> pa: i told you, set a dummy avio context [19:16] <ubitux> and write to it [19:24] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:ab6f4fd4a70a: libavformat/dv: Make unchanged function arguments const [19:24] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:280ef183db55: libavformat/ffmetadec: Make unchanged function arguments const [19:29] <localhost_> Hi, i'll like to know the number of slots for OPW this year [19:33] <wm4> El3ra: probably, also see https://trac.ffmpeg.org/wiki/SponsoringPrograms/OPW/2014-12 [19:33] <El3ra> wm4, thanks, I heard some orgs are given more slots than others, so just looking at the chances [20:45] <aparna014> Hi. [20:45] <aparna014> Im new here [20:46] <aparna014> I want to contribute to ffmpeg. From where should I start ?? [20:47] <JEEBcz> 1) grab the current source code from git 2) compile ffmpeg (you will need a compiler, yasm and some kind of shell; ./configure and then make) 3) look at what interests you or see if someone can give you an example of something that could be worked on [20:49] <wm4> contributions are done mostly by sending patches to the ffmpeg-devel mailing list [21:12] <cone-600> ffmpeg.git 03James Almer 07master:70277d1d234b: x86/videodsp: add ff_emu_edge_{hfix,hvar}_avx2 [21:14] <aparna014> Thank you [23:13] <cone-600> ffmpeg.git 03Philip DeCamp 07master:857fc0a71f1b: libavutil/opt: fix av_opt_set_channel_layout() to access correct memory address [23:51] <cone-600> ffmpeg.git 03James Almer 07master:ad26e83f9cb8: avcodec/x86: use function pointers for {put,add}_pixels_clamped [00:00] --- Thu Sep 25 2014

1 0

ffmpeg.log.20140924
by burek 25 Sep '14

25 Sep '14

[01:08] <UWarbs> I would like to use ffmpeg with my rails project. I noticed the ruby gem, ffmpeg, was last updated in 2009. What is the best way to do a rails wrapper? Specifically I want to get user audio, cut the leading silence, then save the files to a databse. Cutting the leading silence is my only hold up. [01:57] Last message repeated 1 time(s). [02:05] <iive> UWarbs: http://digitalcardboard.com/blog/2009/08/25/the-sox-of-silence/ ? [02:07] <UWarbs> That is essentially what I am looking for, I do not know how to write code in a rails app that can use the command line like that though. [02:07] <UWarbs> Thank you [04:44] <MikeJoel> how would you set the screen size of the video? I am trying to convert a swf to mp4 - it just is showing the upper left corner area (really zoomed in) [04:56] <Nosomy|off> MikeJoel, whats is yours cmdlines? [04:59] <MikeJoel> I was just trying the basic -i input.swf output.mp4 [05:04] <MikeJoel> sacarasc: that was it. The output works. it creates a mp4 but the upper left area of the swf fills the entire mp4 video so nothing else is seen. [05:04] <sacarasc> I want what ffmpeg outputs, not the file, but the text on the console. [05:04] <MikeJoel> as if the swf was being zoomed in to that area [07:08] <atmosx> hello [07:09] <atmosx> does this error message means my /dev/video0 doesn't produce any stream? [NULL @ 0x1644bb0] Unable to find a suitable output format for 'pipe:' [07:17] <Nosomy> yet ffplay still broke to playbacks Matroska Files with HEVC Video Streams [07:19] <Nosomy> http://pastebin.com/nKzRYSMp [07:21] <Nosomy> why only ffplay version N-64587-g2571e8f (old version), works fine? [08:15] <ash__> Hi, I'm new to ffmpeg and wanted to know whether is it feasible to have ffmpeg running in browser(as a plug in maybe) to convert the webcam stream to H264 which can be sent to a streaming server? [09:31] <myubuntu> libavformat/matroskadec.c:1173: undefined reference to `BZ2_bzDecompressInit' [09:31] <myubuntu> how to solve this problem? [09:32] <myubuntu> after I installed some packages, I can't compile my application. I was ok a few days age. [09:35] <sruz25> so I'm trying to loop, but -loop doesn't seem to work if used after the video, and loop_input/output are no longer present. Any idea how to go about it? [09:36] <sruz25> oh and loop before the input says Option loop not found [09:54] <relaxed> exit [09:55] <relaxed> :) [10:25] <atmosx> can I run ffmpeg instance on rc.local ? [10:28] <BtbN> Doesn't sound like the best idea, but i don't see why not. [10:31] <atmosx> well, it's a pain to login and start ffmpeg (streaming) at every boot. [10:58] <BtbN> It will hang the boot process, because it's waiting for ffmpeg to finish "starting" [10:58] <BtbN> it will also run as root [10:59] <relaxed> cron [11:08] <Mavrik> mornin [11:08] <Mavrik> myubuntu, you're not linking your app against bz2 library [11:25] <jul13t> hello holden, michaelni I'm interested in the Symmetric-key block ciphers project [11:25] <jul13t> i'm proficient in C and git [11:27] <jul13t> currently working on a proposal , although i've never done cryptgraphy before [11:28] <jul13t> i'm deeply interested in cyphers and stuff [11:31] <sphoorti> Hello folks! I wish to apply for the OPW program for ffmpeg. michaelni are you around ? [11:35] <sphoorti> The projects symmetric block cyphers and basic servers for network protocols caught my interest. Networks and information security have been my interest areas and I have experience in git , C and other networking internals @michaelni [11:36] <jul13t> sphoorti, looks like he's not around [11:36] <jul13t> i'm interested in working on the symmetric key block cyphers [11:36] <sphoorti> hey jul13t ! yeah looks like it [11:36] <sphoorti> oh thats nice [11:37] <jul13t> nice meeting you [11:37] <jul13t> i dropped by here yesterday and no one was online it seems they're all sleeping [11:38] <sphoorti> They might be in some other timezone. Maybe they might come online after a while [11:38] <sphoorti> jul13t: [11:40] <jul13t> aight sphoorti [11:46] <relaxed> jul13t sphoorti try #ffmpeg-devel [11:46] <michaelni> sphoorti, jul13t, another applicant (myra) yestreday already indicated interrest in symmetric block cyphers [11:47] <jul13t> michaelni, great i'll look at something else [11:47] <sphoorti> michaelni: basic servers for network protocols is also of my interest [11:47] <sphoorti> could you tell me something more about it ? [11:48] <sphoorti> or how I should proceed with it ? michaelni [11:48] <ubitux> are you guys here for the opw thing? [11:48] <ubitux> ("guys", right sorry) [11:48] <ubitux> if so, just join us on #ffmpeg-devel, that's where all the devs are [11:49] <sphoorti> thanks a lot ubitux :) [15:26] <sruz25> does -maxrate count in all streams or just the video? Alternatively why does it go over the specified bitrate? Just a link to some more comprehensive 2-pass encoding would be appreciated too. [15:43] <Ko_deZ> Hi. I am playing around with ffmpeg to see if I can make crude video analyzer. I have managed to get both average QP and standard deviation printout by extending some debug output code in h264.c. My only other wish was to print DTS together with this QP info, but I have been uable to figure out how to get that info. Is it awailable trough the H264Context struct somehow? [15:48] <MikeJoel> Making a video out of images. I am not understanding the format for designating the images (0.jpg to 25.jpg) (the docs haven't made it clearer to me) [15:49] <ubitux> %d.jpg should do [15:51] <MikeJoel> yes id does - why? [15:51] <MikeJoel> what if I was doing an image range from 25 to 100 [15:53] <ubitux> %d is a pattern that matches int [15:53] <ubitux> so any int range would do [15:53] <ubitux> if you have more files in the directory, iirc there is a -start_number or an option like that [15:53] <MikeJoel> so if the files were "pic0.jpg" 'pic'%d.jpg [15:53] <ubitux> and then you also use the option -frames:v 75 [15:53] <ubitux> to stop at 100 [15:54] <ubitux> pic%d.jpg [15:54] <MikeJoel> ok (thought I had to designate teh string part)... thats a great help... thanks [15:54] <ubitux> OTOH, if you have pic001, pic002 .. pic321 .. pic%03d (man printf) [15:57] <MikeJoel> %03 designating 3 digits and 0 showing leading 0s [16:00] <MikeJoel> ok - now I am understanding - thank you [16:03] <parkedin> Hey guys. Was following a pretty awesome guide on the website but am encountering an error and was hoping someone may be able to point me in the right direction. Thanks in advance [16:03] <parkedin> Following this guide: https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu [16:04] <parkedin> and receiving this error during the final configuration: ERROR: libfdk_aac not found [16:17] <parkedin> Ran through the libfdk_aac again and removed extraneous spaces from the ffmpeg commands and it appears to be completing. [16:35] <sine0> does anyone use gentoo here and build ffmpeg [16:35] <c_14> What's your question? [16:35] <wintershade> sine0: what a coincidence, a wild Gentoo user connects. [16:36] <sine0> huh ? [16:36] <wintershade> sine0: I use Gentoo. and I build ffmpeg. [16:36] <sine0> im trying to work out what use flags are required as im getting a blocked from building blender, blocking libav [16:36] <sine0> im assuming that ffmpeg is not being built with the required use flags [16:37] <wintershade> sine0: which USE flags do you have enabled, and what is your system profile? [16:37] <sine0> desktop and the default that come with just emerge ffmpeg [16:38] <wintershade> sine0: do you have any other USE flags enabled or disabled in your make.conf? [16:38] <sine0> what i dont know is how it can be blockoing libav and to know what flags it want [16:38] <c_14> According to the blender-2.71-r1 ebuild, it needs: ">=media-video/ffmpeg-2.1.4:0[x264,mp3,encode,theora,jpeg2k?]" [16:38] <sine0> USE="bindist mmx sse sse2 X glamor udev" [16:38] <sine0> ok ok ill just put that in [16:38] <wintershade> sine0: first of all, disable that ugly bindist thing. [16:39] <sine0> ok [16:39] <wintershade> sine0: hang on. I'm trying to see what portage is telling me about blender. [16:39] <wintershade> sine0: can you paste(bin) your exact output after the following line... [16:39] <wintershade> sine0: emerge -pv blender [16:39] <wintershade> sine0: ? [16:39] <wintershade> sine0: tia [16:40] <sine0> hang on its building! [16:40] <wintershade> lol [16:40] <sine0> c_14: way to go with building using the default rtfm instructions [16:41] <wintershade> when all else fails, read the manual. [16:41] <sine0> i dont deserve to use gentoo [16:41] <sine0> hah [16:41] <wintershade> meh [16:41] <wintershade> everybody deserves to use Gentoo. IMHO it's the second best thing ever. [16:41] <sine0> thanks for your time and dont go anywhere the build might fail and ill be running back [16:42] <wintershade> sine0: anyway... you should have python3_4 in your PYTHON_TARGETS in make.conf [16:42] <sine0> i do actually love it but its a steep learning curce and when you just want to do somethign quickly because your gf is behind you and you have to start editing make use accpt keyword files just to run a video on a friday night its pita [16:43] <sine0> PYTHON_TARGETS="python2_7 python3_3 python3_4" [16:43] <wintershade> sine0: yes, but like with oter such distros (Arch, Debian, Slackware) you actually learn to configure and use Linux. [16:43] <sine0> yes im slowly learning through the need [16:43] <sine0> its good fun [16:43] <wintershade> sine0: that's cool. you should also have PYTHON_SINGLE_TARGET="python3_4" in your env for blender [16:44] <sine0> now you have lost me [16:44] <wintershade> sine0: ok, let's go private cos we're deeply offtopic, don't wanna bother people with it here. [16:44] <sine0> but i will be back i hear squeals of a femenine nature as i just popped to the toilet [16:44] <sine0> ok [17:12] <joeflateau> to stream a video loop infinitely? [17:12] <joeflateau> ffmpeg -loop 1 -i "...mp4" -f flv "rtmp://..." doesn't work [17:14] <c_14> loop currently doesn't work for videos [17:14] <joeflateau> oh [17:14] <joeflateau> any workaround? [17:14] <joeflateau> i read about ffplay supporting it, maybe piping output from that into ffmpeg somehow? [17:17] <joeflateau> guess not [17:17] <c_14> Maybe `while true; do cat file; done | ffmpeg -i - foobar' ? [17:17] <c_14> might work [17:17] <joeflateau> hmm, i'll give it a shot [17:30] <joeflateau> c_14: while true; do cat "file.mp4"; done | ffmpeg -i pipe: -f ... seems to have worked. thanks! [17:33] <joeflateau> c_14: welp, actually ffmpeg dies when it tries to actually start over [17:34] <joeflateau> i guess while true; ffmpeg...; done will have to work for now [17:39] <c_14> hmm, it was a 50/50 chance I guess. [17:39] <kepstin-laptop> I'm kind of amazed that ffmpeg can read concatenated mp4 files through a pipe at all. [17:40] <joeflateau> it can't, turns out [17:40] <kepstin-laptop> well, in that case, just remux into mpeg ts :) [17:41] <joeflateau> hmm, remux to mpegts and cat that into ffmpeg? [17:41] <kepstin-laptop> yeah [17:41] <joeflateau> i'll give it a shot [17:43] <joeflateau> that worked! [17:44] <joeflateau> thank you both, kepstin-laptop and c_14 [17:50] <thebombzen> Sappert yay for cheating [17:50] <thebombzen> remember how I said I had to do the reading and write a blog post [17:50] <thebombzen> well I read someone else's blog post that they posted online [17:50] <thebombzen> oops, wrong channel ^_^ [18:11] <w00ds> howdy! [18:12] <w00ds> i have an ffmpeg encoding going on right now, is there a way to check how long it will take to finish? [18:12] <w00ds> is there a way prior to encoding to do a "dry run" and find out the output file size? [18:13] <c_14> Only if you specify the bitrate/ do 2-pass encoding. [18:13] <w00ds> is it necessary to specify bitrate? [18:14] <c_14> No. [18:15] <w00ds> i did this: ./ffmpeg -i XXXXXXXX.2014.3D.1080p.BluRay.x264-SPRiNTER.mkv -y -c:v libx265 -c:a copy XXXXXXX.2014.3D.1080p.BluRay.x264-SPRiNTER.x265.mp4 [18:15] <w00ds> so its re-encoding to x265, copy audio [18:15] <c_14> It'll take a while". [18:16] <c_14> You can set the crf to control the quality. [18:16] <c_14> And look at the fps= in the ffmpeg output to find out the current encoding framerate. [18:16] <c_14> ie how many frames it's encoding per second [18:16] <w00ds> frame=10328 fps=6.7 q=0.0 size= 141221kB time=00:07:11.16 bitrate=2683.1kbits/s [18:17] <w00ds> 4cores @ 4gighz each [18:17] <w00ds> im guessing its pure CPU power, right? [18:17] <w00ds> so the more CPU i through at it ... the faster it'll encode? [18:17] <c_14> yep [18:17] <w00ds> k [18:17] <w00ds> after this encoding ... i'll redo the vmware to be 8 cores @ 4gighz [19:26] <pa> question: why does handbrake generate a mkv output file with cue points every 10 seconds, while ffmpeg generates one with cue points every 4.5 seconds? even if i specify -keyint_min 250 [19:29] <pa> http://paste.ubuntu.com/8419593/ [19:30] <pa> at the end i ctrl-c'ed, like after 800 frames [19:32] <pa> ok, maybe i'm wrong in fact [19:32] <c_14> Ok, now what do you mean by "cue points"? [19:32] <c_14> I-frames? [19:32] <pa> cue points are 8.4secs spaced [19:32] <c_14> Which they should be. [19:32] <pa> that probably means 250frames @29.9fps [19:33] <JodaZ> can i have ffmpeg not touch audio samples/frames? i have a problem with it splitting audio samples when seeking [19:34] <c_14> -c:a copy ? [19:37] <JodaZ> c_14, then you can't seek i think [19:37] <c_14> You can seek, it just might not be as precise. [19:37] <JodaZ> hmm [19:43] <ElAngelo> hi, i have a couple of files ffplay doesn't seem to want to play at all but i think it's related to the video driver [19:43] <ElAngelo> is that possible? [19:43] <ElAngelo> and how to confirm this? [19:44] <ElAngelo> the files are mp42 [19:44] <ElAngelo> and the video driver is r600 [19:44] <ElAngelo> on linux [19:44] <c_14> ffplay -vn ? [19:44] <c_14> If that works, ffplay -nodisp [19:46] <ElAngelo> it says it fails to open audio [19:46] <ElAngelo> that kind of completely stuns me [19:46] <c_14> ffplay -an ? [19:46] <ElAngelo> is that even possible? [19:47] <c_14> Sure, if you say use alsa without dmix and you already have an audio output open. [19:47] <ElAngelo> hmm [19:47] <ElAngelo> right [19:47] <ElAngelo> ok i'll have to do this differently then [19:47] <ElAngelo> xbmc is already running as xbmc user and i'm trying to run ffplay as root [19:47] <ElAngelo> on top of that [19:48] <ElAngelo> probablby the audio device is assigned to xbmc now [19:50] <ElAngelo> oh crap [19:50] <ElAngelo> permission denied if i try to play it as xbmc users [19:50] <ElAngelo> *user [19:50] <ElAngelo> it's just the permissions on the file that are wrong [19:50] <ElAngelo> sigh [19:50] <ElAngelo> wasted countless hours on this [19:50] <ElAngelo> thnx c_14 [19:51] <c_14> np [19:55] <dahat> Are there any known 'newer' RTMP servers that spit out either RTMPE or RTMPS that libRTMP is known not to work with at this time? [21:42] <MikeJoel> what would be the best pixel format for converting from swf? [22:05] <asix3> does anyone know how to find out the version of avformat I'm using? I need to write an sdp file and I belive I've got an error in this line: a=tool:libavformat 55.2.100 [22:10] <sacarasc> If you have the ffmpeg binary, run that and it should tell you. [22:12] <asix3> oh, great [22:15] <rainey> how does the libavcodec decoder deal with partial audio packets? Will I have to buffer a full block of say mpeg1 layer 2 audio to pass it to the decoder, or does it have its own internal buffers that I can just pass a bitstream to? [22:15] <asix3> well, that wasn't the issue. I'm getting this error: Could not find codec parameters for stream 0 (Video: h264): unspecified size Consider increasing the value for the 'analyzeduration' and 'probesize' options [22:16] <asix3> how can I set analyzeduration and probesize? [22:17] <roninhack> Hi , i did build ffmpeg , but i cant enter multiple image in loop , forexample -loop 1 -i image.png -i audio.mp3 -i video.mp4 , im getting error only at image input s , [22:17] <Nosomy|off> asix3, the last ffplay have the same problem with matroska files with hevc video stream [22:17] <roninhack> anyone help me about it? i'm thinking , i miss library or etc? [22:18] <asix3> here's the sdp file I'm working with, along with a gstreamer pipeline that can pick up the stream: http://pastebin.com/CQLb2PVp [22:19] <roninhack> im dont streaming [22:19] <asix3> Nosomy: I don't think I'm using matroska or hevc [22:20] <asix3> and I'm not using ffplay [22:20] <Nosomy> ok... [22:20] <asix3> or am i? [22:21] <asix3> here's the file I'm working with: http://pastebin.com/AgbtFmmW [22:22] <Nosomy> only i typed, ffplay reports the same error message when try playback matroska with hevc.... [22:28] <asix3> ok [22:32] <ebfull> is the raw AVPacket data handed to the muxer of matroska going to be placed within the container without modifications? obviously it'll place metadata around it, but will it change the packet? [22:32] <MikeJoel> having a hard time finding this in the docs - still images indexed, say 1-50, 100-150, 200-250 - I thought I saw a flag for skipping missing indexes but can't find it now. [22:32] <c_14> MikeJoel: just glob? [22:33] <MikeJoel> glob? just a sec will have to read about that. [22:33] <c_14> https://ffmpeg.org/ffmpeg-formats.html#image2-2 [22:36] <MikeJoel> says globbing is not supported in the build I have [22:43] <asix3> huzzah -- the problem was that I was specifying the wrong IP. Now I'm getting this error: [h264 @ 0x7fca7b01f200] non-existing PPS 0 referenced [h264 @ 0x7fca7b01f200] non-existing PPS 0 referenced [h264 @ 0x7fca7b01f200] decode_slice_header error [h264 @ 0x7fca7b01f200] no frame! -- any ideas? [22:44] <c_14> MikeJoel: zeranoe? [23:07] <MikeJoel> hmmm actually does not appear to be a way. I guess I need to renumber. [00:00] --- Thu Sep 25 2014

1 0

ffmpeg-devel.log.20140923
by burek 24 Sep '14

24 Sep '14

[01:47] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:22844132069e: ffmpeg: replace impossible dts/pts combinations by a reasonable guess instead of hard failing [02:08] <BBB> jamrial: ok [02:09] <pross> llogan: yes, ffmpeg is used to fix a corrupted avi file (showing a murder taking place). the console output is blurred so can't make out the version number :( [02:09] <BBB> (michaelni: ^^ for his avx2/vp9 patch in https://github.com/jamrial/FFmpeg/commit/1d408433e2d3fc951d26d688d15c4fe08c…) [02:09] <pross> vlc also makes an appearance! [02:11] <jamrial> just noticed it's missing preprocessor checks for avx2 in the asm file, to prevent old yasm versions from failing to compile, so don't merge it yet [02:11] <BBB> hm& good point [03:10] <jamrial> BBB: https://github.com/jamrial/FFmpeg/commit/ed81dcc1ec0e6c634417900c477f7e2891… i'll push in a bit [03:36] <cone-636> ffmpeg.git 03James Almer 07master:6b2caa321fc2: x86/vp9: add AVX and AVX2 MC [03:38] <llogan> sponsoring programs. waste of time, IMO. [03:47] <jamrial> BBB: http://pastebin.com/upCg2xCw [04:06] <BBB> yay [04:06] <BBB> although [04:06] <BBB> we (me) should really optimize decode_coeffs_b [04:06] <BBB> and decode_b [04:06] <BBB> I did some work on it but Im sure theres more low-hanging fruit [06:53] <shipra> Hello [06:53] <shipra> I want to apply for OPW this year [06:54] <shipra> The deadline is a month from now and I want to start contributing to this org as soon as possible. [06:54] <shipra> Could someone here please guide me through the startup process. [06:59] <jamrial> shipra: stick around if none of the OPW admins answers, or come back later. it's early in the morning for most after all [06:59] <shipra> Ok :) [07:00] <jamrial> in any case check https://trac.ffmpeg.org/wiki/SponsoringPrograms/OPW/2014-12 if you haven't already. it lists some of the projects [10:11] <nevcairiel> too bad the Git url he referenced will go invalid when i rebase on top of ffmpeg master the next time [10:13] <rcombs> does 1f0 do git gc often? [10:14] <nevcairiel> i never do it manually, but i imagine it does it occasionally when i push a new rebase to shrink the repo [10:16] <nevcairiel> i was planning to release a new version tho, maybe i wont update ffmpeg until then, and i tag the ffmpeg version used for a release, so it may persist [10:58] <Diogo> hi iam using ffmpeg branch:master of git.. [10:59] <Diogo> i am trying to create a hls but i received this warning [10:59] <Diogo> debug: [mpegts @ 0x7f1934070360] Using AVStream.codec.time_base as a timebase hint to the muxer is deprecated. Set AVStream.time_base instead. [10:59] <Diogo> ? [13:27] <ubitux> wm4: i got an answer from Behdad, saying he will reply to me later (traveling) [14:16] <wm4> ubitux: eh, ok [16:17] <Diogo> /msg NickServ identify compal [17:09] <cone-883> ffmpeg.git 03Michael Niedermayer 07master:0bb5ad7a06eb: avcodec/asvenc: Fix integer overflow in level [17:10] <cone-883> ffmpeg.git 03Michael Niedermayer 07master:b0f7de3e7c73: avcodec/asvenc: fix AAN scaling [17:37] <cone-883> ffmpeg.git 03Pascal Massimino 07master:e5b3112996c3: avcodec/webp: fix default palette color 0xff000000 -> 0x00000000 [18:26] <cone-883> ffmpeg.git 03Benoit Fouet 07master:6843b9dc78bc: avformat/riffenc: Filter out "BottomUp" in ff_put_bmp_header() [21:32] <cone-78> ffmpeg.git 03Bernd Kuhls 07master:6b733be75552: Fix compile error on arm4/arm5 platform [21:33] <rcombs> what still uses arm4? [21:49] <wm4> dude we support OS/2 [21:49] <wm4> there's lots of obscure and ancient crap we "support" [22:17] <cone-78> ffmpeg.git 03Marton Balint 07master:32f1a288e2e6: ffplay: handle fullscreen status on resize event [22:17] <cone-78> ffmpeg.git 03Marton Balint 07master:9e0d1c00b583: ffplay: factorize decoder functions to generic decoder [22:17] <cone-78> ffmpeg.git 03Marton Balint 07master:1f5a3cf688bc: ffplay: move pts calculating code to generic decoder [22:17] <cone-78> ffmpeg.git 03Michael Niedermayer 07master:55179987e767: Merge remote-tracking branch 'cus/stable' [22:52] <cone-78> ffmpeg.git 03Carl Eugen Hoyos 07master:9f0ba52f3444: Skip decklink_common_c.h when running make checkheaders. [22:57] <Lvrbny> Hello, is anyone willing to help a ffmpeg noobie? Just need some raw PCM audio converted to AAC. [23:05] <gnafu> Lvrbny: I believe #ffmpeg is better suited for that type of question. [23:09] <Lvrbny> Yes, I was hoping to get the answer from ffmpeg but I can't find a code nor do I know how edit the code line to get what I need. [23:10] <Lvrbny> Oh,... the channel.. [23:10] <Lvrbny> thanks [00:00] --- Wed Sep 24 2014

1 0

ffmpeg.log.20140923
by burek 24 Sep '14

24 Sep '14

[00:30] <jverce> Hi there. I want to use FFmpeg from Java. What is the most used binding for the framework? What I need to do is mainly redirect video stream from one source to another, store the input video stream into chunks, and also take thumbnails in a regular pace. [00:49] <Rathann> jverce: you can use JNI to call FFmpeg libs directly [00:52] <foobarz> what cpu type or cpu feature (instructions, AVX etc) are most important for fast 2160p HEVC playing? [00:52] <Rathann> jverce: looks like the few ffmpeg-java projects which were there are mostly dead [00:52] <Hello71> vaapi. [00:53] <BtbN> vaapi for h265? nope. [00:53] <BtbN> CORES are most important [00:54] <BtbN> throw 8 3GHz physical cores at it, and it should work [00:54] <BtbN> If the cores are not Pentium 4 or something like that. [00:56] <foobarz> So that means Core i5 cpus are not really adaquate for 2160p HEVC? Wow. That is a lot of mainstream CPUs. [00:57] <foobarz> I just am trying to figure out these requirements for my next PC build. [01:01] <Suchiman> well i could test how 2160p playback performs on a 4670K but i do not have 2160p HEVC Content lying around xD [01:02] <foobarz> there is a downloadable sample of that 2160p HEVC on the web [01:02] <foobarz> it shows cows in a field and stuff [01:03] <llogan> someone offers to test and you say "it's on the web". [01:04] <foobarz> the time is nearly come for a hw website to do a benchmarking article on how 2160p HEVC plays on various PC configurations [01:04] <theholyduck> llogan, itsn ot like we have all these links lying around :P [01:04] <theholyduck> foobarz, but, there was a long period of h264, where the large majority of desktop cpus people had [01:04] <theholyduck> wouldnt be able to play it back [01:05] <theholyduck> but, decoders got faster, and cpus got faster. [01:05] <theholyduck> and, now its not that much of an issue in most cases [01:06] <foobarz> samples are here... http://www.elecard.com/en/download/videos.html [01:06] <theholyduck> foobarz, also, thats quite a bit of looking forward to, a video format and standard thats still quite a bits off? [01:06] <foobarz> i downloaded the 4K of cows [01:08] <theholyduck> like, i fully expect to be on an all new build by the time i start running into HEVC in the wild [01:08] <foobarz> 2160p HEVC is going to take over in just a few more months [01:08] <theholyduck> foobarz, why? [01:08] <BtbN> The next nvidia GPUs have hardware decoding for up to 8K anyway, so that problem is gone in a year or so. [01:08] <BtbN> The more interesting thing is: They also have hardware _en_coding support for it. [01:09] <theholyduck> BtbN, most hardware encoders are pretty poop though [01:09] <theholyduck> see, all the hardware h264 encoders ever [01:09] <theholyduck> foobarz, put it like this, from the time i saw my first h264 file in the wild [01:09] <BtbN> Of course they are not as good as x264, but they are more than ok. [01:09] <theholyduck> to, it being in common usage online. [01:10] <theholyduck> foobarz, was probably atleast 5 years [01:10] <BtbN> the problem is, CPUs aren't getting faster anymore at the same rate as 5 years ago [01:10] <theholyduck> thats basicly the time it took, for most people to have the cpu power to deal with it. [01:10] <theholyduck> and, the people creating videos having confidence in the encoding tools [01:10] <theholyduck> etc.etc [01:11] <BtbN> h265 will not be common before 16 or more physical cores have become a common thing [01:11] <theholyduck> and even, then, it was quite a while, before like, the scene, started using it [01:12] <BtbN> The thing is, more storage and bandwidth is cheaper than getting more encoding power. So h264 at a higher bitrate it is. [01:12] <theholyduck> BtbN, also, encode time. [01:12] <theholyduck> i havent played much with HEVC yet [01:12] <theholyduck> but, vp9 encode times are just atrocious [01:12] <BtbN> I can get 1 fps out of my i5-2500k for encoding 1080p [01:12] <theholyduck> sure, you get better video quality, but it takes you like 10 times longer to encode the same file with vp9 to h264 [01:13] <theholyduck> er, vs h264 [01:13] <theholyduck> and thats x264 --placebo [01:13] <theholyduck> i havent kept up on x265 development, but last i heard, it was just the concept encoder, with some slight tweaks [01:14] <theholyduck> no relation to x264 [01:14] <foobarz> i did notice that it took my cpu 16 threads to play the cows smooth [01:14] <BtbN> It aims to get for h265 what x264 is for h264 [01:14] <theholyduck> BtbN, but, they started by just copying the concecpt encoder source base [01:14] <BtbN> Their goal for encoding is: 1080p, 30fps, in realtime on a 16 core xeon. [01:14] <theholyduck> and starting from that [01:14] <theholyduck> x264 started from scratch with no badly done structure to start with [01:15] <theholyduck> and it still, took forever to become decent [01:15] <theholyduck> what was it, like, 2-3 years of x264 development before it stopped being terrible [01:15] <theholyduck> another 3-4 before it was commonly used [01:15] <theholyduck> foobarz, theres a reason all the initial blurays were mpeg2, and then vc-1 [01:16] <theholyduck> cause even though h264 had been a spec for quite a while at that point [01:16] <theholyduck> people didnt have decent encoders to create content for it [01:16] <JodaZ_> mpeg2 ?_? [01:16] <JodaZ_> not even mpeg4? [01:16] <theholyduck> JodaZ_, narp :P [01:16] <BtbN> I'm somewhat into video streaming, and the if you do stream 1080p60, you loose like 10-30% of your viewership, because their PCs are too weak to decode it in realtime. [01:16] <Suchiman> foobarz: configuration is i5 4670K overclocked to 4,4GHz (Default clock is 3,4, Turbo up to 3,8), 1x8GB 1600MHz RAM, AMD Radeon R9 270X, W8.1 Pro x64. With a native 1920x1080 Monitor attached through HDMI, ffplay consumes between 70% and 90% CPU. Tested on the 4K material: Elecard 4K video about Tomsk, part 3 [01:17] <JodaZ_> dvd's are mpeg4 tho aint they? [01:17] <kepstin-laptop> JodaZ_: mpeg4 isn't in the bluray spec. dvd is mpeg2. [01:17] <BtbN> So it will take decades for h265 to become even somewhat interesting. [01:17] <theholyduck> JodaZ_, narp [01:17] <theholyduck> JodaZ_, dvds are mpeg2 [01:17] <JodaZ_> °.° [01:17] <theholyduck> mpeg4 was mostly a disaster, outside of the mpeg4-asp [01:17] <JodaZ_> divx not good? [01:17] <theholyduck> JodaZ_, divx/xvid is mpeg4-asp [01:18] <theholyduck> kepstin-laptop, well to be fair, in theory, h264 is mpeg4-avc [01:18] <theholyduck> so, in a sense, mpeg4 is on bds [01:19] <theholyduck> but, most people dont refer to h264 as mpeg4, so the point is sort of moot [01:19] <kepstin-laptop> yeah, I suppose. assume I said mpeg4-asp when I said mpeg4 :) [01:19] <theholyduck> kepstin-laptop, mopst people do mean thatr [01:19] <theholyduck> i just find its better to be absolutely clear about those things [01:19] <theholyduck> but yeah, initial bds were mpeg2, and then vc-1 [01:19] <theholyduck> cause, h264 encoders were still shit [01:20] <theholyduck> and even when they started using h264 on bds, the encoders were still pretty subpar [01:20] <theholyduck> foobarz, so yeah, give it a while for people to write, decent encoders that can actually be used on a day to day basis, and actually utilize a decent chunk of the formats capabilities [01:21] <theholyduck> and the majority of people having the ability to play back the content without having to worry about a thing [01:21] <Suchiman> foobarz: VLC nightly didn't Play the video material, it played about 0,5 seconds, then it stopped and only consumed CPU [01:21] <theholyduck> h264s main benefit was that it was supported, by flash [01:21] <theholyduck> so, absoluytely everyone had a h264 decoder that just worked [01:21] <theholyduck> well, 1 of its main benefits [01:21] <theholyduck> with no need to install anything, or do much of anything. [01:22] <foobarz> Suchiman: i recommend you test with freshly compiled MPlayer and ffmpeg (internal) from snaphots. Somehow mplayer is a lot faster for me. [01:22] <Suchiman> foobarz: well, it was the ffplay from the recent zeranoe build [01:23] <Suchiman> foobarz: i will test mplayer [01:25] <Suchiman> foobarz: MPlayer only consumed 30~35% CPU (mhh wondering wether that means it's only consuming one core), it is noticeable lagging and spaming the console with: [01:25] <Suchiman> No pts value from demuxer to use for frame! [01:25] <Suchiman> pts after filters MISSING [01:27] <foobarz> Suchiman: mplayer command for me was: mplayer -lavdopts threads=16 [01:28] <Suchiman> foobarz: i only haz 4 threads and cores :P [01:28] <foobarz> Suchiman: mplayer -lavdopts threads=16 -really-quiet [01:29] <foobarz> because mplayer seems to work so well, i got the idea usage of HEVC is coming in the next few months already... with the new GTX 900 cards hw encoding is supposed to be coming too [01:30] <foobarz> seems like all the pieces to use HEVC are showing up working [01:32] <Suchiman> foobarz: much better, mplayer -lavdopts threads=4 -really-quiet 140803_4k_hm130_4s_sao_dbf_qp27.265 <-- 40~50% at slow Motion and up to 65% at quick Motion (cows) (4 cores / 4 threads, all together clocked at 4,4) [01:33] <foobarz> you could try 8 threads. it seemed to like threads for me [01:33] <Suchiman> foobarz: why should i try 8? well anyway lets see :P [01:34] <foobarz> Suchiman: i also wonder how it works for you if you were not overclocked [01:35] <Suchiman> foobarz: doesn't makes any difference (8), same usage and both very smooth [01:35] <Suchiman> foobarz: 1GHz Bonus - more Cache misses in the worst case [01:36] <foobarz> Suchiman: that video looks awesome quality... this is why i want to see this encoder working... it makes smaller files too [01:36] <Suchiman> foobarz: yeah [01:36] <Suchiman> foobarz: but the Encoder Needs to get faster and more efficient :o [01:36] <Suchiman> veryslow h264 is almost as good and fast as normal h265 [01:37] <Suchiman> making h265 slower is almost like choosing placebo [01:38] <foobarz> Suchiman: i use mplayer as my regular player... it is the best for me [01:39] <Suchiman> foobarz: but it's way to cmdline and stuf [01:40] <foobarz> i'm okay with cmd line... i make text file playlists and have made right-click start tools on my file browser [01:40] <Suchiman> using mainly Windows media player enhanced with http://shark007.net/ (less intrusive than k-lite and almost invisible, also makes preview for all the files in explorer without problems), for the rest vlc and for experiments ffplay [01:43] <Suchiman> foobarz: oh well, windows media player also plays this file ;) using about 10% more CPU than mplayer [02:05] <foobarz> Suchiman: thx for testing that. interesting to know a quad core i5 can play it smooth :) [02:06] <Suchiman> i wonder how much difference hyper threading would make [02:08] <foobarz> Suchiman: another factor for all video playback smoothness is correct amount of disk caching if the HDD is slow. i'd have to try different cache sizes to see if it matters but i barely got smooth play near 100% of my cpu usage [02:08] <Suchiman> foobarz: heh, i guess my 840 SSD will deliver enough Speed for dat [08:56] <K4T> is here somebody who ave some experience with nginx rtmp module + ffmpeg? [10:37] <fling> http://lurkmore.so/images/6/69/Blacklines.JPG [10:42] <fling> how to fix? ^ [10:51] <Sokolio> hmm, perhaps providing the commandline input would be more helpful? [12:04] <Diogo> debug: [mpegts @ 0x7f1934070360] Using AVStream.codec.time_base as a timebase hint to the muxer is deprecated. Set AVStream.time_base instead. [12:04] <Diogo> do you know if this warning already correct in branch master? [12:29] <wintershade> hey guys! me again :P [12:29] <wintershade> question. how come the hxq filters don't work anymore? have they been replaced by something? [12:31] <ubitux> hxq? [12:32] <wintershade> ubitux: this --> https://www.ffmpeg.org/ffmpeg-filters.html#hqx [12:32] <ubitux> hqx, right [12:32] <ubitux> what do you mean that dont work anymore? [12:33] <wintershade> oops. typo on my end. [12:33] <wintershade> ffmpeg gives me this error: "No such filter: 'hqx'" [12:33] <ubitux> what version? [12:33] <wintershade> 2.2.7 [12:33] <ubitux> was it present in this release? [12:34] <wintershade> no idea. I know earlier releases had it... [12:34] <ubitux> they were added in 2.3 [12:34] <wintershade> that's why I'm asking what happened to it. [12:34] <ubitux> there weren't in earlier releases [12:34] <wintershade> uh... 2.3? are you sure? I clearly remember it working a few months ago. [12:35] <ubitux> we are in 2.4 currently [12:35] <ubitux> they were added after 2.2, and were in the 2.3 release [12:35] <ubitux> check the Changelog [12:36] <wintershade> I see. for some reason, I had 2.3 installed on my system previously, then it... updated(?) to 2.2.7 then. [12:36] <wintershade> note - I'm on Gentoo. Has there been some regression that made Gentoo mask the 2.3/2.4 version? [12:36] <ubitux> no idea what broken system you use [12:36] <ubitux> not sure [12:36] <wintershade> Gentoo is not broken... [12:36] <wintershade> but let's not get into religious discussions. [12:37] <ubitux> if your system is upgrading packages from 2.3 to 2.2.7 then it's broken, or you're doing something wrong [12:37] <ubitux> anyway, ask your distro [12:38] <wintershade> hmm, okay. anyway, ffmpeg 2.3.3 is marked as unstable on amd64 here, and there is no sign of ffmpeg 2.4 in the portage tree. [12:38] <wintershade> ...probably due to the fact that it's been released yesterday :D [12:38] <wintershade> alrighty, I'll hang on for an update. [12:39] <ubitux> it wasn't release yesterday [12:39] <ubitux> only the revision [12:39] <wintershade> right. I see that now, just browsing the main website. when was 2.4 released? [12:40] <ubitux> not long ago [12:40] <ubitux> about 10 days ago [12:40] <wintershade> ugh. I'm looking at the Gentoo's "roadmap" for the ebuild... 2.2.7 is marked as unstable, 2.3.3 is hardmasked, and 2.4 doesn't exist yet. [12:40] <wintershade> I'll ask the guys there. [12:43] <wintershade> another question. any idea/recommendation for a good deblocking filter? I've found mp=uspp, but I'm not sure if it will do the trick properly... [12:44] <wintershade> btw, ubitux I must say I really like your blog ;) [12:45] <ubitux> thanks, but i'm not posting much on it :p [12:45] <ubitux> maybe i will when i'm done with subtitles, but that's boring stuff [12:46] <wintershade> ubitux: subtitles? [12:46] <ubitux> no opinion on pp filters, -vf pp is native if you want, but uspp might provide better results [12:46] <wintershade> also, I feel your pain about classifying music. [12:46] <wintershade> yeah, although uspp is called ULTRA SLOW for a reason... [12:47] <ubitux> mp={pp7,fspp} might be interesting [12:47] <ubitux> ah, and -vf spp [12:47] <ubitux> forgot i ported that one [12:48] <wintershade> good on you about that one! [12:48] <wintershade> does it work faster when ported? [12:48] <ubitux> i doubt it really makes a difference [12:49] <ubitux> it probably save the mp overhead, that's all [12:49] <wintershade> as I thought. [12:49] <ubitux> gtg [12:49] <ubitux> bye [12:49] <wintershade> cheerio~ [13:08] <pa> noob question: how to compile the sample programs shipped with ffmpeg sources? [13:30] <pa> i tried with simply using Make, but it seems that the makefile does not properly configure PKG_CONFIG_PATH [13:31] <c_14> export PKG_CONFIG_PATH=/path/to/wherever/lib:$PKG_CONFIG_PATH ? [13:44] <pa> that did it, thanks! but now i get a weird compilation error: ffmpeg_sources/ffmpeg/libavutil/time.c:60: undefined reference to `clock_gettime [13:50] <ubitux> pa: did you see the README in the doc/examples directory? [14:01] <pa> yes [14:03] <pa> i tried make examples, it doesnt work. i then tried to make in the examples directory, it kind of builds, but then it fails with that error [14:05] <ubitux> what doesn't work? [14:05] <ubitux> if you ran configure it should work [14:06] <pa> ffmpeg built just fine [14:06] <pa> but make examples fail [14:06] <ubitux> how [14:13] <anshul_mahe> pa pb! [14:13] <anshul_mahe> pa !pb [14:15] <anshul_mahe> I think you need to install ffmpeg and see if .pc are there in default PKG_CONFIG_PATH [14:19] <ubitux> anshul_mahe: it's !pb <nickname> [14:20] <anshul_mahe> thanks, for next time i will remember [14:35] <pa> ok no, it works, it's that i make clean'ed before. [14:36] <pa> (as per instructions) [14:39] <Akagi201> How can I use ffmpeg to push rtmp stream to a rtmp server? [14:41] <pa> anyhow, in the example transcoding.c, what output encoders are used? [14:42] <pa> or how to choose them? [14:51] <anshul_mahe> change the transcoding.c [14:51] <anshul_mahe> there is enum inside [14:54] <pa> ah great thanks [14:57] <pa> so by simply calling transcoding somefile.mp4 somefile.mkv i get an x264 error: http://paste.ubuntu.com/8410635/ [14:59] <pa> maybe it's easier if i modify ffmpeg.c [15:17] <pa> forgive my ignorance, but how does ffmpeg_parse_options initialize the context? it declares a private context object and returns an int.. [15:17] <pa> it doesnt even have dependency injection [15:29] <pa> is there some page describing the design of ffmpeg? [15:29] <Sokolio> nope [15:29] <pa> i see [15:29] <Sokolio> use the source, Luke [15:29] <pa> i am trying [15:30] <ubitux> pa: what design exactly? [15:30] <ubitux> there is actually. [15:30] <pa> still dont get how ffmpeg_parse_option communicate the parsing results to the rest of the program [15:30] <Sokolio> ubitux: pelas elaborate on this [15:30] <csepulvedab> hello guys. [15:30] <Sokolio> There's doxy alright [15:30] <csepulvedab> any one try to cut a video useing seek option? [15:31] <ubitux> Sokolio: well, maybe http://ffmpeg.org/ffmpeg.html#Detailed-description and http://git.videolan.org/?p=ffmpeg.git;a=blob_plain;f=doc/ffmpeg.txt;hb=HEAD [15:31] <ubitux> some stuff might have been renamed since i did that though [15:31] <spaam> soo ffmpeg is using libav. hmmmm [15:31] <ubitux> spaam: note the * [15:31] <Sokolio> oh, right [15:32] <Sokolio> I'm usually using libs [15:32] <spaam> ubitux: nice asciiart :) [15:32] <Sokolio> but yes, all the stuff can be queried, like AVOptions [15:33] <Sokolio> is there a separate irc channel for ffmpeg libs? [15:33] <ubitux> csepulvedab: http://trac.ffmpeg.org/wiki/Seeking%20with%20FFmpeg [15:33] <ubitux> Sokolio: not really, you're supposed to ask here [15:33] <csepulvedab> we use seek and the result video show twice the first frame [15:33] <csepulvedab> for example: [15:33] <csepulvedab> ffmpeg -y -i gop_10.mp4 -acodec aac -strict -2 -vcodec libx264 -ss 5.00500 -vframes 60 -level 31 -profile:v baseline 1.mp4 [15:34] <csepulvedab> the result video has 60 frames [15:34] <Sokolio> alright, here it goes: is there a DVB subtitle encoder in ffmpeg which encodes bitmap subs (nit plain text)? [15:34] <csepulvedab> but the first frame show frame 0 from second 5 [15:34] <csepulvedab> and second frame show too the frame 0 from second 5 [15:35] <csepulvedab> we know that because the original video has timestamp burned [15:35] <csepulvedab> any one can reproduce this? [15:35] <ubitux> Sokolio: ffmpeg -v quiet -codecs|grep dvbsub says yes [15:35] <Sokolio> I know [15:36] <Sokolio> but it still doesn't render subs in the most common format, which is runlength encoded bitmaps [15:36] <Sokolio> I was thinking about developing one but where to start [15:37] <Sokolio> or maybe develop the existing one [15:37] <Sokolio> there probably exists a #ffmpeg-devel channel [15:37] <Sokolio> yup [15:39] <csepulvedab> anyone? :) y try with different videos [15:40] <csepulvedab> and always happen this to me. [15:40] <Sokolio> and looking at the source, there is a possibility to render bitmaps, I wonder how to use it [15:40] <Sokolio> csepulvedab perhaps this is a bug [15:41] <csepulvedab> https://ffmpeg.org/pipermail/ffmpeg-user/2014-September/023590.html [15:41] <csepulvedab> i want someone els can reproduce this before report it. [15:48] <Sokolio> ok [15:48] <Sokolio> I have 2.3.3 at hand [16:12] <Sokolio> @csepulvedab, I confirm [16:12] <Sokolio> With ffmpeg 2.3.3 I also got duplicate first frames [16:22] <pa> why am i not able to find av_write_header with rgrep? [17:16] <Phlarp> Is it possible to skew video by 30 degrees and place it on top of a static image with FFMPEG? [17:19] <ubitux> yes [17:19] <ubitux> with rotate and overlay filter [17:22] <pa> i see there are different h264 profiles in ffmpeg, like baseline, main, high, high10, etc.. [17:22] <pa> which one is the most efficient in terms of space? [17:23] <BtbN> Those don't come from ffmpeg [17:23] <wintershade> pa: the newest, but it breaks compatibility with hardware players. [17:23] <BtbN> they are the normal h264 featureset profiles [17:24] <wintershade> pa: but the differences are minor. you should use compression presets along with the crf option (or 2-pass with target bitrate if you *really* need full control over your filesize). [17:24] <xixi10111011> they are spec defined [17:25] <pa> wintershade, what is the newest? [17:28] <wintershade> pa: no idea. you should check the specs of x264. but like I said, you shouldn't really bother. from my experience, you'll get a 0,1 - 1% in filesize. [17:29] <wintershade> pa: on top of that, if you use some really high profile and give it some moderately standard settings, you will end up with a lower profile that's only tagged as a higher one. [17:30] <wintershade> pa: they're basically constraints for encoding options, which will enable your video to be reproduced on various hardware players. [17:31] <pa> oh i see [17:31] <pa> thanks [17:31] <wintershade> pa: in other words, using (for example) baseline level 3 is practically telling the encoder "don't use bitrates or framerates that are too high for iPhone 3 to decode". [17:32] <wintershade> pa: while, OTOH, Linux and Android will always decode practically everything. My ancient Archos Gen8 with Android 2.2 Froyo had no trouble decoding high profile level 5. [17:33] <wintershade> pa: so if you aren't using some not-so-tolerant hardware player, don't bother. just use the settings that are best for your eyes and be happy about it. [17:34] <wintershade> pa: what you are probably looking for are *presets*, not *profiles* (such as ultrafast, veryfast, fast, medium, slow, slower, very, placebo - or something along those lines). use one that is acceptable to you. with crf, it will have no impact on video quality, just filesize (slower presets at crf will produce smaller files). [17:34] <wintershade> pa: HTH [17:40] <pa> so the profile does not change the filesize as much as the preset [17:40] <pa> interesting [17:40] <pa> another question: does anybody here know who wrote the matroska support in libavcodecs? any chance i could have a short chat with him/her? [17:41] <wintershade> pa: preset will impact filesize if you are using crf. if you are using 2-pass with target bitrate, it will affect quality (slower presets will give better quality - or so they say). [17:42] <wintershade> pa: no idea about matroska. sorry. [17:54] <roninhack> hi , anyone here ? [17:54] <roninhack> i have a question about ffmpeg built with opencl , when i build im getting "opencl not found" error [17:55] <roninhack> anyone here? [18:00] <pa> hm [18:01] <pa> maybe you miss opencl headers? [18:01] <pa> donno because i never built ffmpeg with opencl [18:01] <pa> also, never understood how to select the runtime [18:01] <pa> e.g., pick between nvidia and intel, for example [18:10] <anshul_mahe> I want to broadcast rtmp stream, is there anything else i need other then ffmpeg [18:27] <soulless> hi [18:27] <soulless> hi guys, I need help. [18:28] <soulless> I try use the -hls_flags but I always get the same error Unrecognized option 'hls_flags' [18:28] <soulless> I downloaded the last binary for OS X [18:29] <anshul_mahe> how to stream rtmp stream to youtube [18:31] <soulless> I try mp4 input file [18:31] <soulless> ffmpeg -i Divergent.mp4 -hls_flags single_file out.m3u8 [18:32] <soulless> ok, sorry [18:32] <soulless> I will [18:33] <soulless> http://pastebin.com/j7WpNZ1c [19:00] <anshul_mahe> do anyone know how to form youtube rtmp url [19:00] <anshul_mahe> to stream to youtube server [19:11] <soulless> https://trac.ffmpeg.org/ticket/3972 [19:23] <ChocolateArmpits> Is anyone familiar with metadata in MXF container ? [20:00] <zenny1> hi, how can one crop videos vertically? If one wants to crop equal amount on both sides (left and right), one can use crop filter like crop=in_w-2*10:in_h-2*20, but if I need to crop 40 pixels on left and 60 pixes on right and none on top and bottom, what should be the above command? Thanks! [20:00] <dahat> Are there any known 'newer' RTMP servers that spit out either RTMPE or RTMPS that libRTMP is known not to work with at this time? [20:10] <llogan> zenny1: crop=iw-100:ih:40:0 [20:57] <benlieb> I'm chopping up a large video into many small sections. All is well except for the last video in the series. I can't hear the audio. No matter what I do, I can't seem to make this work. What am I missing? [20:59] <benlieb> I guess it could be on my side, but I wonder if this is a common thing. Audio syncing issue? [21:09] <benlieb> It seems it was user error. I was fading starting at -33000 seconds. [21:09] <benlieb> bug involving times larger than one hour. [21:09] <benlieb> in my code [21:41] <Muthu> hello [21:42] <Muthu> I have already UDP multicast headend support available. Now I want to take that and stream to HLS for apple devices [22:25] <zenny1> llogan: Thanks, it worked. But the syntax is going above my head. [22:31] <anshul_mahe> how to setup rtmp server using ffserver [22:37] <BtbN> By using nginx instead. ffserver is dead [22:40] <anshul_mahe> BtbN: ok, i will try that, thanks [23:17] <Phlarp> http://pastebin.com/hZafkine I have this command which scales and rotates a video, however the scaled video is placed directly in the middle of the output, is there anyway to control this offset? [23:22] <Lvrbny_> Hello, I was wondering if someone might want to help out a ffmpeg noobie. I'm just trying to code some audio PCM to AAC. [23:23] <sacarasc> Using the libraries? Or the command line? [23:25] <Lvrbny_> i believe i'm using the cmd line [23:25] <sacarasc> https://trac.ffmpeg.org/wiki/Encode/AAC [23:25] <sacarasc> That should help. [23:26] <Lvrbny_> Thanks, i'll read that and post back in a few. [23:39] <llogan> Phlarp: you could use crop, pad, and/or overlay to move it [23:39] <llogan> pad then crop [00:00] --- Wed Sep 24 2014

1 0

ffmpeg-devel.log.20140922
by burek 23 Sep '14

23 Sep '14

[00:20] <jamrial> BBB: https://github.com/jamrial/FFmpeg/commit/1d408433e2d3fc951d26d688d15c4fe08c… [00:24] <cone-560> ffmpeg.git 03Nicolas George 07fatal: ambiguous argument 'refs/tags/n2.4.1': unknown revision or path not in the working tree. [00:24] <cone-560> Use '--' to separate paths from revisions [00:24] <cone-560> refs/tags/n2.4.1:HEAD: lavu/bprint: add const to av_bprint_is_complete() argument. [00:53] <cone-560> ffmpeg.git 03Jörg Krause 07master:56b8d106769c: configure: Refactor setting of feature test macro _XOPEN_SOURCE [01:26] <pross> 'what to pull from ffmpeg' makes up third of libav vdd report. good job guys. [01:28] <jamrial> we can't really make fun of them for that tbh, because we literally pull *everything* from libav [02:40] <iive> well, they are not going to literally pull stuff from ffmpeg. Its history is way too ... non-linear, bumpy, hairy [03:08] <cone-560> ffmpeg.git 03Deti Fliegl 07master:a5e040ee3c75: avdevice/decklink: move general code of decklink encoder to common file [03:08] <cone-560> ffmpeg.git 03Deti Fliegl 07master:bac6cfcb3a1d: avdevice: add decklink input support [03:10] <pross> iive: "smelly" [03:18] <jamrial> michaelni: should the decklink indev addition bump minor for lavd? [03:45] <cone-560> ffmpeg.git 03Michael Niedermayer 07master:f3aaec781a9e: avdevice/version: bump minor for the decklink input [04:26] <cone-560> ffmpeg.git 03Thomas Volkert 07master:e1cddd1a0c21: rtpenc_h263_rfc2190: avoid misleading error output [06:25] <rcombs> I was under the impression that the only thing people ever did with AVIs was remux them to something else [10:32] <ubitux> mmh is there a record of the ebu-tt talk @ vdd? [10:34] <cone-636> ffmpeg.git 03Benoit Fouet 07master:591e06b0e24f: avformat/img2: remove useless 'pix' duplicated entry. [11:02] <cone-636> ffmpeg.git 03Benoit Fouet 07master:9c843fb1d081: avformat/avidec: ensure that palette does not contain the BottomUp info. [11:56] <kierank> ubitux: I think yes [11:57] <ubitux> does it look sane? [13:53] <ubitux> > I would like to see avpicture_deinterlace improved, renamed, but retained as [13:53] <ubitux> > part of the main ffmpeg libraries. [13:53] <ubitux> seriously.. [14:10] <J_Darnley> renamed to what? [14:11] <ubitux> avpicture_deinterlace_broken() is the only suggestion that comes to my mind [14:12] <ubitux> they guy is like "please i know it's broken, but please keep it, and improve it so it's not broken" [14:15] <wm4> he just dislikes the terrible lavfi API [14:15] <wm4> seems justified [14:15] <ubitux> that's right, but the correct thing to do is improve the interface with it [14:15] <ubitux> not keep and "improve" this hack [14:17] <wm4> sure [14:47] <Daemon404> wow avpicture_deinterlace still exists? [14:47] <Daemon404> that makes me a sad panda [14:48] <wm4> ffmpeg never removes things [14:48] <Daemon404> is it at least deprecated [14:48] <Daemon404> even people in ffmpeg hate it [14:48] <Daemon404> afaict [14:49] <wm4> yes [14:49] <ubitux> libav also has it [14:49] <wm4> and it's in one of those major-bump-removal-guards [14:49] <Daemon404> ubitux, doesnt mean its not terrible [14:49] <Daemon404> wm4, i see [14:49] <Daemon404> so 'soon' [14:49] <ubitux> sure, just saying it's not that we are eagerly keeping it [14:49] <wm4> 2 years, give or take [14:50] <Daemon404> ubitux, sure [14:50] <Daemon404> i dont think anyone is encouraging its use [14:50] <Daemon404> ;) [14:51] <wm4> I think all of avpicture should die [14:52] <Daemon404> there was a single function i used from it iirc... [14:52] <Daemon404> avpicture_layout or something [14:52] <Daemon404> to make a contiguous buffer [14:52] <wm4> yeah [14:53] <wm4> rawdec or so uses that too [14:53] <Daemon404> i mean [14:53] <wm4> or was it rawvideo [14:53] <Daemon404> it's useful, but it should be in avutil [14:53] <Daemon404> or smth [14:53] <Daemon404> i mean a proper util func [14:53] <Daemon404> not avpict [15:38] <cone-636> ffmpeg.git 03Thomas Volkert 07master:5820358bbc4a: Add missing entry for maintainer of rtpenc_hevc.* [17:01] <wm4> ubitux: "We at XBMC got the first complaints about broken subtitles. Why was this removed?" [17:01] <wm4> hahahaha [17:01] <ubitux> i'm replying [17:01] <ubitux> that was expected [17:01] <ubitux> vlc might break too [17:02] <wm4> I just replied too [17:02] <nevcairiel> that reminds me that I wanted to check if i need to adapt anything to that [17:03] <nevcairiel> did you change the ass output of avi by any chance [17:03] <nevcairiel> (i know, ass in avi, big hack, etc) [17:04] <wm4> I don't think we ever supported that? [17:04] <nevcairiel> well it works somehow, i use avformat afterall [17:04] <wm4> got a sample file? [17:04] <nevcairiel> i was hoping to avoid that part [17:05] <nevcairiel> my archives of that age arent well sorted [17:06] <ubitux> it's not supported in avi afaik [17:07] <nevcairiel> found it! [17:08] <nevcairiel> http://files.1f0.de/samples/mewmew-ssa.avi [17:08] <nevcairiel> still works for me at least [17:09] <wm4> doesn't work here [17:09] <wm4> no subtitle stream [17:10] <nevcairiel> wonder if i hacked the detection in [17:10] Action: nevcairiel checks hack backlog [17:10] <wm4> [avi @ 0x68d920] Could not find codec parameters for stream 17 (Unknown: none): unknown codec [17:10] <wm4> Consider increasing the value for the 'analyzeduration' and 'probesize' options [17:10] <wm4> and I get streams like: [17:10] <wm4> Stream #0:4: Unknown: none [17:10] <wm4> Metadata: [17:10] <wm4> title : English [17:10] <nevcairiel> http://git.1f0.de/gitweb?p=ffmpeg.git;a=commitdiff;h=603d3c063787895bbf10ec… [17:10] <wm4> (ffprobe) [17:10] <nevcairiel> this maybe did it [17:11] <nevcairiel> only one remotely related [17:11] <wm4> could be [17:11] <nevcairiel> but of course its in the old format [17:11] <nevcairiel> but i have codez that convert it! [17:12] <ubitux> if the avi didn't hack the packet, nothing might actually change [17:12] <ubitux> also is see no mapping to the codec itself [17:12] <ubitux> (and never saw any) [17:12] <nevcairiel> i guess there is a probe function somewhere [17:12] <wm4> Stream #0:4: Subtitle: ssa [17:12] <wm4> Metadata: [17:12] <wm4> title : English [17:13] <wm4> black magic? [17:13] <wm4> which that patch above applied [17:13] <ubitux> O_O [17:13] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:fe5093aafc8f: Revert "configure: Refactor setting of feature test macro _XOPEN_SOURCE" [17:13] <nevcairiel> it can use the probe function of the ass demuxer to set the codec [17:13] <nevcairiel> its one of the neat tricks in avformat [17:13] <wm4> it even works in my player [17:13] <wm4> oh, my ffmpeg copy isn't up to date [17:13] <wm4> might be before ubitux's changes [17:14] Action: ubitux is extremely confused at what's going on [17:14] <nevcairiel> hah [17:15] <nevcairiel> which is supposed to be the short mkv format, SSA or ASS? [17:15] <wm4> how do you make ffprobe to show packet contents? [17:15] <wm4> mkv supports both [17:15] <Daemon404> *contents*? i'd be surprised if it had such a feature. [17:15] <wm4> the mangling should be the same [17:15] <nevcairiel> ASS is the short one, i looked it up [17:16] <nevcairiel> wm4: i was talking av codecs [17:16] <wm4> ah [17:16] <nevcairiel> AV_CODEC_ID_SSA is with Dialogue: and timestamp and whatnot [17:16] <wm4> yeah [17:16] <nevcairiel> _ASS is without [17:16] <wm4> ok so avi uses the old format? [17:16] <nevcairiel> yes [17:16] <wm4> awesome [17:17] <nevcairiel> whatever tool muxed this, it just copied the line 1:1 from a .ass file [17:17] <nevcairiel> maybe even ffmpeg muxed this, and just copied the stuff in there [17:18] <nevcairiel> what other containers could ass/ssa possibly be in? [17:18] <nevcairiel> pretty much just mkv? [17:19] <wm4> ogm? [17:19] <wm4> (maybe) [17:19] <nevcairiel> mkv is the only place it seems mentioned in avformat [17:20] <nevcairiel> short of the raw demuxers [17:20] <wm4> hm seems like ogm just has plain text subs [17:22] <ubitux> wm4, Daemon404 -show_packets -show_data [17:22] <ubitux> and -select_streams s:0 might help [17:22] <nevcairiel> that avi has a lot of subtitle streams otherwise [17:23] <wm4> thanks [17:23] <nevcairiel> the first stream is kinda funky tho, i think it has only one packet [17:23] <wm4> Dialogue: Marked=0,0:00:00.77,0:.... [17:23] <ubitux> oh. my. god. [17:23] <nevcairiel> i told you [17:24] <nevcairiel> Writing application : AVI-Mux GUI 1.17.8 [17:24] <nevcairiel> fwiw [17:24] <JEEBsv> oh yes [17:24] <ubitux> current ffmpeg doesn't demux them anyway [17:24] <JEEBsv> from the "yes avi can do this" guy [17:25] <nevcairiel> i think a user complained about lack of support to me one day, and it was like a 2 line fix, so... [17:26] <ubitux> i'll keep that sample for when i get a hand on a death note [17:27] <Daemon404> lol [17:27] <Daemon404> didnt look but i bet it was from circa 2005 gg [17:32] <ubitux> ooh that request_probe is so evil [17:32] <ubitux> actually that's awesome [17:32] <ubitux> but still.. [17:33] <Daemon404> the best probe is no probe [17:33] <Daemon404> everything is mp3 [17:33] <alpa_astero> I agree with the first half of that [17:33] <Daemon404> the latter is true in lavf though [17:34] <nevcairiel> i think even avformat would be hardpressed to identify a text subtitle as mp3 [17:34] <Daemon404> you may regret those words one dya [17:34] <Daemon404> day [17:34] <wm4> lol [17:35] <ubitux> try starting the subtitles "\x7ELF" or something like that [17:35] <ubitux> might trigger the mp3 probing [17:35] <Daemon404> ubitux, as a bonus it could actually work with some ass parsers [17:35] <nevcairiel> i suppose some chinese unicode subs could easily identify as binary something [17:35] <Daemon404> because some will skip any leadign two bytes [17:35] <wm4> yeah, lavf still thinks ELF binaries are mp3 [17:35] <Daemon404> because some files have a BOM [17:36] <nevcairiel> wm4: probe score 2? [17:36] <nevcairiel> :D [17:36] <ubitux> Daemon404: wm4 fixed that ;) [17:36] <nevcairiel> luckily lavf exposes the probe score now, so you can say like, probe score < 10, i dont want to open you [17:36] <wm4> nevcairiel: no, 50 [17:37] <nevcairiel> i dunno what value is required for realisticly opening most valid files [17:37] <Daemon404> nevcairiel, but some demuxers use a low probe score [17:37] <Daemon404> and only such a probe score [17:37] <Daemon404> it's entirely arbitrary [17:37] <nevcairiel> Daemon404: but not as low as 10 [17:37] <nevcairiel> its usually 25 or so [17:37] <ubitux> nevcairiel: anyway, i guess we'll probably need to keep the SSA decoder... [17:37] <ubitux> so& thanks for the sample :) [17:38] <wm4> you could make the avi demuxer depend on the ass demuxer, and share the header parser [17:38] <wm4> the novelty of "avi demuxer depends on ass demuxer" would be amusing [17:38] <Daemon404> just add it to dsputils [17:38] <ubitux> :D [17:38] <nevcairiel> i thought diego killed dsputils now [17:39] <Daemon404> true [17:39] <Daemon404> The Context was cleaned up to [17:39] <Daemon404> i dont have a whipping boy anymore [17:39] <ubitux> yes it's called videodsp now [17:39] <nevcairiel> there is still mpegvideoencctx or how its called, and its still evil [17:39] <Daemon404> i thought elenril broke it up [17:40] <nevcairiel> its not used in h264 anymore at least, but i think the evil remains in the true mpeg decoders [17:40] <Daemon404> o ok [17:40] <nevcairiel> i may have missed something tho [17:40] <nevcairiel> never know [17:44] <ubitux> btw, i wonder how we're going to be able to decode properly these subtitles [17:44] <ubitux> typically, on seek [17:44] <wm4> ubitux: because of the ReadOrder? [17:44] <ubitux> yes [17:44] <wm4> set ReadOrder to timestamp [17:44] <wm4> not very nice but works [17:44] <ubitux> ooh [17:45] <wm4> the ass muxer will buffer them, but who cares [17:45] <nevcairiel> if this is the same format as .ass files, how do .ass files work :d [17:45] <ubitux> libass won't shit bricks with the readorder gaps? [17:45] <wm4> ubitux: no [17:45] <ubitux> yeah right [17:45] <ubitux> interesting [17:46] <ubitux> nevcairiel: ass demuxer increment the readorder at each line [17:46] <ubitux> also it's not incremental since it needs to read the whole file (to stort by ts etc) [17:47] <ubitux> wm4: anyway, thanks for the trick, that sounds pretty good [17:47] <ubitux> i hope libass read the readorder as int64 ;) [17:50] <wm4> no [17:50] <wm4> just int [17:55] <nevcairiel> just use ass timebase, there is still plenty of room in 32-bit then [17:56] <nevcairiel> or 31 if its signed [17:56] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:bd6890975303: postprocess: prefetch* dont change anything, thus their arguments should be const [17:56] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:3c4fc6a78251: postproc/postprocess_template: mark unchanged function arguments const [17:58] <ubitux> note that pts are not unique [17:58] <ubitux> and typically in your sample [17:58] <ubitux> the first 2 have the same pts [17:59] <ubitux> i should probably use pos [18:16] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:3a7f9db180d9: avcodec/snow: Make block argumrnt of ff_snow_pred_block() const [18:16] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:544380aaf0da: avcodec/4xm: Make src of decode_p_block() const [18:59] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:d902a3f4cb97: avformat/adtsenc: buf isnt changed in adts_decode_extradata(), make it const [18:59] <cone-636> ffmpeg.git 03Michael Niedermayer 07master:1cf28fd5f3f6: avformat/asfenc: Make asf_write_indexs index argument const [19:10] <wm4> how do I enable debugging for a certain source file? I keep forgetting this [19:10] <nevcairiel> i know how to do it for msvc [19:11] <nevcairiel> #pragma GCC optimize ("O0") i guess? [19:11] <wm4> oh, it'd be awesome if that works, I'll try [19:12] <nevcairiel> thats what i do in msvc, and it works perfectly [19:12] <nevcairiel> different pragma of course [19:13] <wm4> this probably worked [19:13] <wm4> thanks a lot [19:21] <ubitux> hey saste [19:21] <saste> ubitux, hey [19:22] <ubitux> saste: are you going to write a summary? [19:22] <saste> ubitux, a summary about VDD2014? [19:22] <ubitux> yes [19:22] <saste> ubitux, we had a meeting with Libav, I'm going to write a summary about that, yes [19:23] <saste> let's say before the end of the week [19:23] <saste> then I missed the ffmpeg whining session, because I already left the venue when the session begun [19:36] <rcombs> can the ASS muxer be made to not buffer output? I might need that for a thing. [19:37] <ubitux> buffer output? [19:37] <ubitux> like, not flushing after every packet? [19:38] <rcombs> like, not waiting until the end if it doesn't get ReadOrder 1 [19:38] <ubitux> aah, not buffer output, no [19:38] <ubitux> why do you need that? [19:38] <rcombs> might need to stream subtitles to a client live [19:38] <ubitux> how evil [19:38] <ubitux> and you pick ass for that? [19:39] <ubitux> well, i guess we could add an option [19:39] <rcombs> libjass is a thing that works pretty well [19:39] <wm4> libjass wat [19:39] <ubitux> like, "max buffer count size" or something [19:39] <rcombs> I've been poking at it for in-browser streamed subtitle rendering [19:39] <ubitux> libjackass [19:40] <rcombs> ubitux: and I suppose I'd set that to 0 if I wanted subs output RIGHT NOW, then? [19:40] <ubitux> 1 actually i'd say [19:40] <ubitux> but yeah [19:40] <wm4> rcombs: why would you mux anyway? [19:40] <rcombs> sane values are 0, 1, and infinity, I suppose [19:40] <ubitux> you're aware that will break the readorder right? [19:40] <wm4> you'd just transfer the ass packet over http or so [19:41] <rcombs> wm4: still gotta output in some sort of format [19:41] <rcombs> ubitux: yeah, but if I'm rendering in-browser it probably doesn't matter [19:41] <ubitux> why not use webvtt? oO [19:42] <ubitux> (you need advanced markup?) [19:42] <rcombs> ubitux: because a lot of implementations suck, and a lot of browsers don't have implementations [19:42] <ubitux> ok [19:42] <rcombs> and you can't even do color properly with webvtt [19:43] <rcombs> libjass is great for dialogue with basic styling, and can even handle simple kara and TS [19:43] <wm4> hehe webvtt [19:43] <rcombs> but if you're anywhere near the point where ReadOrder matters, then you're probably beyond its limitations [19:43] <wm4> <rcombs> wm4: still gotta output in some sort of format <- well, you need some framing [19:43] <wm4> maybe wrap the ass text in json [19:43] <wm4> oh god what am I suggestion... but should work [19:43] <wm4> *suggesting [19:44] <ubitux> o_o [19:44] <rcombs> wm4: that'd require a custom ASS-in-JSON muxer and would be altogether more complex than just streaming .ass and chucking lines at libjass's parser [19:44] <rcombs> also, it'd make me feel like Crunchyroll (which does something like that) [19:45] <wm4> oh so I guess your issue is that even the code for it is trivial, you have to have it in ffmpeg? [19:45] <ubitux> maybe we can make a jass muxer [19:45] <rcombs> more that I have absolutely no reason to do it that way [19:45] <ubitux> actually, you can do that with ffprobe [19:46] <ubitux> ah mmh, not for the payload [19:47] <wm4> so, what is libjass? a new impl, or emscripted libass? [19:48] <rcombs> Arnavion's basic ASS implementation in JS using HTML nodes and CSS [19:49] <rcombs> it's limited to what you can do with, well, HTML nodes and CSS, but that encompasses pretty much everything you can rationally call "subtitles" [19:51] <wm4> that's... not what ASS is [19:52] <wm4> ASS what you irrationally call "subtitles" [19:53] <kepstin-laptop> you see the lightning bolts in https://www.kepstin.ca/dump/railgunsop.jpg ? those are done in ASS subtitles. [19:53] <ubitux> yeah that stuff is "common" [19:57] <rcombs> kepstin-laptop: that's not that complex [19:57] <rcombs> libjass couldn't handle it (it doesn't have drawing support), but there's _way_ more complex stuff done regularly in ASS than that [19:57] <wm4> oh I've seen this thing before [19:57] <wm4> someone complained that it wasn't correctly clipped [19:57] <wm4> but I forgot about it [19:57] <kepstin-laptop> yeah, that's just the example I happened to have handy [19:57] <kepstin-laptop> wm4: that image was done with mplayer's libass rendering on the opengl video output. [19:58] <wm4> same issue [19:58] <wm4> I _think_ libass should be clipping the drawing to the margins [19:59] <Guest50004> pross: did you see "the code"? a tv series in your area (I think). apparently ffmpeg makes an appearance (or maybe it's not our ffmpeg?) [20:05] <ubitux> so, fieldmatch doesn't seem to like progressive content at all [20:05] <ubitux> or, i mean, switch between the two [20:36] <llogan> is anyone getting autoresponder messages from a ffmpeg-user ML subscriber as very recently complained by Reindl H. in same ML? [20:40] <ubitux> nope [20:41] <llogan> i guess i'd have to reply to something to test [21:00] <wm4> michaelni: dalias (from musl, you might also know him from mplayer times) has said you can ask him anything about feature test macros and what exactly to do, in order to resolve the confusion [21:01] <nevcairiel> the musl people are idealists though, the real world doesnt function like that [21:02] <wm4> they also know how the real world works, though [21:02] <michaelni> wm4, i know dalias but this is really not my patch [21:03] <michaelni> Jörg Krause is the author of that (and other musl) patches [21:04] <michaelni> i just reverted it after it broke almost all bsds [21:04] <wm4> oh well I guess that guy will try ask on the musl ML again if he needs more advice [21:04] <wm4> +to [21:05] <michaelni> yes, hopefully [21:05] <michaelni> also ive no musl setup here so iam really not the right one to work on this [21:14] <cone-636> ffmpeg.git 03Nicholas Robbins 07master:c7d21dee2856: libavcodec/dvdsubdec: Add option forced_subs_only to only decode forced subtitle frames. [21:48] <mjuszczak> Are there ffmpeg 2.x packages for Ubuntu 14.04? The PPA only goes up to 1.2.x. Looking to find something that has this bug resolved: https://trac.ffmpeg.org/ticket/1918 [21:48] <JEEBcz> no packages most probably since most packages are just trying to be a replacement for the libav installed, including the libraries [21:48] <JEEBcz> and 1.2.x is IIRC the last thing with a similar API to libav 9 [21:53] <mjuszczak> I know we talked about this earlier. So is it better to just move to libav at this point...? [21:53] <BtbN> no? [21:53] <BtbN> You just can't libav with a binary incompatible ffmpeg, that would break half the system [21:53] <BtbN> so it's a compatible ffmpeg version, which turns out to be 1.2 [21:54] <BtbN> just be happy you're not on 13.10 or earlier, it's ffmpeg 0.10 there. [21:54] <wm4> I've came across this PPA: https://launchpad.net/~mc3man/+archive/ubuntu/trusty-media [21:55] <mjuszczak> My issue is: right now we're running ffmpeg 1.2 on Ubuntu 14.04 using the ffmpeg PPA (jon-severinnson). Because of that bug above (we think), we need to move to 2.x - local tests of 2.x fix the bug. Either that or libav but even then we aren't sure if libav 9.x has the bug fixed. [21:55] <ubitux> debian has packaged 2.4 in experimental, i suppose it won't take long before ubuntu follows [21:55] <BtbN> Yeah, just a few more years [21:55] <nevcairiel> ubuntu would inherit it if it moves into sid in debian [21:55] <nevcairiel> if we're extremely lucky, maybe for 15.04, but probably not [21:57] <ubitux> sid ` experimental? [21:57] <ubitux> never remember. [21:57] <wm4> sid is unstable [21:58] <mjuszczak> I did find 2.3.x packaged here for trusty: https://launchpad.net/~archivematica/+archive/ubuntu/externals [21:58] <wm4> 2.3 is now unmaintained [21:58] <wm4> because no distro used it [22:04] <mjuszczak> wm4: So the mc3man PPA might be a better bet? [22:04] <wm4> possibly [22:04] <wm4> 2.3 was dropped because no distro uses it [22:05] <jamrial> 2.3 is api compatible with 2.2 (which is api compatible with libav 10), so distros using any of the latter two should be able to use 2.3 without any issue [22:05] <jamrial> maybe we shouldn't be so hasty with calling it unmaintained [22:05] <wm4> michaelni: 2.3 is discontinued, right? [22:05] <wm4> jamrial: 2.4 should be API compatible with 2.3 (mostly) [22:06] <wm4> just like libav 11 is mostly API compatible with libav 10 [22:06] <jamrial> quite a few things were dropped, so not really [22:06] <jamrial> that didn't happen with 2.2 -> 2.3 [22:06] <jamrial> http://upstream-tracker.org/compat_reports/ffmpeg/2.2.8_to_2.3/src_compat_r… [22:06] <michaelni> 2.3 doesnt have a volunteer to maintain it, but ill very likely backport stuff anyway for a while when its easy or critical [22:09] <wm4> jamrial: no removed symbols? [22:10] <jamrial> no. just a bunch added as you can see in that link [22:10] <nevcairiel> his complaing was about the red in this one: http://upstream-tracker.org/compat_reports/ffmpeg/2.3.3_to_2.4/src_compat_r… [22:10] <wm4> ah I didn't read the link [22:10] <jamrial> whereas 2.3 -> 2.4 removed quite a few [22:11] <wm4> not that bad [22:11] <wm4> mostly things that were deprecated ages ago [22:12] <wm4> but yeah might break some things [22:13] <mjuszczak> Is there an easy way to figure out what version of ffmpeg has this fix? https://trac.ffmpeg.org/ticket/1918 -- the earliest version? [22:13] <nevcairiel> it doesnt even mention in w hich change it was fixed, so .. no [22:13] <michaelni> wm4, posted a slightly changed fix for musl support [22:14] <mjuszczak> nevcairiel: okay, thanks. [22:16] <jamrial> mjuszczak: earliest version with that fix seems to be ffmpeg 1.0 [22:16] <mjuszczak> jamrial: thank you! [22:17] <mjuszczak> weird then that we're running 1.2 and encountering a bug but 2.x fixes it - perhaps I'm looking at the wrong one. [22:17] <mjuszczak> I'll keep researching [22:17] <jamrial> commit 6f77122bf5712da1d860a0ad7174181fd0bcffd9 mentions that ticket [22:17] <mjuszczak> Awesome - thank you [22:17] <wm4> sigh, mp3 seeking is so broken [22:18] <wm4> adding some sort of full scanning/indexing pass to utils.c would probably easier than trying to fix it [23:45] <cone-636> ffmpeg.git 03Nicholas Robbins 07master:bdb7f0866683: doc/decoders: adding documentation for lavc/dvdsubdec.c option "forced_subs_only" [00:00] --- Tue Sep 23 2014

1 0

ffmpeg.log.20140922
by burek 23 Sep '14

23 Sep '14

[00:00] <t4nk705> hi, I have a trueHD audio file in .mka container and i'd like to convert it to dts-hd ma or lpcm [00:01] <t4nk705> how do i do this via command line using latest ffmpeg [00:01] <t4nk705> not sure what arguments to use after ./ffmpeg [00:36] <slimsag> Hi, I have a semi-legal question for you all. I know that you're not lawyers and can't give me any legal advice -- but I want to ask strictly of your *opinion*. [00:36] <slimsag> I want to make Go bindings to (LGPL) FFMPEG. Go *really* encourages the idea of statically linking things. I don't want to statically link FFMPEG because the LGPL license would destroy any commercial application using my Go bindings. [00:38] <slimsag> Dynamic linking *is possible* but makes deployment a pain. My idea is to make a Go library that *installs* FFMPEG for you silently, it would still be dynamically linking and would therefor fall under LGPL *technically*. [00:40] <slimsag> e.g. it would extract FFMPEG dynamic libs into a folder like $HOME/.goffmpeg/ [00:41] <slimsag> Now my question: do you think from a legal standpoint (your opinion) that this is okay to do? Obviously installers already do this -- but it would be done silently when the application was ran rather than obviously with a GUI etc [00:55] <Lac3rat3d> is there some metadata that determines default audio stream? [00:56] <benlieb> Lac3rat3d: i think the first audio stream is selected [00:56] <benlieb> if not specified [00:57] <Lac3rat3d> ok [00:59] <Lac3rat3d> i have some mkv's that are 264 dca. i converted them to 0:0 x264 0:1 aac 5.1 0:2 dca [00:59] <Lac3rat3d> but my plex server still recognizes the file as x264/dca [01:00] <c_14> Lac3rat3d: there is a flag-default for the matroska format, but FFmpeg does not currently support setting it [01:00] <c_14> You can use mkvtoolnix to set it. [01:00] <Lac3rat3d> hmm well i also changed the container to mp4 [01:00] <Lac3rat3d> so plex is recognizing it as mp4/x264/dca [01:02] <eago> :( I don0t think is possibl to do what I want [01:05] <eago> I can see how can I use the webcam as input.... but not from the browser with ffmpeg installed in a remote server [01:07] <vmBenLub1r> I'm encoding some gifs into webm using the concat demuxer and the command line output looks like this: http://i.imgur.com/0VB4yoW.png [01:07] <vmBenLub1r> the output video looks fine, but I'd like to disable whatever is giving that warning [01:58] <eago> is possible to boradcast from a machine that isn't a server to a machine that has ffserverinstalled and is a server, so that second machine streams to viewers? [02:07] <c_14> yes [02:19] <eago> and is it possible to have ffmpeg in a remote server and capture from a local webcam? [02:23] <c_14> if the webcam is forwarded to the network, yes [02:25] <eago> ok [02:44] <eago> you can think anyway I could send live video from a browser to ffmpeg on a server to then stream using ffserver? [02:44] <c_14> "send from a browser" <- how is the browser accessing the webcam? [02:45] <eago> using javascript [02:45] <eago> getUserMedia [02:46] <c_14> If you can send the video out using http,udp,tcp or any of the supported ffmpeg input protocols, yes. [02:46] <c_14> But don't ask me how to do that in javascript. [02:46] <eago> ok [02:47] <c_14> You just need ffmpeg to read/wait for input on it's internet-accessable ip address and then do whatever you want with it. [02:49] <eago> so... what happens if connection isn't great, it just pause and then resumes or it falls? [02:49] <eago> I mean, crashes [02:49] <c_14> if the connection breaks off, the ffmpeg process will stop [02:50] <eago> ok [02:52] <eago> thank you [02:59] <eago> need some rest, thank you very much [02:59] <eago> cya [06:11] <active8> if I run -filter_complex "[0:v]setpts=2*PTS[v];[0:a]atempo=0.5[a]" on an mp4 and out to mp4 twice to get 1/4 speed, all else being equal I should expect more loss in the audio than if I just ran -filter_complex "[0:v]setpts=4*PTS[v];[0:a]atempo=0.5,atempo=0.5[a]" once, right? leading up to another question, so I'm just trying to avoid an assumption [06:13] <active8> getting pretty good sound either way, btw; but this might not always work - using -c:v libx264 -preset slow -crf 0 -c:a libmp3lame -q:a 0 [06:14] <active8> ok -c:a libmp3lame -q:a 0 (since the question concerns the audio) [11:30] <voxadam> Is it possible to use ffmepg/ffserver to stream adaptive bitrate DASH or FLS streams while transcoding to lower bitrates as needed on-the-fly? [11:34] <voxadam> Ideally, what I'm interested in doing is to stream an MKV (h264/AAC) file via DASH while remuxing and transcoding on-the-fly as needed. [11:43] <JazzCZ> Hi, is there a build including yesterday's updates? I wanna use TrueHD without Dolby Atmos [11:44] <JazzCZ> can't build it myself though, hard as shit [11:48] <JazzCZ> sigh [14:25] <bearish> anyone familiar with DashCast or MP4Box? [15:13] <pa> was looking at mp4box just now, but for the first time.. :) [16:09] <bearish> pa: MPEG-DASH looks cool, but damn if i can get it to work [16:40] <active8> if I run -filter_complex "[0:v]setpts=2*PTS[v];[0:a]atempo=0.5[a]" on an mp4 and out to mp4 twice to get 1/4 speed, all else being equal I should expect more loss in the audio than if I just ran -filter_complex "[0:v]setpts=4*PTS[v];[0:a]atempo=0.5,atempo=0.5[a]" once, right? leading up to another question, so I'm just trying to avoid an assumption [16:41] <active8> getting pretty good sound either way with -c:a libmp3lame -q:a 0, btw; but this might not always work [16:41] <active8> can I do better? [16:43] <active8> if not using atempo, atempo (twice in the same filter string) maybe save to lossless and then run again and encode? Or what about doubling the samples so the ones that get dropped don't degrade the signal? [16:43] <pa> bearwhat are you trying to do? :) [16:44] <active8> [0:v]setpts=2*PTS[v];[0:a]atempo=0.5[a] is half speed slow-mo and the other is 1/4 speed [16:45] <active8> this is so I can see fingers flying and hear what they play. other apps mo0ght be differen't where I;d need better video, but this is about sound quality [16:45] <active8> pitch is prevserved - couple places sound crappy [16:46] <active8> actually, the fingers blur a bit, too, and that could be improved -one thing at a time, i guess [16:47] <relaxed> pastebin your command [16:48] <active8> relaxed, pa: http://pastebin.com/fzG5JALB [16:49] <active8> g'morning 8) [16:51] <eago> hello... I'm trying to find a protocol to stream using ffmpeg... I will have ffmpeg in a remot server and I want to do a live stream from a local computer. I am able to post blobs from my local computer to an apache server [16:51] <active8> rtmp? [16:52] <eago> ok, cool, I'll try that, thank you [16:52] <relaxed> active8: you could use -c:a pcm_s16le for lossless but it will increase the size [16:53] <active8> eago: how are you sending the blobs to the server? [16:53] <eago> from javascript with a XMLHttpRequest [16:53] <active8> ok relaxed - duh, i forgot and just did that for screencast capture. [16:53] <relaxed> active8: with -crf 0 it should be lossless as well, so I don't know why you see blurring [16:55] <active8> yeah. real fast acoustic intro. what if I doubled the samples into an intermediate file first? Think I'd have to -- not sure... increase fps and decrease rate? [16:55] <active8> or vice versa [16:57] <active8> btw -crf 0 and -crf 18 seem to be the same - like gold speaker cables don't do much, either 8) [17:10] <active8> on mp4 to mp4 -- like if i'm just ctting something out with -ss and -t, why does q:a 1 always say [libfdk_aac @ 0x3100680] VBR quality 118 out of range, should be 1-5[libfdk_aac @ 0x3100680] Note, the VBR setting is unsupported and only works with some parameter combinations [17:12] <active8> like here: http://pastebin.com/ppZwYNzc [17:24] <sacarasc> active8: You can probably use -c:a copy [17:42] <active8> sacarasc, I would, but it's not allowed with filtercomplex [17:43] <sacarasc> That command didn't show that. [17:43] <active8> it works with -c:a libmp3lame, but it's odd that -q:a 1 is in 1-5 and it tells me I'm asking for 118 [17:44] <active8> http://pastebin.com/fzG5JALB [17:44] <sacarasc> Maybe try using a newer ffmpeg, yours is over a year old. [17:44] <active8> line 31 of the paste [17:45] <active8> ah, the rebuild mess. wonder if nux-dextop has a really recent build yet [17:45] <active8> did it last month and still have to use my build for everything but VLC [17:47] <active8> sacarasc, oh yeah. I used a quick example of the error. c:a copy works when I can get away with it 8) [18:31] <pa> when transcoding to h264 using a mkv container, does mplayer know at once how many frames and cluster there will be in the output file (not the offset, obviously, but just the amount) [18:33] <benlieb> what is the best way to denoise audio? [18:58] <benlieb> Why doesn't my audio work in Quicktime (but does in other players) when combining an mp3 and video file using FFmpeg? [18:58] <benlieb> I've written up a stack overflow question if anyone cares to take a stab at this: [18:58] <benlieb> http://superuser.com/questions/815125/why-doesnt-my-audio-work-in-quicktime… [19:00] <c_14> Does quicktime play the mp3? [19:00] <voidDotClass> hey guys, i need to trim just the first second of a .3gp video. can anyone tell me the command for doing that? [19:01] <c_14> You want the first second or you don't want it? [19:02] <voidDotClass> i dont want it [19:02] <benlieb> c_14: yes it does [19:02] <voidDotClass> c_14: so it should trim the first second from the video [19:02] <c_14> voidDotClass: ffmpeg -ss 1 -i file outfile (you can use -c copy if you don't want to reencode but the cut won't be as precise) [19:04] <voidDotClass> file = input path, outfile = output path? [19:04] <c_14> yep [19:05] <voidDotClass> [h263 @ 0x9d5b3a0] Invalid pixel aspect ratio 65536/65536, limit is 255/255 [19:05] <benlieb> c_14: what do you think might be the problem [19:05] <benlieb> ? [19:05] <voidDotClass> Error while opening encoder for output stream #0.0 - maybe incorrect parameters such as bit_rate, rate, width or height [19:06] <benlieb> voidDotClass: is your height an odd number [19:06] <benlieb> ? [19:06] <benlieb> voidDotClass: as in odd/even [19:06] <voidDotClass> no idea, its a video recorded by an iphone [19:06] <benlieb> voidDotClass: it would say it in the output [19:06] <benlieb> in red [19:07] <voidDotClass> this is what it says in red: [h263 @ 0x9d5b3a0] Invalid pixel aspect ratio 65536/65536, limit is 255/255 [19:07] <c_14> benlieb: I'm guessing quicktime doesn't like the major brand/ compatible brands header. Or it just doesn't like mp3 in mp4, I don't know. Quicktime is weird. [19:07] <c_14> Do you have any such files you could test, or other media players maybe? [19:07] <c_14> voidDotClass: Is the input video h263? [19:08] <benlieb> c_14: I've been using mp3 in mp4 for a while, and haven't seen any problem. [19:08] <benlieb> What could I do to test your idea? [19:08] <voidDotClass> c_14: http://pastebin.com/iRxfYQ5z [19:08] <c_14> Try encoding the mp3 to AAC and merge that. [19:08] <benlieb> I could change the codec on the audio from copy to something else... [19:09] <benlieb> could I just specify that in the codec? [19:09] <c_14> probably [19:09] <c_14> voidDotClass: First of all, you're not using FFmpeg, you're using Libav. Either see #libav for support or download a recent copy of FFmpeg. ie one of the static builds from http://johnvansickle.com/ffmpeg/ [19:10] <voidDotClass> huh. but i'm using the ffmpeg command [19:10] <JEEBcz> then you are using a lol old ffmpeg from the time when libav still had both ffmpeg and avconv binaries [19:10] <JEEBcz> avconv being the updated one [19:10] <JEEBcz> and ffmpeg being the old ffmpeg [19:11] <voidDotClass> it says built in feb 2014 [19:11] <JEEBcz> yes [19:11] <JEEBcz> but the version probably says 0.7 or 0.8 [19:11] <benlieb> c_14: arc is experimental but I'm trying libfaac [19:11] <JEEBcz> the build date has nothing to do with how new/old the code is [19:11] <voidDotClass> well, any chance you could help me with this issue? i just need to trim that 1 second. [19:11] <voidDotClass> it is 0.8 [19:12] <JEEBcz> yes, the last Libav version with ffmpeg [19:12] <voidDotClass> you can't help me with that issue? [19:13] <JEEBcz> which was stuck in debian and ubuntu for a long long time due to the fact that there were large API changes in the next version and all packaged applications had to be updated :V [19:13] <c_14> First, download the latest static build from that link I had. Then you can just use h264 instead of h263 and it should just work (probably). [19:13] <voidDotClass> ok, trying that [19:13] <JEEBcz> lol [19:13] <JEEBcz> I'm just going to ignore all the things overlooked in that sentence [19:14] <JEEBcz> because at this point I'm tired and want to eat [19:15] <rickbol> I'm trying to tee the output of my transcoded video, but vlc launches two players for the resulting ts file, and... [19:15] <rickbol> ffmpeg no longer "sees" the -pix_fmt yuv420p... directive in my -filter_complex definition. http://pastie.org/9584834 [19:16] <c_14> rickbol: What's your command [19:16] <rickbol> c_14: long, ugly and complex, http://pastie.org/9584834 [19:16] <c_14> That's the autput [19:17] <c_14> *output [19:17] <c_14> Not the command. [19:17] <rickbol> hmmm, I thought I included my sorry, one second [19:18] <benlieb> voidDotClass: have you tried to reset the SAR or DAR manually? [19:18] <benlieb> c_14: it seems to work when using libfaac. I wonder why id doesn't like the mp3 [19:18] <rickbol> http://pastie.org/9584899 [19:19] <benlieb> c_14: but all of this is just to clean up the audio. What do you recommend for audio denoising? [19:20] <rickbol> benlieb: I haven't used ffmpeg for audio processing, but SoX might have filters and stuff for denoising, but I'm not sure. [19:21] <benlieb> rickbol: right now my method is to split the audio from the vid, and then use audacity or an eternal program. [19:21] <voidDotClass> benlieb: i haven't, how would i do that [19:21] <benlieb> It would be great if I could do something right in ffmpeg. I have thousands of videos to deal with. [19:22] <benlieb> voidDotClass: dunno if it will work but http://www.ffmpeg.org/ffmpeg-filters.html#setdar_002c-setsar [19:24] <c_14> benlieb: I don't think there's anything in FFmpeg to denoise audio. [19:24] <benlieb> c_14: that seems to be the case. in that case what would you recommend. The best denoiser I've found is in iMovie. [19:24] <c_14> rickbol: What is exact problem again? [19:25] <benlieb> but it won't do just audio. [19:25] <c_14> benlieb: hmm, I'd probably try sox or audacity. There's a few others I've heard of, but none I've tried. [19:25] <benlieb> it takes forever to import and then reexport, mostly for video processing, but I don't even need the vid processed [19:25] <benlieb> c_14: i tried audacity. the quality of the denies pales compared to imovie [19:26] <benlieb> which is probably using the same tech in iphone [19:26] <benlieb> woh autocorrect [19:26] <benlieb> the quality pales... [19:27] <benlieb> so I'm importing the vid, denoising, then re-exporting to a very small vid size, extracting the audio, then recombining. [19:27] <benlieb> not exactly efficient [19:27] <benlieb> is sox command line? [19:27] <benlieb> never heard of it. [19:27] <rickbol> c_14: couple things: 1) with the "tee...-map..." in the last line, ffmpeg reports the -pix_fmt error; 2) in playback in vlc, two vlc instances are launched, one has video, the other mostly grey with some blotches... [19:28] <c_14> benlieb: yes it is, might be a bit complicated but there should be guides on the internet [19:28] <rickbol> c_14: ... in vlc (when playing the resulting ts file from disk)... [19:29] <c_14> rickbol: 1) the -pix_fmt isn't an error, just a warning. 2) can you try playing the file with ffplay or some other video player? [19:29] <benlieb> yeah, looks involved. [19:29] <benlieb> maybe for another day. [19:29] <benlieb> maybe another life, lol [19:29] <pa> if i want to encode in h264 and control the generation of the container, how do i do? should i use libavcodec and libavformats directly? [19:30] <rickbol> c_14: it plays from totem, but this vlc behavior doesn't happen if I don't tee the ffmpeg output. If I just write to a ts file only, vlc plays fine. [19:31] <c_14> try manually setting the format for the ts file? [19:32] <rickbol> c_14: the biggest problem seems to be that processing bit rate goes thru the roof, and my machine can't do it. But I'm really only taking the same transcoded output and writing to disk and a socket. [19:33] <rickbol> c_14: if i don't use the -pix_fmt switch (without tee-ing), vlc won't play the ts file at all. [19:34] <c_14> vlc might not like yuv422p [19:34] <rickbol> c_14: it should be yuv420p [19:35] <rickbol> c_14: that's why I'm using the switch. [19:36] <c_14> You mean the format filter? If you only want to use the format filter and don't want to specify -pix_fmt then you'll have to place it after the overlay filter. [19:36] <rickbol> c_14: I'm thinking there's something going on with the -map statements between my -filter_complex and my output, but I jsut don't understand -map yet. [19:37] <c_14> Your map is fine. [19:37] <rickbol> c_14: so what are the "[xxx]" things, just names? [19:38] <c_14> I think they're called pads, you can visualize them as buckets. [19:38] <c_14> You put things into the bucket. [19:38] <c_14> And then you take it out again. [19:38] <rickbol> yea, ok [19:39] <rickbol> c_14: are you saying I need to move the yadif, denoise and format, elsewhere in my command? [19:39] <c_14> just the format [19:41] <rickbol> c_14: ok, but I don't get the warning if I leave the "tee.." out of my output... which seems strange. [19:42] <c_14> It does... [19:45] <voidDotClass> c_14: so i've downloaded the static library and tried it again, here's my output this time: http://pastebin.com/4Sj65dCh [19:45] <voidDotClass> any ideas? [19:45] <vmBenLubar> what does "DTS -922328613750660, next:2260000 st:0 invalid dropping" mean? [19:45] <c_14> voidDotClass: add -c:v libx264 [19:46] <rickbol> c_14: tee-ing for my case, since my file container is a transport stream (ts) and my I'm just routing that to a socket, that there's way to much processor being consumed [19:46] <voidDotClass> ./ffmpeg -ss 1 -i ~/Vid*/VID*.3gp -c:v libx264 out.3gp ? [19:47] <c_14> ye [19:47] <voidDotClass> now i get this: http://pastebin.com/Pd1y8U7f [19:48] <c_14> add -c:a copy [19:48] <vmBenLubar> here's my command line and input and output: http://pastebin.com/wRmJA6zv [19:49] <voidDotClass> replace the -c:v with the -c:a ? [19:49] <c_14> nah [19:49] <voidDotClass> or add a new one in addition to the last? [19:49] <c_14> just add another one [19:49] <voidDotClass> ok [19:50] <voidDotClass> nice, finally seems to be working [19:50] <c_14> vmBenLubar: As long as the output file[s] play, ignore it. [19:50] <vmBenLubar> c_14: it does, but can I get it to not print that message a billion times? [19:50] <c_14> You can probably decrease the loglevel. [19:51] <vmBenLubar> I tried that, but it also removed "frame= 112 fps= 55 q=0.0 q=31.0 size= 63kB time=00:00:02.26 bitrate= 227.5kbits/s dup=1 drop=0" [19:51] <rickbol> c_14: my purpose for tee-ing was so I could view the stream (confidence monitoring) of the file, But this tee-ing uses too much cpu. I can [19:51] <rickbol> jsut open the file in mplayer w/o cpu problems. taht doesn't seem right. [19:52] <c_14> It shouldn't use more cpu, it might burn a bit of io, but hmm. [20:05] <rickbol> c_ [20:08] <rickbol> c_14: there's some interaction between my filter and tee-ing the output. I removed "filter=yuv420p" from filter_complex, and added -pix_fmt yuv420p elsewhere, and now... [20:08] <XHFHX> Hi there. I have a problem with the pipe command. At first I tried "-f avi pipe:" but the quality wasn't good enough. So now i changed it to "-c:v rawvideo -f avi pipe:" which works, but only until the file hits 360MB, after that the file doesn't grow anymore although it's still rendering. I'm piping to ffmbc [20:08] <rickbol> c_14: I see that the second vlc instance shows me the video, but w/o the overlay. [20:09] <c_14> And the first instance shows the video? [20:09] <rickbol> c_14: so my file has two video streams, one with overlay, and the second, without. [20:09] <rickbol> both have the main video. [20:10] <c_14> oh, wait, yes [20:10] <c_14> You have too many maps [20:10] <rickbol> ah, I had hoped so! [20:10] <c_14> get rid of the -map 0:v and the -map 0:a [20:11] <c_14> Didn't see those last time. [20:11] <pa> is there a guide somewhere to transcode using libav* directly and not ffmpeg? [20:11] <rickbol> c_14: in the output line? [20:11] <pa> like for example to create the header and the trailer by hand [20:11] <pa> and the raw data stream [20:11] <llogan> XHFHX: why pipe to ffmbc? [20:12] <pa> i tried to look into the source code a little, but its large and i got confused pretty quickly [20:13] <llogan> pa: take a look in doc/examples. it may or may not have some useful stuff for you [20:13] <pa> thanks! [20:13] <XHFHX> because ffmpeg doesn't support XDCAMHD/xd5c [20:13] <llogan> pa: also https://www.ffmpeg.org/doxygen/trunk/index.html [20:14] <llogan> XHFHX: so why pipe from ffmpeg instead of using ffmbc to decode? [20:15] <XHFHX> because ffmpeg doesn't support concatting of videos which I need [20:15] <rickbol> c_14: much better!!! I'm able to write to file and monitor @ < 100%cpu ! thx. Now I need to tweak for max video quality while not maxxing cpu. [20:15] <llogan> XHFHX: do you mean ffmbc in that last sentence? [20:16] <XHFHX> llogan: sorry, yes :) [20:17] <XHFHX> llogan: http://pastebin.com/aTGNtQMH [20:17] <llogan> and if ffmpeg doesn't support something you can submit a feature request. it may be possible to port from ffmbc (despite the lack of a version control repo and the license) [20:18] <llogan> XHFHX: and the COMPLETE console output [20:18] <llogan> many people ignore that part. i'm not sure why. [20:20] <XHFHX> llogan: oh, sorry. but i don't know how to output the complete console output in windows, as the first lines begin to disapear after some time [20:21] <XHFHX> currently im making a test with piping from ffmpeg to ffmpeg to see if it's an pipe error or a ffmbc error [20:21] <XHFHX> because when I used the pipe without the -c:v rawvideo it worked but the quality was awful [20:24] <_dunno_> ffmpeg by mistake considered my IN to be OUT - now the file still exists in the file table but is of zero-size. Is there some approved procedure which could be used to revive my movie? [20:24] <llogan> did you encounter a bug that caused this to happen? [20:25] <active8> if using setpts to slow down a video and atempo to keep the sound pitch the same - i get some blurring and maybe a little bit of sound degradation. mp4 to mp4. so those filters work on timestamps and I'm not sure if this will work, but how could I double - make redundant frames - and get that to play like the original, and then maybe if I slowed it down with setpts and atempo, I wouldn't be losing info. [20:26] <_dunno__> sry - no - not a bug - but i ended up with an error msg [20:26] <active8> or just fake the filter by changing bitrates and fps? [20:27] <active8> i'd still need redundant info, i think [20:30] <llogan> active8: sox may do a better job for audio, and others have mentioned using slowmovideo for video to perfrom interpolation [20:32] <debianuser> _dunno_: no easy way, sorry. :( Try regular file recovery tools (or shut the PC down, disconnect a disk where your movie was and bring it to some file recovery service) (it's better to not write anything to that disk, that would increase a chance to recover a file) [20:34] <active8> llogan: i wonder if openshot can do slow mo (it's not a built in effect) also wonder if syncing will be a chore after using two different tools [20:34] <llogan> sorry. i don't know the answer to either of those questions. [20:35] <debianuser> (for some systems if you notice a file loss and instantly cut of the power from the dist within ~5 seconds after file being overwritten there's a close to 100% chance to manually recover the filesystem with the original file on it) [20:35] <active8> i heard sox is good and it''s in one of my trusted repos already [20:44] <XHFHX> llogan: http://pastebin.com/G1yHAjKD hope this is ok, somehow it didn't store the console output into a file, so I only have the beginning and the ending of the log, but I think this is the intresting part [20:46] <active8> i seem to be lacking a resource that explains the relation between fps, bitrate, etc for both audio and video - in case I want to tweak things (including speed up and slow down, maybe) Where's a good article? I saw some equation posted once. I hope the info isn't scattered across 100 bad articles [20:47] <kepstin-laptop> well, there's no direct relation between fps and bitrate, except that in most codecs you need more bitrate to encode something with higher fps at the same quality [20:50] <active8> bitrate is related to how fast it spits out to the screen, right? like bandwith in ntsc - pixels per second = bitrate/bits-per-pixel (roughly) ? [20:51] <active8> so you'd need more bps to fill a hiher fps [20:51] <kepstin-laptop> active8: only for raw video. For modern encoded video, it's more complicated [20:52] <active8> yeah, that's the prob [20:52] <active8> i would be converting to raw if it wasn't already raw, doing my thing, and reencoding [20:53] <active8> just want the nuts and bolts engineering in one article to get a handle on it all. [20:53] <active8> without the huffman coding explanation, wavelets, etc [20:54] <kepstin-laptop> pretty much the only reason to set an output bitrate nowadays is that you need the file to fit a specific size, or to fit in a bandwidth restriction due to format (e.g. bluray drive speed, internet connection speed); for most uses, quality-based encoding (e.g. x264's crf) is preferred. [20:55] <kepstin-laptop> but as far as bitrate goes, what happens is that you pick the bitrate you want, and then the codec is responsible for allocating bits to try to keep the best quality it can with the bits you're allowing it to use. [20:56] <active8> what if I want to take each audio or video frame, duplicate it, and encode it at a different fps and have it play at the same "speed". if I can do that, I can them change the speed without messing with setpts dropping frames I need - maybe. [20:57] <active8> say raw or mp4 to raw (and mp3 or aac to wav) and then mes with the info before reencoding at [insert equation i asked for] bps/fps [20:58] <active8> double # of frames and set fps to oldfps*2 or same-fps to get half speed [20:59] <active8> then VLC chokes on an off FPS and I'm back to how can it be done [20:59] <kepstin-laptop> I'm not clear on what exactly your final goal is [21:00] <kepstin-laptop> do you have a source video, and want to make a new video that's slow motion - i.e. it takes twice as long to play? [21:00] <active8> not using setpts and atempo [21:00] <active8> i should extract the pictures and flip them like a deck of cards [21:01] <XHFHX> hi there, can anyone help me with this? http://pastebin.com/G1yHAjKD no matter how big the video is it renders, it doesn't grow beyond 360mb but still renders the whole video [21:02] <kepstin-laptop> active8: it sounds like you had something that worked, but the final encoded video and audio quality was too low? [21:03] <llogan> XHFHX: did it work as expected when you piped from ffmpeg to ffmpeg? what if you try a different encoder? what if you try a different format? [21:03] <XHFHX> ffmpeg->ffmpeg is still rendering [21:04] <llogan> if that works then you'll have to ask in #ffmbc I guess [21:05] <active8> kepstin-laptop, yeah. little bit of sound degradation, but blurring and it looks like the sync is off, but who knows with a youtube of a dude playing a lightning fast classical acoustic intro to the best song ever written (almost) [21:05] <kepstin-laptop> well, you'll get some audio degredation with the audio stretch, but it sounds like your main issue is that you didn't set the quality on the encoders for the output video sufficiently high. [21:06] <active8> try youtube rush la villa strangiatto acoustic intro (you'll see that from the thumbnail) cut the intro with -ss t and see what I mean - i'll post the command [21:06] <active8> -c:v libx264 -preset slow -crf 0 -c:a libmp3lame -q:a 0 [21:06] <active8> -filter_complex "[0:v]setpts=4*PTS[v];[0:a]atempo=0.5,atempo=0.5[a]" -map "[v]" -map "[a]" [21:07] <active8> in reverse order, of course [21:08] <active8> the docs talk about presets, but not really how they affect things other than compression and file size [21:08] <active8> and they're not even clear on that [21:08] <kepstin-laptop> hmm, yeah, those settings should give you pretty much the best audio and video you can [21:09] <active8> "the choice is about this over that" but nothing about which does what [21:09] <XHFHX> llogan: ffmpeg->ffmpeg is fine, so it must be a ffmbc problem :/ [21:09] <active8> that's why I want to try it the ugly hardcore way like [21:12] <llogan> XHFHX: they may tell you to try 0.7.1 (if they even respond at all). [21:12] <kepstin-laptop> if you're looking at the samve video I see, youtube id bOyEerX6k9M, the original video is pretty bad quality, so it's never gonna not look blurry [21:12] <active8> not just for this project, either. i could maybe just slow down VLC and do alright, but there has to be some knowledge of how to do this without setpts and google isn't obeying me on this one [21:14] <llogan> XHFHX: until then keep trying other stuff. -vcodec huffyuv. -f mov, -f matroska, etc. just shooting blind here... [21:14] <active8> the only other one with an acoustic intro has the camera a bit far from the fretboard, but I might try that one, anyway. IT's motion blur big time at half and quarter speed. [21:14] <kepstin-laptop> active8: but yeah, many videos that don't appear blurry at full speed will appear blurry when slowed down, simply because of how your mind perceives motion. Both the original and the slowed down have the same blur [21:15] <active8> true [21:16] <kepstin-laptop> and no matter what, you will get audio degredation when slowing down audio to Œ original speed and attempting to keep pitch. Might be some tools that can do a better job than ffmpeg's atempo filter, but I dunno which. [21:18] <active8> and there are two different positions to play this in. I might have to load XP on a machine and get the music printed out and just use my ear on the slowed down audio. I'd still like to find a way to avoid setpts/atempo. I know there was a way cause the docs refer to an old way. And I also know thre was an equation out there with bps fps kinda stuff in there maybe sample rate, fps, and speed. [21:18] <XHFHX> llogan: i tried the same command again, but now I'm getting a "broken pipe" error message... [21:18] <active8> sox might work on the audio [21:19] <active8> a pipe's supply was terminated - just got that on ctrl-c from a pipe into less [21:19] <kepstin-laptop> active8: you're not going to be able to get any better video than what that command is already giving you. But you might be able to use sox to do the audio, it might give better results. [21:19] <XHFHX> llogan: nevermind, somehow the ffmpeg.exe file deleted it's concetnt itself so it was 0kb [21:20] <kepstin-laptop> (the only way to get better video would be to use a higher quality source video) [21:20] <active8> garbage in, garbage out [21:20] <active8> next time alex lifeson gets locked up, i'll have the judge order community service right here [21:21] <kepstin-laptop> all that setpts command does is make each video frame appear on the screen for 4 times as long, and the x264 settings are making it save the frame exactly as it was in the original video. [21:21] <active8> ok. not sure why i was thinking it dropped frames. that would be fast motion [21:23] <active8> well thanks, kepstin-laptop. At least I have something that will work on future endeavors with good video. think i'll go record my neighbor kicking his car and busting his knuckles with a wrench [21:25] <mjuszczak> Are there ffmpeg 2.x packages for Ubuntu 14.04? The PPA only goes up to 1.2.x [21:29] <pa> llogan, but in theory does ffmpeg itself allow to produce only the "container" of a transcoded file? that is, without the av data itself? [21:38] <XHFHX> @llogan: it seems -c:v rawvideo makes some problems. do you know what I could use instead? when I use nothing it's a damn ugly quality :/ [21:39] <XHFHX> looks like 256 colors with a resolution of 200x400 upscaled to 1080p [21:54] <roninhack> hi [21:54] <roninhack> anyone here? i have a question about building opencl [21:55] <roninhack> anyone here? [21:56] <MikeJoel> ffmpeg? told it would convert flash or avi to mpeg4? [21:59] <MikeJoel> is ffmpeg an app or dll, or?? [22:04] <c_14> https://ffmpeg.org/ [22:06] <MikeJoel> thanks [22:14] <llogan> XHFHX: i mentioned huffyuv before [22:15] <llogan> pa: i don't know [22:16] <pa> does it happen that av_write_trailer actually modifies the header too? [22:20] <Sokolio> pa; yep [22:21] <Sokolio> in mp4, when moving moov [22:21] <pa> also in mkv i suppose [22:22] <Sokolio> maybe [22:22] <pa> do you know, however, if , for example in mkv, it is possible to know beforehand the number of cue points? (not their offset, obviously) [22:23] <pa> it looks like that ffmpeg api are a little how to say [22:23] <pa> inflexible , if one wants to do something weird [22:24] <Sokolio> maybe libMatroska would help? [22:24] <pa> is it used by ffmpeg? [22:24] <Sokolio> I like to use faac or x264 directly [22:25] <Sokolio> so why not use a muxing lib directly [22:25] <Sokolio> i have no idea, I guess ffmpeg has its own parser [22:25] <pa> you mean muxing raw x264 streams? [22:25] <Sokolio> no, I meant encoding through x264 directly and not by using libx264 codec [22:26] <Sokolio> sometimes it's not bad to walk around ffmpeg [22:26] <pa> could be an option.. can x264 write mkv by itself? [22:27] <Sokolio> the lib? nope [22:27] <Sokolio> it's just an encoder [22:27] <pa> and the output is? mp4? [22:27] <Sokolio> the output is h264 bitstream, no container, no format [22:27] <pa> so raw h264 [22:28] <Sokolio> yes [22:32] <kepstin-laptop> x264 can optionally write to mkv, flv, or mp4 depending on how it was compiled. [22:32] <kepstin-laptop> (the executable) [22:33] <Sokolio> yep, the binary executable can mux several formats [22:33] <Sokolio> but the lib is just for encoding [22:34] <JodaZ_> kepstin-laptop, how does it do that? [22:35] <kepstin-laptop> JodaZ_: for mp4 it uses gpac; it has built-in minimal muxers for the other formats. [22:35] <pa> for mkv? libmatroska? [22:36] <JodaZ_> k [22:37] <Sokolio> I think I've never used x264 through CLI [22:38] <Sokolio> I peeked ino the source [22:38] <Sokolio> there;s a minimal muxer of their own design [22:39] <Sokolio> so it's not libmatroska [22:39] <Sokolio> there's also a flv muxer [22:40] <pa> ah i see [22:42] <Sokolio> but that won't probably solve your issues with cue points [22:46] <pa> well what i need in fact is just the amount of necessary cue points and their position in time [22:46] <JodaZ_> hmm, question, will ffmpeg get fame exact seeking any time soon? (so one can make HLS/DASH fragments with it) [22:46] <pa> i guess this could be derived by the number of frames in the movie and selected number of frames per keyframe [22:46] <kepstin-laptop> JodaZ_: seeking should be frame exact by default in current ffmpeg versions (unless you're using -v:c copy, of course) [22:47] <JodaZ_> kepstin-laptop, having problems with audio [22:48] <pa> kepstin-laptop, seeking as in -ss? [22:48] <JodaZ_> with audiosamples split and filled with silence [22:55] <pa> but is there a way to make ffmpeg seek by frame/keyframe instead of time offset? [22:55] <pa> i mean either ffmpeg, or via ffmpeg apis in some custom example [23:03] <JodaZ_> pa, you also interested in perfect splitting? [23:07] <llogan> kepstin-laptop: or l-smash instead of gpac [23:14] Action: Suchiman hears perfect splitting [23:14] <Suchiman> yeah thats something i could also need :P cutting away ads from recordings [23:14] <Suchiman> on the other Hand still searching for something making cutting as easy as possible [23:20] <pa> yes, i need it for correct transcoding when seeking [23:22] <Suchiman> pa: well, if you write -ss before the Input file, the Navigation will be keyframe based [23:23] <Suchiman> pa: https://trac.ffmpeg.org/wiki/Seeking%20with%20FFmpeg#Fastseeking [23:26] <JodaZ_> Suchiman, well for ads the cuts don't need to be perfect [23:27] <JodaZ_> Suchiman, problem there is that its only keyframe based for the video, the audio is still cut inside frames [23:27] <Suchiman> JodaZ_: is there a recommend technique? i'm currently cutting out only the movie material (copy), then put them back together using the concat demuxer [23:28] <JodaZ_> don't think there is one [23:29] <Suchiman> mhh.. a skip Video filter would be interesting [23:29] <JodaZ_> that stuff exists [23:29] <JodaZ_> but filters only work for reencoding i think not for copy codec [23:30] <Suchiman> would be okay for the final stage, comming from raw / lossless compressed Input to cutted and filtered into transcoding [23:35] <vmBenLubar> how do I find out what pix_fmt are supported with webm? [23:35] <Suchiman> vmBenLubar: should be dependent on the codec, do you mean vp8 / vp9 ? [23:35] <vmBenLubar> vp8 [23:36] <Suchiman> vmBenLubar: according to the RFC: VP8 works exclusively with an 8-bit YUV 4:2:0 image format. [23:36] <Mavrik> https://www.ffmpeg.org/doxygen/trunk/libvpxenc_8c_source.html [23:37] <Mavrik> 896 .pix_fmts = (const enum AVPixelFormat[]){ AV_PIX_FMT_YUV420P, AV_PIX_FMT_NONE }, [23:37] <Mavrik> so that's that [23:37] <Mavrik> oh sorry, that's VP9 [23:37] <Mavrik> VP8: 867 .pix_fmts = (const enum AVPixelFormat[]){ AV_PIX_FMT_YUV420P, AV_PIX_FMT_YUVA420P, AV_PIX_FMT_NONE }, [23:40] <vmBenLubar> thanks [23:48] <asix3> hello everyone. I'm trying to grab video streamed over rtp from a raspberry pi -- the code I have to do so is here: http://pastebin.com/1ALcP0WH . I get an error saying: Unable to receive RTP payload type 96 without an SDP file describing it [23:48] <asix3> I'm streaming it from the pi using this gstreamer pipeline: raspivid -t 0 -h 720 -w 1080 -fps 25 -hf -b 2000000 -o - | gst-launch-1.0 -v fdsrc ! h264parse ! rtph264pay config-interval=1 pt=96 ! udpsink host=131.156.68.41 port=5000 [23:48] <asix3> thoughts? [00:00] --- Tue Sep 23 2014

1 0

ffmpeg-devel.log.20140921
by burek 22 Sep '14

22 Sep '14

[01:12] <cone-123> ffmpeg.git 03Michael Niedermayer 07master:28dce3cdbc33: avcodec/alacenc: Remove unused variable [01:29] <cone-123> ffmpeg.git 03Michael Niedermayer 07release/2.2:5df02760dd2f: update for 2.2.8 [01:53] <BBB> oh was/is vdd this weekend? do we get an executive summary? [02:11] <cone-123> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n2.2.8': unknown revision or path not in the working tree. [02:11] <cone-123> Use '--' to separate paths from revisions [02:11] <cone-123> refs/tags/n2.2.8:HEAD: avcodec/alacenc: Remove unused variable [11:11] <kierank> hmmm interlaced chroma upsample + normal scale won't work the way i propose [11:12] <wm4> so ilpack is useful? [11:15] <kierank> dunno let me implement and see [11:36] <kierank> michaelni: will swscale understand the concept of chroma lying outside the plane? [11:38] <kierank> http://i.msdn.microsoft.com/dynimg/IC500902.png [11:38] <kierank> specifically in 420 mpeg-2 [11:38] <kierank> the second line has a chroma sample that lies above it [11:47] <wm4> I thought swscale can now do arbitrary chroma posititioning, but not sure if it would go out of bounds in your case [12:17] <JazzCZ> hi, where do i get the truehd patch? [12:17] <wm4> Warning: Encountered more data after announced end of track (frame 14054/14054). Frankenstein! [12:17] <wm4> ... [12:18] <wm4> This was a Frankenstein track. [12:18] <wm4> ^ how mpg123 handles concatenated mp3s [12:18] <wm4> JazzCZ: this one? https://ffmpeg.org/pipermail/ffmpeg-devel/2014-September/163189.html [12:19] <JazzCZ> yes man, that is right [12:20] <JazzCZ> yeah, that's what I mean [12:24] <michaelni> kierank, ill push a fix for that in a moment [12:31] <nevcairiel> considering the one Blu-ray that uses this feature wasn't even released yet, a lot of people sure seem to have pirated it :p [12:34] <wm4> so is this yet another rarely used extension? [12:34] <nevcairiel> who knows [12:34] <nevcairiel> its new [12:35] <nevcairiel> impossible to say how often it will be used [12:38] <wm4> so I assume it's strictly optional, and the issue is really just that there was a parsing bug? [12:38] <nevcairiel> yeah, plus that you need to skip the new data, as its not directly audio data [12:39] <nevcairiel> they use the truehd substream to transport another compressed bitstream inside [12:40] <nevcairiel> afaik, there is no software implementation of this yet, commercial or otherwise, the only way to decode it are external devices through HDMI streaming of the truehd stream [12:41] <rcombs> is it actually more compressed audio data, or just channel mapping/panning meta? [12:42] <nevcairiel> i assume its compressed audio data, its big enough for that at least. Its object-based audio data, which you can then render depending on your speaker configuration, i don't think you can reconstruct that from the other pcm streams [12:43] <rcombs> fun [12:44] <nevcairiel> its a fascinating concept for sure, but the problem is that you can never render that properly unless you have detailed information about the room and setup you're rendering to, which means you ideally need a measuring mic [12:44] <nevcairiel> works for the HDMI AV Receivers, those come with mics usually anyway [12:45] <rcombs> mine did! [12:45] <wm4> lol really [12:45] <wm4> fascinating [12:45] <wm4> so they calibrate themselves [12:45] <rcombs> it also only has 3 HDMI input ports, and can't convert analog video to HDMI, and can't play HDMI audio on its second zone [12:46] <nevcairiel> the idea behind it is pretty good, why pre-render audio into 8 channels based on some fixed speaker setup when you can just encode the audio sources directly and do the mixing at the users end based on their actual physical setup [12:46] <rcombs> wm4: mine only calibrated gain levels for each speaker with it [12:47] <rcombs> and maybe some sort of phase-related thing I'm not remembering the details of [12:47] <rcombs> I should probably recalibrate it [12:47] <cone-560> ffmpeg.git 03Diego Biurrun 07master:103391ca90b2: dca: Remove some commented-out cruft [12:47] <cone-560> ffmpeg.git 03Michael Niedermayer 07master:950ce21d4a7c: Merge commit '103391ca90b2f7c56ae756d76c76f7c3dfa28dd4' [12:47] <nevcairiel> the high end models try to adjust for the frequency response of the speakers as well [12:47] <cone-560> ffmpeg.git 03Michael Niedermayer 07master:61af6bebb457: swscale: Allow chroma samples to be above and to the left of luma samples [12:47] <cone-560> ffmpeg.git 03Michael Niedermayer 07master:e927682e1b25: avfilter/vf_scale: Allow chroma samples to be above and to the left of luma samples [12:47] <rcombs> the mic itself also functioned as a decent desk microphone with a really long mono 3.5mm cable [12:47] <nevcairiel> i would think with that long cable its prone to interference [12:48] <rcombs> probably :3 [12:48] <rcombs> feel free to yell at yamaha [12:48] <wm4> you can sell better cables separately to the audiophiles [12:48] <nevcairiel> for calibration is probably fine, as he calibration sounds are pretty loud and would probably drown out any interference :P [12:48] <rcombs> I'm not sure if a bit of EM interference would make a significant difference during calibrationyup [12:50] <nevcairiel> those HDMI devices are not really targeted at the audiophiles anyway, they use separate USB DACs with their own amplifiers after [12:50] <rcombs> that sounds dumb [12:51] <nevcairiel> its really the same thing, only split in two devices, and no stupid HDMI [12:51] <rcombs> meanwhile, I get users complaining about decoding AC3/DTS audio on their computers and sending to their receivers as PCM instead of passing it through on the HDMI cable [12:51] <wm4> rcombs: of course [12:51] <rcombs> because they think the receiver will somehow do a better job of decoding the audio than lavc, I guess? [12:52] <wm4> rcombs: welcome to a world of pain [12:52] <wm4> no [12:52] <wm4> because they can't send PCM over their shitty link [12:52] <nevcairiel> users are stupid like that, they need to see the light for bitstream decoding go on, or its not good sound [12:52] <nevcairiel> wm4: even if they can, they dont want to [12:52] <kierank> Dolby atmos is serious business [12:52] <wm4> or maybe that isn't an issue anymore with actual HDMI [12:52] <rcombs> wm4: HDMI can handle 8-channel PCM [12:52] <rcombs> wm4: that was an issue with S/PDIF over coax [12:52] <wm4> oh, good [12:53] <wm4> there still might be issues with speaker config etc. [12:53] <nevcairiel> the thing is, in my experience if you bitstream to the receiver, its typically a bit louder than sending PCM of the same audio [12:53] <nevcairiel> and louder is perceived as better by most people [12:53] <wm4> haha [12:53] <wm4> I wonder if that's an intentional marketing trick [12:53] <nevcairiel> no idea [12:53] <rcombs> of course, then there are the audio streams that have embedded dynamic range compression information [12:53] <wm4> but most likely it's just differences in remixing? [12:54] <nevcairiel> might be lavc ignoring stuff like dialnorm or dr [12:54] <rcombs> lavc parses dialnorm, though I'm not sure if it uses it [12:54] <kierank> Doesn't use it [12:54] <kierank> On some codecs DRC is applied [12:55] <rcombs> of course, dialnorm is just stream-global fixed gain, so& [12:55] <rcombs> that could indeed be why bitstreaming is sometimes louder [12:57] <rcombs> and from what I've heard, DRC meta on DVDs and BDs is often really poorly-executed [12:58] <rcombs> "<A> that will make it decode the TrueHD/DTS-MA streams to 7.1 LPCM. <B> yes but isnt that shit?" <-- actual (anonymized) conversation between two people who both supposedly knew what they were talking about [12:58] <wm4> there was a patch for DTS-MA [12:58] <wm4> but it's not bitexact yet, and now it appears to be bitrotting [12:58] <nevcairiel> on windows i hijacked a binary decoder for dts-hd ma to make those people happy [12:59] <JEEB> lol [12:59] <rcombs> (A actually did; B took a fair bit of explaining to understand that PCM is, in fact, lossless) [12:59] <wm4> <rcombs> of course, dialnorm is just stream-global fixed gain, so& <- could export that as replaygain data [12:59] <nevcairiel> we get our fair share of audiophools on the support forum, i have to bite my tongue way too often [12:59] <rcombs> wm4: huh, that could work [13:00] <nevcairiel> WAV sounds better than FLAC, etc [13:00] <wm4> rcombs: someone would just have to get the math right [13:00] <rcombs> I wonder, how's it exported now? [13:00] <wm4> which probably involves woodoo and secret constants [13:00] <wm4> voodoo even [13:00] <wm4> rcombs: side data [13:00] <rcombs> "Dialnorm is an integer value with range 1 to 31 corresponding to a playback gain of -30 to 0 dB (unity) respectively" [13:01] <rcombs> well that's simple enough [13:02] <wm4> h [13:02] <wm4> m [13:04] <cone-560> ffmpeg.git 03Thomas Volkert 07master:6821a5a4adcb: rtpenc: HEVC/H.265 support [13:05] <nevcairiel> but if its negative values, doesnt it make stuff quieter [13:05] <nevcairiel> or i'm missing some part of it :D [13:11] <rcombs> nevcairiel: I think it's meant to represent the average dialogue level in dBFS, so you'd apply (-dialnorm)dB of gain if you wanted dialogue level to match between multiple streams [13:16] <rcombs> 'Dialnorm is defined in ATSC A/85 as An AC-3 metadata parameter, numerically equal to the absolute value of the Dialog Level, carried in the AC-3 bit stream. This unsigned 5-bit code indicates how far the average Dialog Level is below 0 LKFS. Valid values are 1-31. (zero value is reserved) The dialnorm values of 1 to 31 are interpreted as -1 to -31LKFS.' [13:16] <wm4> holy wat https://github.com/git/git/pull/107 [13:16] <michaelni> Loriker, maintainers get OP in ffmpeg-devel and as you maintain rtpdec/enc_h261 and rtpdec_hevc, ... [13:16] <rcombs> "The loudness unit of LKFS is dB and is used the same way as a dB of gain. For example, a -15 LKFS program can be made to match the loudness of a -22 LKFS program by attenuating 7 dB." [13:17] <Loriker> okay, thx [13:18] <rcombs> wm4: wat [13:19] <wm4> I wonder why that repo has 15 PRs, even if the repo info says they accept no PRs [13:20] <wm4> and then I saw this [13:20] <wm4> oh well [13:26] <rcombs> because ~Linus~ [14:23] <ubitux> heh, found this in an old mkv http://b.pkh.me/touch-ssa.ssa [14:23] <ubitux> ("Command:" but not actually one) [14:23] <ubitux> also "JACOsub" :) [14:36] <wm4> ubitux: wat [14:36] <ubitux> ah and finally found a mkv sample with some Marked=<X> thing in it [14:37] <wm4> so it has a jacosub script embedded or so [14:37] <wm4> as comment events [14:37] <ubitux> wm4: no not really, it was probably converted and hackdjusted [14:37] <ubitux> mmh wait yeah [14:37] <wm4> so ass is inspired by jacosub? lol [14:37] <ubitux> indeed it looks like jacosub [14:38] <ubitux> wm4: yeah of course [14:38] <wm4> lesson of history [14:38] <ubitux> so yeah anyway there is some jacosub embedded in the comments for some reason [14:39] <ubitux> but with ass markup [14:39] <ubitux> i have no idea what's going on here [14:39] <cone-560> ffmpeg.git 03Hendrik Leppkes 07master:ff34b2d6d35b: mlpdec: support major sync headers with optional extension blocks [14:39] <cone-560> ffmpeg.git 03Hendrik Leppkes 07master:36bf549b2706: mlpdec: support TrueHD streams with an Atmos substream [14:39] <ubitux> i see some jacosub timestamps [14:40] <ubitux> oh whatever :) [14:40] <wm4> yeah [14:41] <ubitux> "Central Anime - Kansas - Fansubbing since 1992 -" [14:42] <ubitux> yeah right ok. [14:42] <nevcairiel> they started with vhs subbing [15:18] <cone-560> ffmpeg.git 03wm4 07master:6c7f1155bb64: avformat/mp3dec: avoid early EOF with concatenated gapless mp3s [15:18] <cone-560> ffmpeg.git 03Thomas Volkert 07master:dcdc1cbf4321: rtpdec_hevc: do not print an error message if the received packet has a valid header but lacks additional bytes as payload [17:15] <cone-560> ffmpeg.git 03Pascal Massimino 07release/2.4:b7f271995137: libavcodec/webp: treat out-of-bound palette index as translucent black [17:15] <cone-560> ffmpeg.git 03Gianluigi Tiesi 07release/2.4:3b57d7769a76: avcodec/libilbc: support for latest git of libilbc [18:22] <ubitux> hum [18:23] <ubitux> av_bprint_is_complete() should take a const parameter [18:23] <ubitux> is it safe to change that without a major bump? [18:24] <nevcairiel> personally i would say yes, adding a const to a parameter should be save, but i believe we tried that before and for some reason it was changed later to be version dependent and only activate on the bump [18:24] <nevcairiel> although i have no clue as to why [18:24] <ubitux> iirc last time it happened was because of cb pointers assignment or something along these lines [18:25] <ubitux> i doubt something using a cb system on top of bprint is set up somewhere [18:25] <ubitux> also, it's an inline function here [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:d210c0e777c1: avcodec/ass: add ff_ass_add_rect_bprint() helper [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:4c85073044a6: avcodec/jacosubdec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:ac95b436db73: avcodec/microdvddec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:e833b02f2fb5: avcodec/movtextdec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:27a9bee243c7: avcodec/mpl2dec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:8e7808b524d1: avcodec/realtextdec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:6a65da3a182c: avcodec/samidec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:592716227c5f: avcodec/srtdec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:947a5111dd58: avcodec/subviewerdec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:36c3a0167a32: avcodec/textdec: add some memory checks [18:46] <cone-560> ffmpeg.git 03Clément BSsch 07master:ce8dc93aefb4: avcodec/webvttdec: add some memory checks [18:50] <cone-560> ffmpeg.git 03Clément BSsch 07master:e60770679b56: avformat/assenc: return correct error code [18:58] <cone-560> ffmpeg.git 03Clément BSsch 07master:08e2b0da2ca9: avformat/assenc: mux all extradata at once [19:02] <BBB> so no vdd summary?/ [19:02] <BBB> :( [19:03] <ubitux> i suppose we'll get one after everyone is back home [19:04] <ubitux> i don't see beastd, compn, saste, nicolas, ... here on irc :p [19:08] <jamrial> libav sent their own summary already. which is basically "these are the things we'll remove, and these the things we will port from ffmpeg" :p [19:09] <BBB> well there was also an email about opw right? [19:09] <BBB> I was just looking at their list [19:09] <BBB> one wrote an unfinished rm demuxer; one wrote an unfinished asf demuxer; one made all kind of cosmetic patches; yay!" [19:10] <BBB> (I think thats what it said) [19:10] <wm4> oh you're so mean [19:10] <wm4> I can totally understand that something related to asf or rm can remain unfinished [19:10] <wm4> that they trained someone to make cosmetic patches is rather idiotic though [19:10] <BBB> I started the fork, remember? [19:10] <BBB> we had such great goals [19:10] <BBB> to see that these are the accomplishments of the current state of the fork [19:11] <BBB> ... [19:11] <wm4> look at it this way: michaelni can now maintain things [19:11] <wm4> but yeah if you can go back in time and unfork, maybe that's a good idea [19:12] <wm4> libavtimemachine [19:13] <ubitux> jamrial: porting is going to be a real pain for them if they actually try to do it [19:15] <iive> why are they going to remove get_buffer? [19:17] <wm4> iive: because it was replaced with get_buffer1 [19:17] <wm4> err [19:17] <wm4> get_buffer2 [19:17] Action: iive facepalm [19:18] <wm4> nothing to facepalm about [19:18] <wm4> get_buffer2 is the refcounted version, and it's awesome [19:19] <BBB> so what does get_buffer do that is not refcounted? [19:21] <wm4> it simply isn't refcounted [19:21] <wm4> it's for applications which use the old method [19:21] <wm4> they expect that the decoder calls release_buffer [19:23] <michaelni> get_buffer/release_buffer() left the buffers to the user application which could implement ref counting, get_buffer2() moves the ref counting into libavutil [19:23] <michaelni> dunno if any user apps did implement ref counting with get_buffer(), ffmpeg itself didnt [19:24] <BBB> I think chrome did [19:24] <wm4> mplayer [19:24] <wm4> vlc (now uses get_buffer2 I'm sure) [19:25] <wm4> yeah mplayer still uses get_buffer [19:25] <wm4> so I guess ffmpeg will keep it, hurr [19:28] <BBB> or you could send a patch to fix mplayer [19:29] <wm4> hahaha [19:29] <wm4> yeah that "should be easy" [19:30] <wm4> in my mplayer fork I've replaced that stuff with refcounting before even libavcodec had it [19:31] <iive> i'm not familiar with the new api changes. [19:31] <iive> let's say the app allocates 16 buffers and ffmpeg get_buffer2() alll. [19:32] <iive> how the program would know what buffer is free when get_buffer2() is called again? [19:32] <wm4> AVBuffer has a release callback [19:33] <wm4> but it's called asynchronously, and may happen in foreign threads [19:34] Action: iive doublefacepalm [19:35] <wm4> ? [19:38] <iive> so, basically the new api adds a flag and moves release to another structure... [19:39] <wm4> it's a saner design [19:40] <wm4> you could emulate refcounting int eh old api (ffmpeg.c did for libavfilter), but it was very complicated [19:42] <wm4> it was "fun" http://git.videolan.org/?p=ffmpeg.git;a=blob;f=cmdutils.c;h=27e01307d20d7b2… [19:44] <iive> i don't see the improvement... [19:44] <iive> i actually fail to see a difference. [19:45] <wm4> compared to this highly hacky code, ffmpeg.c/cmdutils.c now needs only 1 line of code to setup refcounting [20:47] <cone-560> ffmpeg.git 03Nicolas George 07master:eb7a6d0813ff: lavu/bprint: add const to av_bprint_is_complete() argument. [20:50] <ubitux> wtf burek [20:51] <burek> hmh? [20:52] <nevcairiel> if gcc crashes, report it to gcc :P [20:53] <burek> :) [20:53] <ubitux> it looks totally unrelated to ffmpeg [20:54] <ubitux> burek: cc just got oom killed because the compilation ate too much memory [20:54] <wm4> lolwut [20:54] <ubitux> there is nothing we can do about it [20:54] <burek> well ok, we can always delete the report, but it might be the issue with make file and dependencies not sorted out properly: http://stackoverflow.com/questions/1564195/gnu-makes-j-option [20:54] <nevcairiel> oom would be a decent explanation [20:54] <wm4> how much memory does it use? [20:54] <burek> could be, let me try with more mem [20:54] <burek> 1gb [20:58] <jamrial> dependencies not being sorted properly would give errors like missing objects (because they haven't been compiled yet) and such, but not internal compiler bug errors [20:58] <burek> you're probably right, i'll give it a try in a sec [21:00] <ubitux> burek: while you're here, could you fix fflogger so rcombs has his trac nickname (11rcombs) properly displayed? [21:00] <ubitux> (the number at the start of the nickname are altering the color displayed on irc) [21:01] <burek> sure [21:36] <burek> ubitux, can irc nicknames ever begin with a comma character? [21:37] <burek> -irc +trac [21:37] <ubitux> i don't know the range of characters [21:37] <ubitux> you might even be able to put utf-8 [21:38] <BtbN> Only plain ascii characters iirc [21:38] <burek> because there is no neutral background color code and i really would like to avoid specifying bckgnd color [21:38] <BtbN> I'm not even sure if it allows numbers in front [21:38] <burek> wouldn't [21:38] <burek> would.. man.. i need some sleep obviously [21:38] <ubitux> BtbN: trac nicknames, not irc [21:39] <burek> trac, yes, i didn't type correctly my question [21:39] <ubitux> burek: you just need to put a space after the color change tag [21:40] <burek> i prefixed the single-number color codes with 0, that should do it [21:43] <burek> i guess we'll have to wait for rcombs to do something :) [21:45] <ubitux> thx [21:45] <burek> :) :beer: [22:36] <ubitux> http://uselessd.darknedgy.net/ hehe [22:36] <ubitux> (yeah, sorry, unrelated to ffmpeg, but i heard ppl like systemd drama around here) [23:05] <jamrial> BBB, ubitux: some low hanging fruit i picked up https://github.com/jamrial/FFmpeg/commit/eb9e527ca4fe81ac47f51eef13a74c018d… [23:07] <BBB> is vbroadcasti128 slower than just extending the arrays? [23:07] <BBB> (for filter coeffs) [23:07] <BBB> it seems to me we should be able to keep using mova and just x2 the array [23:07] <BBB> (?) [23:08] <nevcairiel> ymm registers are weird sometimes [23:08] <jamrial> I guess extending the arrays would be faster, yeah. vbroadcasti128 ymm, m128 is afaik slower than a mova ymm, m256 [23:09] <BBB> ok so lets do that [23:09] <BBB> LOCAL_ALIGNED_16 -> 32, just make that a macro parameter [23:09] <BBB> i.e. only use LOCAL_ALIGNED_32 for avx2, keep using 16 for ssse3 [23:10] <jamrial> ok [23:10] <BBB> (since you already know the size in the macro, so that should be trivial) [23:10] <BBB> so what functions exist in yasm and which are the C wrappers? [23:11] <BBB> it looks like 16/32 (width) are raw yasm, and 64 is a c wrapper? [23:11] <jamrial> yeah, fox avx2 32 is yasm, 64 a wrapper. ssse3 is 16 yasm, 32/64 wrapper [23:11] <jamrial> there's no 16 avx2 [23:12] <BBB> can 16 be avx2 also? [23:12] <BBB> or does that suck because of the laning? [23:12] <jamrial> the separate lanes thingy from ymm registers would kill any gain [23:12] <jamrial> yeah [23:12] <BBB> hm.. ok [23:12] <BBB> thats all I have [23:12] <BBB> some numbers on speedup would be nice [23:12] <BBB> (in commit msg) [23:13] <jamrial> it was around 20% to 25% faster in my tests [23:13] <BBB> for mc functions right? [23:13] <BBB> so total gain is & 10% rougly then? [23:13] <BBB> roughly* [23:13] <jamrial> yeah. benching it was a PITA, btw. finding the calls to mc and makings sure to only check the 32/64 cases and such [23:13] <BBB> yes I remember that, its a pain [23:14] <BBB> (sorry I didnt make that easy, this way the code is sweet and small, but benching becomes kind of nightmarish) [23:14] <BBB> I do feel its worth it though, since we dont micro-bench very often so ... [23:15] <BBB> actually 10% doesnt make sense since it only affects part of the mc functions [23:15] <BBB> maybe 5-7% overall? [23:15] <BBB> (which is still cool btw) [23:15] <jamrial> no idea, didn't check overall decoding times or fps as reported by ffmpeg [23:16] <BBB> maybe nice to include in commit msg& but yeah rest is ok [23:18] <jamrial> the timer macros showed changes like 1784 cycles -> 1221 cycles on a single function (16k runs total on a 10 seconds 1080p sample) [23:18] <jamrial> So i don't really expect it will make that much of a difference overall :p [23:25] <BBB> Id test it, mc tends to be the biggest cpu consumer in video decoding [23:25] <BBB> along with coefficient decoding, I guess [23:26] <BBB> so if mc goes down by a significant bit, it should affect overall decoding time [23:26] <BBB> also depends on content ofc& that is, content with smoother motion fields is more likely to use the big blocks that profit from this [23:43] <cone-560> ffmpeg.git 03Michael Niedermayer 07release/2.4:d38943829649: swscale: Allow chroma samples to be above and to the left of luma samples [23:43] <cone-560> ffmpeg.git 03Michael Niedermayer 07release/2.4:bb5c0ac922ef: avfilter/vf_scale: Allow chroma samples to be above and to the left of luma samples [23:43] <cone-560> ffmpeg.git 03Michael Niedermayer 07release/2.4:e1ce4f805f31: update for 2.4.1 [00:00] --- Mon Sep 22 2014

1 0

ffmpeg.log.20140921
by burek 22 Sep '14

22 Sep '14

[02:23] <judu> Hi, I would like to know, in ffmpeg 2.4 what is supposed to replace the SWR_CH_MAX that's defined in libswresample/swresample.h? [02:26] <judu> In the "Macro Definition Documentation" section of http://ffmpeg.org/doxygen/2.4/group__lswr.html, the SWR_CH_MAX is documented, but in the file it's not defined anymore -> "#if LIBSWRESAMPLE_VERSION_MAJOR < 1" is false since 2.4. [05:54] <Mephistos> hi, quick question, I would like to merge two videos together, but concatenate the first one, the second video being a replacement for the credits, is this possible to do with ffmpeg? My first attempt failed. [05:54] <c_14> What do you mean "merge"? [05:55] <Mephistos> two input files, one output file, converting from lossless x264 to a vbr x264, with the credits part of the video replaced at the end (second video source). [05:56] <c_14> sure [05:56] <Mephistos> I have most of the command line, but when I did this, it didn't appear to replace the credits. [05:57] <c_14> ffmpeg -i 1 -i 2 -filter_complex '[0]trim=cred_start:cred_end[tmp];[tmp][1]concat=n=2:v=1:a=0[out]' -map '[out]' -c:v x264 -crf 18 out.mkv [05:57] <c_14> or something [05:58] <c_14> eh, no wait [05:58] <c_14> replace cred_start with 0 and cred_end with cred_start [05:58] <Mephistos> ffmpeg -i source1-1080p-09b-x264-lossless.mp4 -t 4:04 -i source2-1080p-09cF-x264-lossless.mp4 -c:v libx264 -crf 23 -tune film -acodec aac -strict experimental -b:a 256k -ac 2 -f mp4 -threads 3 merged-final-1080p-09d-ffmpeg-x264-`date +%Y%m%d%H%M%S`.mp4 [05:58] <Mephistos> That's my command line right now [05:59] <c_14> source2 is the credits or the to be cut one? [05:59] <Mephistos> the credits [05:59] <Mephistos> I want to cut it at 4:04 seconds exactly [05:59] <c_14> put the -t 4:04 in front of source1 not source2 [05:59] <Mephistos> ok [05:59] <c_14> and you're not actually concatting [06:00] <c_14> add -filter_complex '[0][1]concat=n=2:v=1:a=1[v][a]' -map '[a]' -map '[v]' [06:01] <Mephistos> I'm getting a tonne of buffer overflows [06:02] <c_14> probably not fatal [06:02] <c_14> audio related? [06:02] <Mephistos> no idea [06:02] <Mephistos> it says it's dropping frames [06:29] <Mephistos> I did it as a two step process copying the codecs to remove the credits at the end. [06:31] <Mephistos> and I'm using -f concat now, it's not giving me errors or dropping frames. [06:32] <c_14> If it works, it's good. [06:34] <Mephistos> I have to wait and see, but I think it will work no problem. [07:59] <goulard> What is the purpose of AAC/ADTS extradata? [08:00] <goulard> I can't find any info about it. A small chunk... usually 5 bytes of data in between each ADTS packet [10:02] <frickenate> sacarasc: Haha damn. I was going with AAC for HLS streaming. Quick search seems to indicate that mp3 + aac are both supported on all worthy platforms. [10:06] <frickenate> Ah then again, maybe not. Native html5 support varies for mp3 and aac. Of course. [10:23] <brontosaurusrex> frickenate: ffmpeg -codecs | grep mp3 < if you get the "D" then you have a decoder [10:24] <frickenate> Ooooh I like the D. [10:24] <frickenate> Still compiling for the first time.libavcodec takes a loooong time to compile :) [10:25] <brontosaurusrex> and a nice browser test page http://hpr.dogphilosophy.net/test/ [10:31] <brontosaurusrex> and I don't think firefox has any mp4/aac support, but don't quote me [10:35] <frickenate> I looked that up already. Firefox has AAC now except on OS X, which is coming next month with ff 34. [10:35] <brontosaurusrex> ok, I stand corrected [10:35] <frickenate> But yeah, still fragmented. All hail flash player fallback. [10:36] <brontosaurusrex> yeah, makes you wonder why bother with html5 if we have to flash fallback ... [10:37] <brontosaurusrex> so ff 34 will have that cisco decoders working ... ? [10:38] <frickenate> ooh fancy: http://areweplayingyet.org/ [12:09] <JazzCZ> Hey [12:10] <JazzCZ> where can I find the TrueHD patch? [12:23] <JazzCZ> anyone knows how to apply truehd patch? [12:31] <sara> hello [12:32] <sara> I have a quick q on ffmpeg, I keep getting "no such file or directory" when trying to encode from flv to mp4 [12:32] <sara> any help? :) [12:36] <sacarasc> sara: If the file name has spaces in them, you'll have to either escape it or put quotes around it. [12:37] <sacarasc> Or, the file doesn't exist in the directory you're calling ffmpeg from. [12:37] <sara> I psted the file in the same directory as the ffmpeg executable files [12:37] <sara> lemme try the quote thingy, [12:38] <sara> nope, nothing [12:39] <sara> this is the command I'm using: ffmpeg -i sample sample.mp4 [12:39] <sara> I've tried sample.flv too [12:39] <sara> but nothing [12:39] <sara> as my input i mean [12:40] <sacarasc> Is there a file called "sample" in the directory? [12:40] <sacarasc> Or sample.flv? [12:40] <sara> yes [12:40] <sara> the file i pasted in there [12:40] <sara> named merely sample [12:40] <sara> I pasted it in the "bin" folder [12:41] <sara> where the 3 ffmpeg executable files are found [12:41] <sara> when i downloaded ffmpeg [12:41] <sacarasc> And you're in that directory trying to convert? [12:42] <sara> yes [12:42] <sara> (i think) [12:43] <sacarasc> Which OS are you on? [12:43] <sara> so the file is in the ffmpeg directory (under my c drive( [12:43] <sara> and the command line i'm putting includes the ffmpeg directory [12:43] <sara> is that right>? [12:43] <sara> if I'm inserting as my command line [12:43] <sara> ffmpeg -i sample sample.mp4 [12:43] <sara> windows# [12:43] <sara> windows 7 [12:45] <sacarasc> You're in a directory similar to C:\ffmpeg\bin ? [12:45] <sacarasc> If do you the command "dir", do you see ffmpeg.exe and the file sample? [12:45] <sara> that's correct [12:45] <sara> oops [12:45] <sara> lemme check [12:46] <sara> so merely type in the command line... [12:46] <sara> dir ffmpeg? [12:47] <sara> if i put dir simple [12:47] <sara> *simpoly [12:47] <sara> i don't see ffmpeg [12:47] <sara> but i can my sample file [12:48] <sara> ah, so I need to actually copy pasted the ffmpeg files into this directory? [12:48] <sacarasc> Okay, change your command to something like C:\ffmpeg\bin\ffmpeg.exe -i sample sample.mp4 [12:48] <sacarasc> Replacing the "C:\ffmpeg\bin\" part with the directory that ffmpeg.exe is in. [12:51] <sara> but ffmpeg.exe [12:51] <sara> is in C:\ffmpeg\bin already [12:51] <sara> though my executable file doesn't have the .exe postfix, is that normal? [12:52] <sara> so it's such ffmpeg in the bin folder [12:52] <sacarasc> Where did you download it from? [12:53] <sara> http://ffmpeg.zeranoe.com/builds/ [12:53] <sara> i download the static version for my system [12:54] <sacarasc> Okay. And on the command line "dir" showed the file "sample"? Not "sample.flv"? [12:54] <sara> in dir it says "sample.flv [12:55] <sacarasc> Okay, type this: C:\ffmpeg\bin\ffmpeg.exe -i sample.flv sample.mp4 [12:55] <sacarasc> That should work. [12:59] <sara> nope.... [12:59] <sara> but [12:59] <sara> I just rerun ffmpeg [12:59] <sara> i mean I just re-did [12:59] <sara> dir [12:59] <sara> and ffmpeg is there [13:00] <sara> I copy-pasted the entire ffmpeg folder into my home directory [13:01] <sara> op! [13:01] <sara> seems to be working [13:02] <sara> ,,,,converting.... [13:02] <sara> is there a way to automate this for a batch process? [13:02] <sara> or would I need to do the same thing x 100 for a 100 videos? [13:03] <sacarasc> If you know how to batch script, you can do it. I don't know how, so can't help there. [13:06] <sara> okey dokey [13:06] <sara> thank you! [13:39] <judu> The wI would like to know, in ffmpeg 2.4 what is supposed to replace the SWR_CH_MAX that's defined in libswresample/swresample.h? [13:39] <judu> In the "Macro Definition Documentation" section of http://ffmpeg.org/doxygen/2.4/group__lswr.html, the SWR_CH_MAX is documented, but in the file it's not defined anymore -> "#if LIBSWRESAMPLE_VERSION_MAJOR < 1" is false since 2.4. [14:31] <kaotiko> hi [14:34] <hatsch> hi everybody [14:35] <Sokolio> Hi [14:36] <hatsch> i am trying to use the new icecast protocol in ffmpeg to send vpx video and vorbis audio in a webm container to an icecast server. from the ffmpeg side of view everythinks seems to be okay, i get connected to the server and the input is processing. but receiving the video with eg vlc gives an error about a wrong header ( mp3) http://pastebin.com/WGYLxZHS [14:37] <hatsch> don't know for sure if it's an ffmpeg or icecast issue. but sending webm to icecast with gstreamer is working. [15:57] <sara> hi [15:57] <sara> is it possible to convert and compress a video file at the same time in ffmpeg? [15:58] <sara> e.g. convert a mp4 to flv and compress from 100MB to a smaller size (ideally with minimum loss of quality) [15:58] <sara> i know how the command line to convert [15:58] <Suchiman> usually transcoding (convert) compresses [15:58] <sara> could you give me the command line for compressing+converting at the same time? [15:59] <Suchiman> though you can adjust Quality / Bitrate to Archive compression [15:59] <brontosaurusrex> sara: flv and mp4 are containers [15:59] <Suchiman> but compression is archived through reducing quality [15:59] <sara> how can I adjust the quality/bitrate? [16:00] <sara> is the bitrate value _b:v in the command line? [16:00] <sara> woops sorry, just saw your comm on containers [16:01] <sara> you lost me a bit... [16:01] <brontosaurusrex> meaning flv can contain various codecs [16:01] <brontosaurusrex> perhaps read this http://en.wikipedia.org/wiki/Flash_Video [16:02] <brontosaurusrex> or to refraze; different compression formats [16:02] <sara> I try the command line ffmpeg -i sample.mp4 -ar 22050 sample.flv [16:02] <Suchiman> sara: yeah b:v sets Bitrate but depending on codec there are various Parameters to tweak [16:02] <sara> ok [16:02] <sara> and is the codec dependent on container? [16:03] <sara> or could i use h.264 for all all containers? [16:03] <brontosaurusrex> you could use h.264 for both mp4 and flv [16:03] <Suchiman> sara: Containers can contain from 1 up to almost any codec [16:03] <Suchiman> depends on the container [16:03] <sara> aha [16:04] <Suchiman> sara: the Player at the other end must support both, Container and codec [16:06] <sara> of course, makes sense [16:06] <sara> :) [16:07] <brontosaurusrex> anyway, why would you need flv? [16:45] <sruli> hi all, i hope someone can help me. I am trying to create a video from 3 clips, and an overlay on 1, only 1 of the clips have audio, but i get an error, everytime, please see http://pastebin.com/aCjmQnw5 for command i use and error [17:16] <Lac3rat3d> Does ffmpeg support decoding DTS-ES? I found some threads about supporting it, with patches etc, but when I try to convert a file with DTS-ES to AAC native, I get an error message [17:20] <Lac3rat3d> http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2010-June/091970.html http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2014-April/156698.html http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2010-June/091970.html [18:44] <bearish> hey guys, i'm getting real bad performance with libvpx compared to libx264, any pointers? [18:46] <Nosomy|off> only performance? [18:48] <bearish> So i have this H264 video piping in to ffmpeg [18:49] <bearish> and i'm trying to get it working on a chromecast [18:49] <bearish> which is picky about it's formats [18:49] <bearish> i'm using ffserver to restream [18:53] <bearish> So I can't use MPEG-TS container [18:53] <bearish> can't use mp4, muxer does not support non seekable output [19:00] <hatsch> bearish: did you set the -threads option (it seems 0/auto is not working with libvpx ) [19:03] <bearish> it doesn't seem to work either way, only 1 core gets utilized [19:04] <bearish> and i get 9-15 fps [19:04] <bearish> where on half a core libx264 produces ~30-35fps [19:04] <JEEB> libvpx only has threading with vp8 [19:04] <JEEB> not vp9 [19:04] <theholyduck> theres still no threading on vp9? [19:05] <JEEB> of course not [19:05] <JEEB> google doesn't need it [19:05] <theholyduck> JEEB, well, there wasnt any threading on it like, many months ago [19:05] <JEEB> they just do GOP-based encoding and stitch shit together [19:05] <theholyduck> when i last tested it [19:05] <theholyduck> so i just asumed they would add it at some point [19:05] <theholyduck> to make it not painfully slow [19:05] <JEEB> lol [19:05] <theholyduck> i guess i was wrong [19:07] <theholyduck> also 9-15fps is a lot better than what i got with, libvpx last time around, though, i asked them for some high quality settings to use. [19:07] <theholyduck> it took 13 hours to encode 2000 frames [19:07] <theholyduck> on my poor i7 [19:07] <JEEB> yeah, compression-side settings are lolslow [19:07] <JEEB> but I guess libvpx can get some speed [19:07] <JEEB> (even with vp9) [19:08] <JEEB> vp8 should be kind of faster, and have threading [19:08] <theholyduck> well, who knows, they might have implemented some optimizations, last time i tried using vp9 was right ater it was "released" [19:08] <theholyduck> might not have any threading, but they could atleast write faster code [19:09] <ubitux> am i reading that libvpx is now usable for vp9 encoding? [19:09] <theholyduck> ubitux, i think usable is in the eye of the beholder [19:09] <theholyduck> ubitux, its slow and doesnt offer massive quality gains over x264 [19:10] <theholyduck> so, in my mind, theres no reason to use it at all. [19:10] <ubitux> no, no one except google can use libvpx [19:10] <ubitux> several days to encode a few thousands frames is not usable [19:10] <ubitux> :p [19:10] <theholyduck> ubitux, well, you could reduce the settings so that it makes worse quality than x264 [19:10] <theholyduck> and it wont take several days [19:11] <theholyduck> ;P [19:11] <ubitux> will it be faster than x264? [19:11] <theholyduck> i havent tested it in forever, but seeing as its still not threaded [19:11] <theholyduck> id say no [19:20] <Lac3rat3d> Does ffmpeg support decoding DTS-ES? I found some threads about supporting it, with patches etc, but when I try to convert a file with DTS-ES to AAC native, I get an error message [19:54] <Lac3rat3d> no one? [20:14] <hatsch> anyone got webm streaming to icecast2.4 working? [20:19] <brian_ch> Hi, sorry if it's a dumb question, but are the FF_API_ defines really only for future reference, or it can happen that in some cases they are used otherwise? Is it safe to assume that all code between a FF_API_ ifdef will be compiled in that specific version of ffmpeg? Or the old defines are not always removed right away after the version bump? [20:36] <brian_ch> I ask because I was taking a look at the Pascal translation of the headers, and was wondering if it would be ok to drop these defines, or if instead of simplifying this would bring even more problems. [21:05] <frog_> hi. Some where i read that Ubuntu is changing from ffmpeg to avconv. Is this right? [21:05] <JEEB> quite some time ago they switched from ffmpeg to libav (when libav forked) [21:06] <JEEB> so if you want an up-to-date binary you use avconv on debian-based systems [21:06] <JEEB> (or you install an up-to-date ffmpeg otherwise) [21:06] <frog_> ok. thanks [21:07] <Lac3rat3d> Does ffmpeg support decoding DTS-ES? I found some threads about supporting it, with patches etc, but when I try to convert a file with DTS-ES to AAC native, I get an error message [21:26] <gcl5cp> how can i remove cover art in a M4A? [21:27] <gcl5cp> COVER ART --> Stream #0:1: Video: mjpeg, yuvj420p, 600x600 [SAR 72:72 DAR 1:1], 90k tbr, 90k tbn, 90k tbc [21:33] <gcl5cp> ffmpeg -i original.m4a -map 0:0 -c:a copy non-cover.m4a. do nothing [21:33] <c_14> -vn [21:33] <c_14> probably [21:33] <gcl5cp> do nothing [21:35] <gcl5cp> in ID3Tag editors (easytag puddletag picard) i can't see the cover. [21:36] <gcl5cp> i just need to replace to the correct cover. [21:36] <gcl5cp> ffmpeg -i original.m4a -vn -map 0:0 -c:a copy non-cover.m4a. do nothing [21:43] <gcl5cp> Stream #0:1 seems to be inside Metadata [21:44] <gcl5cp> how can i erase metadata? [21:45] <c_14> -map_metadata -0 or something [21:51] <gcl5cp> only with this works. "-map_metadata -1 -vn -c:a copy" [21:52] <gcl5cp> thank c_14 [21:56] <bearish> What am I doing wrong? All this hls stuff is hurting my brain [21:56] <bearish> http://paste.pound-python.org/show/IBdwS5NzYBch44o8XjnH/ [22:08] <eago_> hello [22:09] <EvolE> eago_: hi [22:09] <eago_> I'm trying to use ffmpeg to stream a live video from a browser [22:10] <eago_> using the webcam [22:11] <eago_> was planning to use webrtc and node.js [22:11] <eago_> but I'm having problems finding docuemtnation on how to do it [22:13] <EvolE> so you gonna get video stream to your nodejs server and then you will direct it to ffmpeg? [22:14] <eago_> yes [22:14] <eago_> or is there a simpler way? [22:15] <EvolE> what you plan to do next with that video after ffmpeg? [22:15] <eago_> stream it using ffserver [22:15] <eago_> basically on user will enter the website and broadcast ysing the web cam [22:15] <eago_> then other users can see that video [22:16] <EvolE> k, now i got it [22:24] <EvolE> eago_: did you see this? https://github.com/muaz-khan/WebRTC-Experiment/tree/master/RecordRTC/Record… [22:25] <eago_> yes.... but that seems to record many small files [22:25] <EvolE> eago_: but there video goes to node like a chunk. not sure if it can be modified to make endless recording and endless chunk sending [22:25] <eago_> instead of a singel stream [22:26] <eago_> ok.... so I call ffmpeg to use that file as input [22:27] <eago_> or can I connect it directly to the data sent by node.js? [22:28] <EvolE> eago_: seems like this is better to modify https://github.com/muaz-khan/WebRTC-Experiment/tree/master/RecordRTC/Record… [22:28] <eago_> ah, yup [22:28] <EvolE> you can make ffmpeg read from pipe. probably there should be something for nodejs also [22:31] <eago_> those look good for first recording and then serving... but not sure how to use it for live constant stream [23:19] <zenny1> Hi, when overlayed two videos side-by-side, the video on right disappears after a few seconds. I am using 'ffmpeg -i Audio.wav -i Rightvideo.mp4 -vf "[in] scale=iw/2:ih/2, pad=2*iw:ih [left]; movie=LeftVideo.mp4, scale=iw/2:ih/2, fade=out:300:30:alpha=1 [right]; [left][right] overlay=main_w/2:0 [out]" -b:v 768k final_output.mp4'. Any input on why the right video disappears after a few seconds? [23:44] <zenny1> I also tried with 'ffmpeg -i Audio.mp2 -i Rightvideo.mp4 -vf "[in] scale=iw/2:ih/2, pad=2*iw:ih [left]; movie=LeftVideo.mp4, scale=iw/2:ih/2, fade=out:300:30:alpha=1 [right]; [left][right] overlay=main_w/2:0 [out]" -b:v 768k final_output.mpg', still the right video vanishes after a few seconds!!! What am I doing wrong?!! [23:55] <benlieb> I'm trying to clean up the audio in a video. I've separated the audio to clean it up externally, but when I try to recombine with the video (with old audio removed) it doesn't play the sound . [23:55] <benlieb> ffmpeg -i videomute.mp4 -i audio.mp3 -c copy found1.mp4 [23:56] <benlieb> The audio is in the file according to ffprobe [00:00] --- Mon Sep 22 2014

1 0

ffmpeg-devel.log.20140920
by burek 21 Sep '14

21 Sep '14

[00:41] <cone-232> ffmpeg.git 03Michael Niedermayer 07master:5c073bbb57b5: avcodec/dvdsubdec: fix VD/SD identifier name [02:52] <cone-232> ffmpeg.git 03James Almer 07master:33c752be513d: x86/me_cmp: port mmxext vsad functions to yasm [10:04] <wm4> michaelni: I just spent an hour wasting my time and found it's because of RELATIVE_TS_BASE; why does Libav not have/need to ridiculous shit? [10:09] <ubitux> ad53c7f9ece7f95cd9538775fb6fd7a17eb1d301 ? [10:10] <wm4> fucked up [10:16] <Case> libav is a piece of crap that can't do simple things right. That's why [10:18] <ubitux> Case: please tone down a bit... [10:19] <ubitux> wm4: is there a bug with this? [10:19] <wm4> probably, but none which I know or observed [10:19] <Case> I was very conservative with my comment. See wm4's [10:19] <nevcairiel> he insulted the code of this project, not the other project, there is a big difference [10:20] <wm4> I just tried to use pkt->pts somewhere, and was naively expecting it would contain the pts [10:20] <wm4> instead of randomly mangled pts [12:18] <cone-281> ffmpeg.git 03Pascal Massimino 07master:7ac6b8cfa7e6: avfilter/idet: typo fix: PROGRSSIVE -> PROGRESSIVE [13:27] <wm4> how do I rsync samples for certain fate tests automatically? [14:00] <J_Darnley> wm4: i believe you can specify a subdirectory in the "source" argument to rsync [14:00] <J_Darnley> ... if that's what you mean? [14:04] <wm4> I think ubitux had a command that automatically invokes rsync for a specific test [14:18] <J_Darnley> Ah, I don't know that. [14:18] <J_Darnley> It sounds like a makefile rule though. [14:34] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:051ac5c0f51c: mpegvideo: Use the current_picture pts [14:34] <cone-281> ffmpeg.git 03Anton Khirnov 07release/2.2:3ac0638d573f: mpegenc: limit the maximum muxrate [14:35] <cone-281> ffmpeg.git 03Anton Khirnov 07release/2.2:7bc37641e3e6: avconv: fix the muxrate values for -target [14:35] <cone-281> ffmpeg.git 03Anton Khirnov 07release/2.2:f7395926f204: avconv: fix parsing the AVOptions for -target [14:35] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:58b5b062b84b: Merge commit '051ac5c0f51c119b33a57f3e137d7344eb1c2b26' into release/2.2 [14:35] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:1a642b72172d: Merge commit '3ac0638d573fc483ba6be3444858b26711c5d67d' into release/2.2 [14:35] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:516ba41f05c8: Merge commit '7bc37641e3e6c24d472ae06fcbecaba4c863829b' into release/2.2 [14:35] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:4b6e46c4277e: Merge commit 'f7395926f204051af9ad459a6d876b96ee6179ee' into release/2.2 [14:35] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:481118615c15: ffmpeg_opt: reduce diff to libav in opt_target() by 16 lines [14:46] <cone-281> ffmpeg.git 03Luca Barbato 07release/2.2:9fcc632249be: pulse: Add a wallclock option to be compatible with other other captures [14:46] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:e940d15a98cb: Merge commit '9fcc632249be3080836a3afce25b3092939743ac' into release/2.2 [14:54] <cone-281> ffmpeg.git 03Diego Biurrun 07release/2.2:1a7d1793d6d2: license: Mention that vf_interlace is GPL, not LGPL [14:54] <cone-281> ffmpeg.git 03Diego Biurrun 07release/2.2:40c7613ecf2f: doc: Fix syntax and logical errors in avconv stream combination example [14:54] <cone-281> ffmpeg.git 03Diego Biurrun 07release/2.2:f2abf8df7a08: Prepare for 10.5 release [14:54] <cone-281> ffmpeg.git 03Diego Biurrun 07release/2.2:d0af7d5745f3: Update Changelog for v10.5 [14:54] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:da1dfea07670: Merge commit '1a7d1793d6d2de4201c2d9fc7056e7c9e202c5fe' into release/2.2 [14:54] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:3bc5aa65bbcf: Merge commit '40c7613ecf2f3f7565309f9780183228605c3684' into release/2.2 [14:54] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:1aab0609968e: Merge commit 'd0af7d5745f3e228293633faa9e57994f3308c31' into release/2.2 [15:15] <kierank> please let me know when the "discussion room" is over [15:17] <wm4> kierank: are there any ffmpeg folks at VDD? [15:17] <kierank> yes [15:17] <wm4> good, who? [15:19] <kierank> stefano, nicolas i saw [15:42] <funman> reimar, thilo [15:57] <ubitux> wm4: make fate-rsync, ^C when the rsync command appears, and adjust to your needs [16:07] <wm4> hm ok seems there's no automagic then [17:33] <ubitux> wm4: the total is 876M [17:33] <ubitux> it's too much? [17:46] <cone-281> ffmpeg.git 03Katerina Barone-Adesi 07release/2.2:d14696c99cca: apetag: Fix APE tag size check [17:46] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:6dd718e416c0: Merge commit 'd14696c99ccac12a052ce10e70859ffc0293ed6a' into release/2.2 [18:13] <ubitux> Dialogue: 0,0:03:59.70,0:04:04.95,Main,,0000,0000,0000,,Umm... I'm just able to move objects without touching them, but... [18:13] <ubitux> Dialogue: 0,0:04:06.10,0:04:08.74,Main",,0259,0000,0000,, Can we stop now?\N No way. [18:13] <ubitux> Dialogue: 0,0:04:08.74,0:04:12.53,Main,,0000,0000,0000,,I want to make those powers a secret between us. [18:13] Action: ubitux wonders what's going on here [18:14] <ubitux> >Main" [18:15] <ubitux> ah, it's actually part of the style name [18:16] <ubitux> how evil [18:21] <ubitux> michaelni: you probably don't remember, but in the ass muxer, you stop writing the extradata header after the first line following [Event] [18:21] <ubitux> and then write the trailing part at the end [18:21] <ubitux> AFAICT these are generally Comments: [18:22] <ubitux> OTOH mkvextract just dump the whole extradata on top [18:22] <ubitux> why this extra complexity? [18:22] <ubitux> were you afraid of junk or something? [18:24] <ubitux> http://lucy.pkh.me/ass-mkvextract-ffmpeg.html [18:25] <michaelni> ubitux, i dont remember [18:25] <ubitux> ok [18:25] <ubitux> do you want to keep the current behaviour? [18:25] <michaelni> ubitux, you know ass much better than i do [18:41] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:1f8e0f7e06af: avcodec/proresenc_kostya: set initial max_slice_size based on frame_size_upper_bound [18:41] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:1a6218954a35: avcodec/proresenc_kostya: allocate 1 slice more to avoid triggering the reallocation warning when the used space is actually less than the allocated [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:f93f739ecac0: avcodec/utils: add GBRP16 to avcodec_align_dimensions2() [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:8d10d6e12738: avcodec/snow: check coeffs for validity [18:42] <cone-281> ffmpeg.git 03wm4 07release/2.2:f1685bd31a19: oggdec: fix invalid free on error [18:42] <cone-281> ffmpeg.git 03Mark Harris 07release/2.2:85b829bff9c3: doc/filters.texi: fix filter name in examples [18:42] <cone-281> ffmpeg.git 03Mika Raento 07release/2.2:0484d7ad7ef7: segment: don't access outside seg->frames array [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:b3f30cb6d63d: avformat/swfdec: Use side data to communicate w/h changes to the decoder [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:b4d2888ce8d1: avformat/swfdec: Do not change the pixel format [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:9e1ce9a8eee1: avcodec/h264: Allow partial escaping [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:9406d3c910bf: avcodec/mpegvideo: Use "goto fail" for all error paths in ff_mpv_common_frame_size_change() [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:fd6230e8f0d2: avcodec/mpegvideo: check that the context is initialized in ff_mpv_common_frame_size_change() [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:d19b55649c55: avcodec/mpegvideo: Set err on failure in ff_mpv_common_frame_size_change() [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:56fb830c302b: avformat/m4vdec: Check for non startcode 00 00 00 sequences in probe [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:252356cf06a4: tools/crypto_bench: fix build when AV_READ_TIME is unavailable [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:6c66ea5e73f2: avcodec/x86/vp9lpf: Always include x86util.asm [18:42] <cone-281> ffmpeg.git 03Pascal Massimino 07release/2.2:99af97ea112d: libavcodec/webp: treat out-of-bound palette index as translucent black [18:42] <cone-281> ffmpeg.git 03James Almer 07release/2.2:16b5df17ea9d: x86/dsputil: add emms to ff_scalarproduct_int16_mmxext() [18:42] <cone-281> ffmpeg.git 03Christophe Gisquet 07release/2.2:dd9b24a48808: wavpack: report if there is no bits left [18:42] <cone-281> ffmpeg.git 03Christophe Gisquet 07release/2.2:da1a8191c524: wavpackenc: proper buffer allocation [18:42] <cone-281> ffmpeg.git 03Michael Niedermayer 07release/2.2:014dee89d00f: avcodec: fix aac/ac3 parser bitstream buffer size [19:42] <ubitux> it's fun that the matroska specs doesn't say to mux [Events] [19:42] <ubitux> but everyone does [19:42] <ubitux> maybe someone should fix that specs [19:42] <ubitux> it's all wrong :D [19:43] <Rodeo_> ubitux: what's that? [19:44] <ubitux> what's what? [19:44] <ubitux> a spec? [19:46] <Rodeo> no, [Events] [19:47] <ubitux> the beginning of the dialogue events [19:47] <ubitux> after [Events] you have a Format: line [19:47] <ubitux> and then all the "Dialogue:" entries (or "Comment:", or even "Command:" or whatever) [19:48] <Rodeo> oh, I failed to read my buffer [19:48] <ubitux> http://www.matroska.org/technical/specs/subtitles/ssa.html it basically says the CodecPrivate shouldn't contain the [Events] and "Format:..." [19:49] <Rodeo> I wonder if we do it wrong too [19:49] <ubitux> but every muxer actually does it; and it's important if you want to store the Comments or whatever [19:49] <ubitux> well if you don't, shit breaks [19:49] <Rodeo> OK [19:50] <Rodeo> yeah, then the spec is wrong and you're right [19:51] <ubitux> there is also a mismatch about the readorder [19:51] <ubitux> (starting at 1 instead of 0) [19:53] <ubitux> i'd be curious if mkvextract actually reconstruct the [Events] header properly when it's missing from CodecPrivate [19:56] <ubitux> mmh interesting, actually it does [19:57] <ubitux> but it completely breaks the Comment: thing [19:57] <ubitux> (at least if they're part of the CodecPrivate [19:57] <ubitux> well, maybe i should do something about it in FFmpeg [20:19] <cone-281> ffmpeg.git 03wm4 07master:d87fe2687fdc: avformat/mp3dec: fix gapless audio support [21:00] <cone-281> ffmpeg.git 03Christophe Gisquet 07master:a0844935354b: alacenc: fix incorrect buffer use [21:00] <cone-281> ffmpeg.git 03Christophe Gisquet 07master:9950073655f5: alacenc: remove unneeded masking [21:22] <nevcairiel> why is audio such a PITA and all the codecs absolutely closed and secret, while for video we have open standards and everything <.< [21:24] <ubitux> it looks mostly the same to me [21:25] <nevcairiel> well i can read specs for h.264, h.265, mpeg2, even vc-1 [21:25] <nevcairiel> i have no decent specs for any dolby audio formats :d [21:25] <ubitux> all of those are open now? [21:26] <ubitux> and aac is (still?) closed? [21:26] <nevcairiel> i havent really tried to find aac [21:27] <iive> nevcairiel: ac3, a51, a52 should be quite open. [21:28] <nevcairiel> a52 and ac3 are the same thing [21:28] <iive> i think a52 is (minor) extension of ac3 [21:54] <nevcairiel> too bad i dont have a copy of the hex-rays decompiler, reading asm is tiresome [22:06] <ubitux> there are so much things to do with text subtitles... [22:06] <ubitux> but i kind of see the end of the tunnel [22:07] <ubitux> i'm pretty sure the text subtitles will be kind of sane after all of this [22:08] <ubitux> wm4: what's the state of the utf16 auto probing? [22:32] <ubitux> what's up with the sup demuxer btw? [00:00] --- Sun Sep 21 2014

1 0

ffmpeg.log.20140920
by burek 21 Sep '14

21 Sep '14

[00:28] <goulard> Sokolio: as it turns out I was doing something very dumb [00:28] <goulard> over writting part of my packet [00:28] <goulard> so as it turns out... its ok to have one ADTS packet per pes [00:29] <Sokolio> Ah it always comes down to this, believe me [00:29] <Sokolio> Good luck [00:30] <goulard> lol [00:30] <goulard> thanks for the help [00:30] <Sokolio> I was more a rubber ducky [00:30] <Sokolio> than actual help, but thanks [00:30] <goulard> haha [11:57] <grepwood> How can I compute a deprecated variable AVCODEC_MAX_AUDIO_FRAME_SIZE? [12:04] <santa1> Hi, trying to capture video from logitech c920 camera using 'ffmpeg -r 24 -s 1920x1080 -f v4l2 -vcodec h264 -i /dev/video1 -copyinkf -vcodec copy output.mp4' which captures well, but playback of captured video is faster (like in cartoons). Any input to overcome the faster playback? [12:05] <grepwood> santa1, your framerate is spoofed? [12:06] <santa1> grepwood: what do you mean? [12:06] <santa1> It is pal standard 24fps, right? [12:06] <grepwood> if your video is captured at X frames per second, and then you write in the stream header that the framerate is 2X, then your player will play it at 2X rather than X [12:07] <grepwood> maybe you need to capture with -r 12? [12:07] <santa1> grepwood: in that case how can one capture either to comply with NTSC or PAL standard? [12:08] <grepwood> you can if your hardware supports it and it was configured for it [12:08] <grepwood> I'm not exactly an expert on cameras, I don't know how would you accomplish that [12:09] <grepwood> sorry :( [12:09] <santa1> grepwood: Thanks for the useful input. Appreciate that! :D [12:10] <grepwood> np :) [12:12] <santa1> grepwood: Do one has to have the screecapture at the same rate as the video to synchronize? [12:12] <santa1> I have specified -r 1 for the screencapture and -r 24 for the video. [12:13] <grepwood> that could end up funny :p [12:14] <santa1> grepwood: What I noticed is the same captured video from webcam is faster in mplayer, but normal when played in vlc! hmmm! [12:15] <grepwood> plot thickens [12:19] <relaxed> pal is 25 fps [12:20] <relaxed> santa1: try with "-re" before the input [12:25] <relaxed> er, scratch that [12:27] <Animedude5555> Anybody here? I have a question about dithering in FFMPEG. [12:28] <Animedude5555> How does arithmetic dithering work? There's bayer dithering and error diffusion, but the first time I have ever heard of aritmetic dithering was here in FFMPEG. [12:29] <Animedude5555> Can somebody explain the algorithm to me? I can't find anything about "arithmetic dithering" on Wikipedia, or any other website at all whatsoever. Did you guys just recently invent the algorithm or something? Please tell me how it works. I've been looking for a decent dithering algorithm to use in a graphics program I'm writing, and this looks like it might be the answer to my problem. [12:30] <Animedude5555> There's two forms, an "addition" and an "xor" form. I'd love to have somebody here tell me just how this works. [12:30] <relaxed> Animedude5555: maybe "man ffmpeg-scaler" will give you some hints [12:31] <santa1> relaxed: Thanks, -re before the input didn't change anything, not -bf values (tried with 0 to disable as well as -1 for auto) :-( [12:32] <Animedude5555> What's "man ffmpeg-scaler"? If that is a document included only in the Linux version then it's useless to me, because I don't have that. I have the Windows version. [12:32] <relaxed> Animedude5555: https://www.ffmpeg.org/ffmpeg-scaler.html [12:33] <Animedude5555> a_dither says "arithmetic dither, based using addition". That's not at all helpful. What the heck is "arithmetic dithering"? [12:33] <relaxed> santa1: pastebin the output of ffmpeg -i on your output.mp4 [12:34] <Animedude5555> Still, I don't know what "arithmetic dither" is. Can somebody here tell me? Or is it some sort of secret? [12:35] <Animedude5555> Maybe FFMPEG trade secret? [12:36] <Animedude5555> Anybody? Anybody know what "aritmetic dither" is? <Relaxed> are you still here? Hello? [12:36] <JEEB> just go read the fine source code :) [12:36] <JEEB> or find the author and try to pry that info from him :P [12:38] <Animedude5555> It should be well documented in the software's online documentation, for those like me who may wish to implement it in their own software, especially since it is NOT DOCUMENTED ANYWHERE ON THE NET AT ALL (not even on the all-knowing Wikipedia). This leads me to think it is an invention of those working on FFMPEG, and a relatively recent invention too. But seriously SOMEBODY should have... [12:38] <Animedude5555> ...reverse engineered it by now and posted it up online in some kind of unofficial documentation (at the very least). [12:39] <JEEB> there's no need to reverse engineer anything [12:39] <JEEB> it's right there in the source code, this is not proprietary software [12:39] <JEEB> there might or might not be comments, too! [12:39] <JEEB> I do agree that more documentation is a good thing, though [12:40] <Animedude5555> Why is it not included on this wikipedia page? https://en.wikipedia.org/wiki/Dither [12:40] <JEEB> possibly because it's something under another name compared to the wikipedia page's author's word selection [12:40] <Animedude5555> It has absolutely EVERY kind of dithering, except FFMPEG's so called "arithmetic dithering". [12:41] <JEEB> I just don't know, you go look at what it actually does or the comments in the source code, and you might find it out [12:41] <JEEB> it's not the part of libav* that I touch :P [12:42] <Animedude5555> Is "aritmetic dithering" a brand new type of dithering, never before seen in the world? In other words, is it litterally a brand new invention of FFMPEG's developers over just the last couple months? [12:42] <JEEB> as I said [12:42] <JEEB> it might just be something called a different way in that article [12:42] <JEEB> it might be something not in that article [12:42] <JEEB> I have no idea, those are just possibilities [12:42] <JEEB> I mean, go check how Wikipedia confuses people with the aspect ratio article [12:43] <JEEB> because many video formats and FFmpeg uses SAR for the single sample's aspect ratio [12:43] <JEEB> yet Wikipedia means something completely different with SAR [12:43] <JEEB> or well, the author of that article [12:44] <JEEB> because wikipedia is not a single entity and all that [12:46] <relaxed> Animedude5555: http://pippin.gimp.org/a_dither/ [12:46] <relaxed> secrets found in the source [12:47] <relaxed> see libswscale/output.c [12:47] <JEEB> yes, which is why taking a look at the source is generally a Very Good Idea [12:47] <JEEB> you can find links to references etc [12:48] <Animedude5555> On the page you linked to he gives a function called "dither". It doesn't say which type of dither is being implemented in his example code. It does have a number of types of dithers listed near the top, but doesn't say which one is being used in his sample code. It could be any one of them. [12:49] <Animedude5555> Which means I may still be missing an explanation to "arithmetic dithering". [12:50] <JEEB> I will have to get a bit rude at this point and point you towards The Fine Source [12:50] <JEEB> that way you should have no questions whatsoever [12:51] <JEEB> there should be a a_dither dither in there implemented :P [12:51] <Animedude5555> It's implemented in C. I'm good ad VB, but not C. I read the FFMPEG C code and my head was spinning. [12:52] <Animedude5555> If somebody just made a simple pseudocode example that could be translated easily into any programming language, that would be what I ned. [12:52] <Animedude5555> *need [12:52] <JEEB> good luck getting spoonfed like that [12:59] <Animedude5555> In the pageyou linked to, it has several parameters to the dither function. They are input, x, y, c, pattern, levels. Nowhere does it explain what the different parameters represent. X and y are obviously the coordinates of a pixel. Input probably means the brightness level of the current pixel (but it is explicitly stated anywhere on that webpage). I have no clue what the parameter "c"... [12:59] <Animedude5555> ...represents, nor what "pattern" represents, nor what "levels" represents. Apparantly these last 2 are the constants "4" and "4". However I'm still confused about what "c" is. Where do I get this value, prior to passing it to the "dither" function? The page has a lot of stuff, but it is presented in a way that has almost NO EXPLANATION of what it is actually doing. This makes taking the... [12:59] <Animedude5555> ...code as it is presented, and actually implementing in anything, EXTREMELY DIFFICULT. [13:00] <JEEB> feel free to go and poke the author, that's all I can think of :P [13:26] <santa1> relaxed: please find the ffmpeg output (http://pastebin.com/DefacsjV) of the file which plays faster in mplayer but normal in vlc. There is also a bit of intermittent random pause in the video which avidemux says is related with b-frames. [13:28] <santa1> relaxed: got to go. Be back in another four hours time. Thanks for your support. [13:59] <Animedude5555> Got a problem here. No matter what I set -sws_dither to, it always uses error diffusion. [14:00] <Animedude5555> I'm trying to convert a series of BMP files to an animate gif. [14:00] <Animedude5555> My commandline is: [14:00] <Animedude5555> ffmpeg -f image2 -r 9.5 -i InputFrames\Frame%%04d.bmp -sws_dither none output.gif [14:00] <Animedude5555> I also tried changing "none" to "0". It still doesn't work! [14:02] <Animedude5555> It appears to always force error diffusion! [14:03] <Animedude5555> Is the Windows version hard-coded to use error diffusion?! [14:05] <relaxed> maybe -option sws_dither=none [14:06] <relaxed> nope [14:08] <Animedude5555> -option does notwork. [14:08] <Animedude5555> *not work [14:10] <Animedude5555> The people at Zerano appear to have compiled it for Windows, and it appears they may have changed some of the source-code prior to compiling for Windows. So the Windows version is not just the main branch of the software, recompiled to run on Windows. It appears to be an entirely different branch of the software. I am assuming this based on the fact that using based on its documentation does... [14:10] <Animedude5555> ...NOT always work (as in this case with the dither). [14:11] <Animedude5555> In effect, it is a "mod" of FFMPEG, not actually FFMPEG. [14:12] <Animedude5555> Can you do a favor for me? If you have access to the Linux source, and a Windows compiler, can you please compile me a "pure" Windows version? [14:16] <rcombs> uh& no, it's straight out of git [14:18] <rcombs> your problem is that there is no "none" option for `-sws_dither` [14:18] <Animedude5555> Then why is it not working with dithering? [14:19] <Animedude5555> I also tried it with "bayer" and "a_dither" and "x_dither". It always looks exactly the same. [14:20] <Animedude5555> The files output are identical (you can check the CRCs in programs that let you calculate the CRC of files). [14:21] <Animedude5555> And those 3 things (bayer, etc) ARE valid, just look in the help file that is output by "-h full". [14:21] <relaxed> -sws_flags lanczos -sws_dither a_dither [14:22] <Animedude5555> But I don't want to resize it. Lanczos is for resizing. [14:22] <Animedude5555> I want to change the dithering mechanism for conversion to gif. [14:22] <Animedude5555> I don't want to change the dithering mechanism for resizing. [14:23] <Animedude5555> I want to keep the same size, just convert to gif (for making an animated gif). [14:23] <Animedude5555> Does it require I use the -sws_flags commandline switch, before it will recognize the -sws_dither commandline switch? [14:23] <relaxed> I don't think so [14:24] <relaxed> just -sws_dither a_dither works here [14:24] <Animedude5555> So I am going to be FORCED to resize the image, if I want to select the dithering type? [14:24] <Animedude5555> Try -sws_dither bayer [14:24] <Animedude5555> You will quickly see that it doesn't work. [14:25] <Animedude5555> It will process the image, and output a file, giving NO ERRORS AT ALL, but the output file will look as if the default dithering (error diffusion) has been used, when with a "bayer" dither, it should have an obvious repeating pattern appearance to the image. [14:26] <Animedude5555> You can immediately tell that it is still using error diffusion, despite having selected "bayer" as the dithering type, because there will be no such repeating pattern. [14:27] <Animedude5555> Do you have a fix for this? Is this a bug in the program? [14:29] <Animedude5555> Hello? [14:29] <Animedude5555> Are you still ther? [14:29] <Animedude5555> *there [14:32] <relaxed> Animedude5555: ffmpeg -i input -vf scale=w=iw:h=ih:sws_dither=a_dither out.gif [14:33] <relaxed> you're welcome [14:34] <Animedude5555> What does "-vf" do? According to the internally generated help file with "-h full" the command line switch "-sws_dither" is supposed to be valid, such that I don't need -vf (followed by a bunch of stuff that I don't even know what it does). [14:34] <Animedude5555> But strangely enough, -sws_dither does not work. [14:34] <Animedude5555> Why? [14:34] <Animedude5555> Did some programmer make an error when typing out his C code for FFMPEG? [14:34] <relaxed> you need the scale filter [14:36] <relaxed> -s WxH -sws_dither a_dither would probably work too [14:36] <Animedude5555> The internally generated help file doesn't say I need to use a scale filter. It says that "-sws_dither" is a valid commandline switch. Is it not working as a stand-alone commandline switch, due to an error introduced into the FFMPEG software by one of the programmers? [14:36] <JEEB> sws_dither as it reads is an swscale option [14:37] <JEEB> swscale does not get included in your encoding chain by default [14:37] <Animedude5555> Huh? What does that mean? [14:37] <XHFHX> Hi there. I currently pipe an HD upscaled video with ffmpeg to ffmbc to create a xdcamhd file. I now wants that the final file has 8 mono-tracks instead of one stereo. how can this be archieved? My current command looks like this: ffmpeg\bin\ffmpeg.exe -i ffmpeg\oasis1.mp4 -vf "scale=1920:1080" -f avi pipe: | ffmbc\ffmbc.exe -i pipe: -target xdcamhd422 -vtag xd5c test2.mov [14:37] <JEEB> Animedude5555, instead of being a separate filter the dithering is within swscale, which is the do-it-all library that does colorspace conversions and resizing, among other things [14:38] <JEEB> and -sws_dither is a valid swscale option [14:38] <JEEB> if you have swscale actually working in your encoding chain [14:38] <JEEB> by default it isn't plugged in there [14:38] <JEEB> adding a scaling filter does that, unsurprisingly [14:38] <JEEB> even if it scales to your input width and height [14:39] <Animedude5555> But it appears that -sws_dither is being run internally during the conversion to gif, because a dither is most certainly being applied to the output. [14:39] <Animedude5555> So I though that meant that the -sws_dither commandline switch was automatically available to the user, when performing a conversion to a GIF file. [14:40] <JEEB> could be something else, or in any case your option was taken in but it had no effect because while you were doing stuff swscale was not called :P [14:40] <JEEB> or something like that [14:40] <JEEB> good luck and have fun with swscale :P [14:40] <JEEB> also it could even be the gif encoder doing extra dithering if needede [14:41] <JEEB> I just don't know [14:41] <Animedude5555> Can you pass this message onto the dev team for me? "Please include in the next version of FFMPEG, an option to set the dithering used on output files, particularly for GIF files." [14:41] <JEEB> no, you do it yourself if you think something is wrong [14:42] <JEEB> the trac is there for it [14:42] <Animedude5555> Are you not part of the development team yourself? [14:42] <JEEB> I have done some code for libav* but I keep the fuck away from swscale [14:42] <relaxed> Animedude5555: I showed you the option to set dithering [14:42] <JEEB> the dithering functionality is in swscale, and you need to use swscale for it to become available [14:43] <JEEB> so relaxed's command line does exactly that [14:43] <JEEB> you add a scale filter (which is swscale) [14:43] <JEEB> and then set the scale size to input width and height, and add the sws_dither option there [14:43] <JEEB> of course you could file a feature request for a separate dithering filter [14:44] <JEEB> so you would do -vf dither_shit=type=a_dither [14:44] <Animedude5555> I'm trying to avoid rescaling the image, using a command that specifies dither type only, and absolutely nothing for width or height. [14:44] <Animedude5555> Is there such a command? [14:44] <relaxed> -vf scale=w=iw:h=ih:sws_dither=a_dither [14:44] <JEEB> well good luck then, since it's in swscale [14:44] <JEEB> you have to scale [14:45] <JEEB> iw and ih set the output to be the same as input, though [14:45] <JEEB> for width and height [14:45] <relaxed> you don't actually scale in the above command [14:45] <relaxed> it passes through [14:45] <JEEB> yes, most sane scalers just skip the scaling part for that [14:45] <oomkiller> mediainfo file 1: http://pastebin.com/Cwz614tG , mediainfo file 2: http://pastebin.com/DNYfNe9W , my command ffmpeg -fflags +genpts -i VTS_01_1.VOB -i VTS_01_2.VOB -ss 00:00:00 -to 01:06:19 -vcodec hevc -x265-params crf=20 -sn -acodec ac3 -map 0:0 -map 0:1 -map 0:2 e01.mkv the error I get: Data stream encoding not supported yet (only streamcopy) . Can someone help me? [14:46] <oomkiller> if I add -dn it works but then I only get one audio stream [14:46] <Animedude5555> But then I'm still having to specify that output width = input width, and that still means I'm filling in height and width parameters. Is there a way to get swscale to just skip asking me about height and width, and accept the desired dithering as the only argument that I plan to provide to it? [14:47] <relaxed> Animedude5555: "JEEB : iw and ih set the output to be the same as input, though" [14:47] <relaxed> can you read? [14:48] <Animedude5555> I just found out that there is. I tried this "-vf scale=sws_dither=bayer" and it seems to work. [14:49] <Animedude5555> This way it isn't even filling in a "width argument" and "height argument" with the image's own width and height. It skips to the important part, of telling it the type of dithering you want to do. [14:49] <JEEB> it is filling it in inside [14:49] <relaxed> it's the same exact thing [14:49] <JEEB> it just happens to be the default [14:49] <JEEB> which of course is good since it makes the line shorter [14:49] <Animedude5555> Ok. [14:50] <JEEB> I of course have no idea if it is the default or not, but it sounds like that from your comments [14:50] <JEEB> and if it is, great [14:50] <JEEB> also for gif you probably want to set the output pix_fmt to rgb8 [14:52] <Animedude5555> Not sure why a chain of "=" signs does what it does anyway, but it works. It would seem that scale=sws_dither=bayer would be the same as saying separately "scale=bayer" and "sws_dither=bayer", but apparantly the second "=" sign has a different meaning than the first. This is kind of confusing, but seams to be very important when it comes to applying filters. Can someone explain it to me? [14:53] <JEEB> welcome to the libavfilter syntax :P [14:53] <JEEB> another part of libav* that I shall never touch [14:53] <rcombs> oomkiller: post the full output of the ffmpeg command that fails [14:53] <rcombs> (in a pastebin, please) [14:55] <oomkiller> rcombs: http://pastebin.com/TVW5R4TQ [14:56] <Animedude5555> it also appears to work to group them in more logical units like -vf scale="sws_dither=bayer" where you are basically saying to take the command "sws_dither=bayer" and pass it to the filter "scale". [14:57] <rcombs> Animedude5555: except those quotes are removed by your shell [14:57] <rcombs> oomkiller: you specified -map 0:0 [14:57] <rcombs> oomkiller: that stream is a data stream [14:58] <Animedude5555> I'm typing in in the Windows command prompt, no the Linux shell. [14:58] <oomkiller> rcombs: oh lol yes you're right [14:59] <rcombs> well, cmd.exe runs on insanity and crack, so good luck there [14:59] <Animedude5555> I think that the text in the command prompt in Windows is passed litterally to the program being run, and it is up to the program to handle the quot marks. [14:59] <oomkiller> rcombs: thx [14:59] <rcombs> oomkiller: also, you might want to do `-map 0:#0x1e0` and similar instead [15:00] <rcombs> oomkiller: the stream indexes (0, 1, 2&) in VOB files are arbitrarily defined, and may change between releases, whereas the IDs (#0x1e0, #0x80, &) are defined by the file and won't change [15:00] <oomkiller> rcombs: ah ok I will do, thx [15:01] <rcombs> it shouldn't make an actual difference in your output here, but I prefer to do it in VOBs and similar [15:03] <oomkiller> rcombs: why do I need the # ? [15:03] <XHFHX> Hi there. Can someone help me with this error message? http://pastebin.com/MFTfKuvs [15:05] <Animedude5555> Is it possible to force no dithering? [15:06] <Animedude5555> I want it to use "nearest color" only in my animated GIF. That's what I've been trying to achieve. However I can't figure out how. Now that I figured out how to specify various types of dithering, I'd like to be able to specify the type of dithering that is in fact NO DITHERING. Is there a way to do this? [15:08] <rcombs> oomkiller: specifies that you want a stream ID and not an index [15:08] <rcombs> XHFHX: I think it's pretty cut-and-dry [15:09] <rcombs> XHFHX: you need to specify a frame rate with -r before -target [15:09] <XHFHX> yeah, it seems so. but which frame rate should I choose? I'm not that much into video conversion [15:09] <rcombs> though I don't quite understand why you're piping ffmpeg to ffmpeg [15:09] <XHFHX> piping ffmpeg to ffmbc [15:09] <Animedude5555> I want it to use "nearest color" only in my animated GIF. That's what I've been trying to achieve. However I can't figure out how. Now that I figured out how to specify various types of dithering, I'd like to be able to specify the type of dithering that is in fact NO DITHERING. Is there a way to do this? [15:11] <rcombs> XHFHX: well, ffmbc is not ffmpeg, so it's not directly supported here [15:11] <rcombs> XHFHX: but the frame rate you should use generally depends on what you intend to use the output for [15:11] <XHFHX> so the output, not what the input is? [15:12] <XHFHX> because i don't understand when i pipe a single file i dont have to enter the framerate, but when i pipe something ffmpeg has created i have to set the framerate [15:12] <Animedude5555> Can you help me set FFMPEG for nearest-color mode? I want to disable dithering on my GIF output. [15:12] <rcombs> XHFHX: apparently (it's an ffmbc error, and I don't know that software, so I'm just guessing reading the error message) [15:12] <Animedude5555> Can you see my posts? [15:15] <XHFHX> ok, thanks rcombs: i now set -r pal and try some tests with different files how this works out! :) [15:15] <rcombs> XHFHX: cool! [15:16] <XHFHX> btw, is there a reason why ffmbc isn't combined with ffmpeg? [15:18] <rcombs> it's by different people [15:18] <rcombs> probably just uses ffmpeg's libs [15:21] <Animedude5555> Is there a way to make it so FFMPEG does not use any dithering when converting to gif? [15:21] <Animedude5555> Please let me know. [15:21] <Animedude5555> I want to set it to just use "nearest color". [15:24] <vlatkozelka> hi , if i want to scale and overlay an icon to a video that would be -filter_complex " [1:v] scale=20:20 ; [0:v][1:v] overlay=10:10 " right ? [15:25] <vlatkozelka> the overlay is working perfectly but the scaling isnt , the icon comes out in its native size [15:25] <rcombs> vlatkozelka: you're not using the output of the scale filter [15:25] <vlatkozelka> i dont understand that sorry [15:25] <rcombs> you're just taking the same input as the scale's input [15:26] <rcombs> you want something like this: " [1:v] scale=20:20 [scaled] ; [0:v][scaled] overlay=10:10 " [15:26] <vlatkozelka> ah [15:26] <vlatkozelka> scaled is like a variable name ? [15:27] <rcombs> yeah, similar [15:27] <vlatkozelka> explains all these examples ive been reading [15:27] <vlatkozelka> didnt know u can do that [15:27] <rcombs> now you know! :D [15:28] <vlatkozelka> thx alot that worked perfectly :) [15:28] <vlatkozelka> with a video too [16:50] <vlatkozelka> i have video1 and video2 , im scaling video2 and overlaying it onto video1 ... video2 is a short 5 sec video , but when it ends it doesnt dispear , how to do that ? [16:51] <c_14> https://ffmpeg.org/ffmpeg-filters.html#overlay-1 [16:51] <c_14> look at eof_action [16:51] <vlatkozelka> command im using : -filter_complex "[1:v] scale=50:50 [sc] ; [0:v][sc] overlay=10:10" [16:52] <vlatkozelka> ok [16:52] <vlatkozelka> btw where does -loop go ? [16:52] <vlatkozelka> if i want a gif to loop [16:53] <oomkiller> I'm getting an error with concatenating two files: http://pastebin.com/BeapSpqn , I guess the problem is the difference in the two video streams. how can I fix that? [16:54] <vlatkozelka> thanks c_14 :) [16:54] <vlatkozelka> added :eof_action=pass [16:55] <c_14> vlatkozelka: I think it's output [16:55] <vlatkozelka> ah [16:55] <vlatkozelka> so if i want to loop a gif i could use eof_action:repeat [16:55] <vlatkozelka> ill give it a try [16:55] <c_14> vlatkozelka: nah, repeat won't work [16:55] <c_14> That just repeats the last frame. [16:55] <c_14> It's the default anyway. [16:56] <vlatkozelka> yeah [16:56] <vlatkozelka> thats what happened indeed [16:57] <c_14> oomkiller: You probably at very least have to get the SARs to match up for both streams, you also might need to have the sizes be equal, I'm not sure if/how x265 handles size changes in the middle of the stream. [17:00] <oomkiller> c_14: I'll try to convert each file to x265 first and then try to concatenate them. [17:01] <c_14> I'd probably just prepend 2 scale filters into the beginning of the filterchain to give the streams square pixels. [17:38] <eldome> hello all, guys. Is there someone i can ask to about linux versions? (the one i installed from jon severinsson's launchpad) [17:45] <c_14> Just ask, if someone can help you they will. [17:46] <eldome> straigth forward so ;) Ok [17:47] <eldome> is there any chanche to see back the rotate filter? [17:48] <c_14> hmm? [17:49] <eldome> i need to rotate an overleyed set of frames. Need to obtain an arbitrary rotation [17:49] <eldome> i cannot run a command withe the rotate filter on the latest version [17:49] <eldome> and actually a ffmperg -filters command don't list the 'rotate' filter [17:50] <c_14> Can you pastebin the output of ffmpeg -version? [17:50] <eldome> sure [17:50] <eldome> just a sec [17:51] <eldome> ffmpeg version 1.2.6-7:1.2.6-1~trusty1 built on Apr 26 2014 18:52:58 with gcc 4.8 (Ubuntu 4.8.2-19ubuntu1) configuration: --arch=amd64 --disable-stripping --enable-avresample --enable-pthreads --enable-runtime-cpudetect --extra-version='7:1.2.6-1~trusty1' --libdir=/usr/lib/x86_64-linux-gnu --prefix=/usr --enable-bzlib --enable-libdc1394 --enable-libfreetype --enable-frei0r --enable-gnutls --enable-libgsm --enable-libmp3lame --enable-lib [17:52] <eldome> sorry, cutted away something... do you need the complete output or is this enough? [17:52] <c_14> That's enough. [17:54] <eldome> (jus saw the pastebin thingie. sorry, kinda n00b: cannot use it :| ) [17:54] <c_14> The rotate filter was added to master 3 (or 4) months after that version of FFmpeg was split from master. [17:55] <c_14> You'll need a newer version of ffmpeg. [17:55] <c_14> (Which is always a good idea anyways) [17:55] <eldome> that's a step.... thank you. Well i also found newer versions [17:55] <c_14> You can probably just use a static build. [17:56] <eldome> a static, right [17:56] <eldome> but the same command produced a horribly wrong result [17:56] <eldome> dropping almost the whole set of frames [17:57] <eldome> can i bother submitting the command ande the results? [17:57] <eldome> (on pastebin, i swear :D ) [17:57] <c_14> Ye, sure. [17:59] <eldome> http://pastebin.com/DYeAJm8W [17:59] <eldome> here's the command [18:00] <eldome> http://pastebin.com/wMJCCTAC [18:00] <eldome> and there's the output [18:01] <eldome> there always is an I/O error on the base frame (but it seems to not affect the reading of that fram) [18:01] <eldome> and it always elabs 2 frames dropping the remaining 173 [18:01] <eldome> (the sets are both made of 175 640x640 pngs) [18:03] <c_14> Ok, there's a few things I notice right off the bat. [18:04] <c_14> wait [18:04] <eldome> waiting and thanks in advance :) [18:05] <c_14> For the sake of my sanity I'm going to convert that filtergraph into a complex filtergraph with multiple inputs. [18:06] <eldome> do whatever you want... if you pop up with a solution, i'll adopt your way and build a commemorative monument to your person ;) [18:08] <c_14> You do know that x264 ignores -qscale, right? [18:09] <eldome> i know... there are some dirt coming from old tries. I guess it shouldn't impact on the whole stuff [18:10] <c_14> Yeah, just noticing things while I go over the command. [18:10] <eldome> fair enough :) [18:10] <oomkiller> c_14: converting it first to x265 before concatenating, didn't work. How does this work with 2 scale filters you mentioned? [18:11] <c_14> eldome: http://ix.io/epo < try that [18:13] <c_14> oomkiller: [input]scale=iw*sar:ih[out] < something like that for each of the streams somewhere at the beginning of the filtergraph [18:13] <c_14> you might need to append a setsar=1 after each of those (before the [out]), not sure though [18:16] <eldome> Output error: No such filter: '' [18:17] <eldome> (On a side note, the command i pasted works on a windows version: ffmpeg version N-61570-gaa86ccc built on Mar 17 2014 22:06:38 with gcc 4.8.2 (GCC) ) [18:18] <oomkiller> c_14: well that gives me the error: Too many inputs specified for the "scale" filter. my command right now: http://pastebin.com/JTzxR1hY [18:21] <eldome> (added the second set, removed the commas after [1] and [2], but... same old result: 173 frames dropped :( ) [18:21] <eldome> (Oh well, 347 frames dropped, actually :O ) [18:23] <c_14> oomkiller: You have to run the scale filter as a separate filterchain with the [input] being the first video stream and again with [input] being the second video stream. Make sure [out] is a different name both times and then take those names and give them as input to the concat filter [18:25] <c_14> eldome: The [2] should be a [1] [18:25] <c_14> and yes, the commas shouldn't have been there [18:25] <c_14> I forgot to get rid of them [18:26] <c_14> The now modified command works for me. [18:27] <c_14> It produces output at least. [18:29] <eldome> it does, but it's a vide with 2 frames [18:29] <eldome> 0 sec duration [18:29] <eldome> (it always did, sorry for being unclear) [18:36] <c_14> Ah, ok. [18:36] <c_14> But it worked correctly with that Windows build? [18:37] <eldome> yes it did [18:37] <eldome> (searching that version source to compile on linux, btwt) [20:06] <JaredBusch> any guides for setup on CentOS7 ? My Google skills are failing. [20:21] <c_14> https://trac.ffmpeg.org/wiki/CompilationGuide/Centos maybe? [22:54] <Animedude5555> I still have a question about dithering in FFMPEG. [22:54] <Animedude5555> I know now how to set the dithering with -vf scale=sws_dither=NameOfDitherType [22:55] <Animedude5555> However, I still can't figure out how to set it to do no dithering. I want to set it just to use "nearest color". [23:03] <relaxed> Animedude5555: Try -vf scale=sws_dither=0 [23:04] <Animedude5555> Tried it, but it seems to automatically use ED dithering anyway. [23:06] <relaxed> then it might not be possible, file a feature request and maybe they'll add the option [00:00] --- Sun Sep 21 2014

1 0