Ffmpeg-devel-irc

Download

ffmpeg-devel-irc@ffmpeg.org

August 2014

1 participants
60 discussions

ffmpeg-devel.log.20140814
by burek 15 Aug '14

15 Aug '14

[00:11] <cone-539> ffmpeg.git 03Andrew Stone 07master:db68ef898a38: ogg: update event_flags with STREAM_/METADATA_UPDATED whenever metadata changes. [00:11] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:a8db787932ad: Merge commit 'db68ef898a3802e51b6f41fd600d0d46d058e3f8' [00:18] <cone-539> ffmpeg.git 03Martin Storsjö 07master:4e629ef80e62: http: Fix authentication, broken since 6a463e7fb [00:18] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:e260c8180ec6: Merge commit '4e629ef80e62a54636cb46033998177dd08cf3ad' [00:23] <cone-539> ffmpeg.git 03Felix Abecassis 07master:159a06dfc83d: stereo3d: initialize AVStereo3D to zero [00:23] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:ad1b571b2682: Merge commit '159a06dfc83d189f753c4583583ddfb571552ff5' [00:28] <cone-539> ffmpeg.git 03Anton Khirnov 07master:aa51b0492bfc: avconv: rename output_packet() to process_input_packet() [00:28] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:097bf149c92a: Merge commit 'aa51b0492bfced6d650fb5ff419e2b13fde6833d' [00:37] <cone-539> ffmpeg.git 03Anton Khirnov 07master:8ddc32629a6d: mem: add av_strndup() for duplicating substrings [00:37] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:c8571c61ec4c: Merge commit '8ddc32629a6d6be77256694c9e322dde134609f3' [00:59] <cone-539> ffmpeg.git 03Anton Khirnov 07master:481a36674954: cmdutils: allow matching by metadata in stream specifiers [00:59] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:b8e4c11d9332: Merge commit '481a3667495425db9fdffb653292b6460fb68208' [01:26] <cone-539> ffmpeg.git 03Anton Khirnov 07master:30e50c50274f: lavf: eliminate ff_get_audio_frame_size() [01:26] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:082d52354f77: doc: fix toolname [01:26] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:0de0c75ebbec: Merge commit '30e50c50274f88f0f5ae829f401cd3c7f5266719' [01:48] <cone-539> ffmpeg.git 03Diego Biurrun 07master:353240541d4e: cpu-test: Add unistd.h #include for getopt() [01:48] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:af5ec182252d: Merge commit '353240541d4ec317471b5cbcaa3e027d00ff8f5c' [01:50] <jamrial> so it was indeed the mp3 muxer buffering 800+ mb of audio packets in memory waiting for the cover art packet to show up [01:50] <wm4> because it was such a good idea to model cover art as video stream [01:50] <Daemon404> lul [01:51] <Daemon404> also, 800 mb of mp3 audio? [01:51] <wm4> how else would anyone have noticed this [01:51] <Daemon404> i wonder what its a recording of [01:53] <jamrial> is it normal for mp3 files to have the cover art at the end? [01:53] <wm4> hm, normally id3v2 is at the beginning [01:54] <Daemon404> was id3v2 the one that base64 encoded the cover? [01:54] <wm4> it does? I wouldn't be surprised though [01:54] <wm4> I once tried to read id3v2 myself and god is it stupiud [01:54] <Daemon404> there was one format that did [01:55] <Daemon404> oh nope. [01:55] <Daemon404> it's ogg. [01:55] <Daemon404> of course it is ogg. [01:55] <wm4> of course! [01:56] <jamrial> technically it's vorbiscomment :P [01:56] <jamrial> it's flac format encoded with base64 inside a vorbiscomment tag [01:57] <Daemon404> not even its parents love it [01:57] <jamrial> lol [01:59] <mark4o> Daemon404: ogg opus is going into last call on the codec list; currently it defers to vorbiscomment but if act soon you can propose something more sane for ogg opus attached pictures [02:00] <jamrial> also, the source file in that ticket was an itunes m4a file. i suppose in those the cover art is at the end then [02:00] <Daemon404> mark4o, the base64 version is listed as deprecated. [02:00] <Daemon404> i assume it wont be used, and the binary version wll be. [02:01] <mark4o> I dont think it is depreceated; where did you see that? [02:01] <Daemon404> https://wiki.xiph.org/VorbisComment#Unofficial_COVERART_field_.28deprecated… [02:02] <mark4o> COVERART comment is deprecated, but replaced with METADATA_BLOCK_PICTURE. Both are base64-encoded. [02:04] <Daemon404> it doesnt explicitly say it is on that page [02:04] <Daemon404> if so, lame. [02:04] <mark4o> The binary FLAC picture structure is base64 encoded and placed within a VorbisComment with the tag name "METADATA_BLOCK_PICTURE". This is the preferred and recommended way of embedding cover art within VorbisComments. [02:04] <Daemon404> fun. [02:04] <Daemon404> luckily i dont give 2 shits about ogg because it is a terrible format and nothing i care about uses it [02:05] Action: Daemon404 is working on opus-in-mp4 spec [02:05] <wm4> but how will you play big buck bunny [02:05] <Compn> mkv ? [02:05] <Compn> or ... nut :) [02:05] <mark4o> I complained about the base64 but couldnt get anyone to care [02:06] <Compn> you could talk to xiphmont [02:06] <Compn> ./q xiphmont [02:08] <j-b> ./ ? [02:09] <Compn> .yes [02:19] <Compn> No claim can be processed if any of these are missing: the original receipt, [02:19] <Compn> the replacement receipt, the airline ticket (travel destination to home) or the [02:19] <Compn> police report. [02:20] <Compn> $40 rail safeguard plan ? no. [02:20] <cone-539> ffmpeg.git 03Luca Barbato 07master:e8049af1325d: mpegts: Do not try to write a PMT larger than SECTION_SIZE [02:20] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:1d7544b752e7: Merge commit 'e8049af1325dd59a51546c15b2e71a0f578e9d27' [02:20] <Compn> haha customer relations department is in quebec! hahahahhaa [02:21] <Compn> This program does not apply to any loss caused by: (a) Rail Pass or Ticket not in the holders actual possession at the time of loss. [02:22] <Compn> so if its stolen out of my hotel room while i'm not there, i'm out of luck. [02:25] <wm4> [PATCH] NULL-check Matroska chapters when reading header <- didn't we have that already [02:25] <wm4> or was it on libav [02:26] <wm4> no, it was on ffmpeg-devel, and the patch was applied [02:28] <iive> Compn: ?? [02:30] <sanjose_kid_> wondering if someone in community can help me... [02:30] <sanjose_kid_> i am trying to add the capability to ffmpeg project for packet-loss concealment into ffmpeg's RTP stack [02:31] <sanjose_kid_> am using RTP networking stack for low-latency video-conferencing [02:31] <sanjose_kid_> By packet loss concealment, I refer to whatever sender sends, receiver needs to receive independent of network conditions. [02:32] <sanjose_kid_> does anyone know how I can do this? [02:32] <Compn> iive : british rail train pass insurance... [02:33] <iive> Compn: oh, I thought it is about the rail coaster trains, like the one that malfunctioned recently [02:33] <BBB> so does anyone volunteer to do the sample cutting from that ticket for a fate test? [02:33] <BBB> (vp9 bug) [02:50] <sanjose_kid_> who is best to work with on ffmpeg's RTP networking stack? [02:51] <kierank> Compn: insurance for what? [02:51] <kierank> oh the ticket [02:54] <jamrial> sanjose_kid_: if nobody's on right now that can help you, you can either try at a later time or email ffmpeg-devel [02:54] <sanjose_kid_> @jamrial thank you [02:56] <sanjose_kid_> @jamrial do u know if there is a specific person who is very familiar with RTP stack in community? [02:58] <sanjose_kid_> @kierank did you mention that you are building your own RTP stack? [02:59] <sanjose_kid_> if yes, have u dealt with packet-loss concealment? [03:04] <kierank> yes we had this discussion yesterday [03:10] <sanjose_kid_> is it feasible to extend library to support this? [03:11] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:986ec3417abd: avformat/utils: Remove demuxer specific frame_size fallback from ff_get_audio_frame_size() [03:11] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:ed488d1535d9: Move frame_size fallback from ff_get_audio_frame_size() to av_get_audio_frame_duration() [03:11] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:7b59217b60cc: Move WMA case from ff_get_audio_frame_size() to av_get_audio_frame_duration() [03:11] <kierank> not easy but possible [03:15] <sanjose_kid_> @kierank do u know calls to make specifically for the communications on packet-loss between server and client ? [03:15] <kierank> calls? [03:15] <kierank> you don't do that [03:15] <kierank> you use the rtp sequence number to measure the packet loss [03:16] <kierank> it's hard in a vbr setting though because you have to buffer [03:17] <sanjose_kid_> so if you're dealing with variable bit rate, how do you buffer? [03:21] <sanjose_kid_> is there any sample project you know of that shows how to handle buffering? [03:24] <sanjose_kid_> such as ability to set it to certain mode depending on latency or packet loss [03:24] <cone-539> ffmpeg.git 03Anton Khirnov 07master:d92550d19129: lavf: eliminate ff_get_audio_frame_size() [04:30] <cone-539> ffmpeg.git 03Christophe Gisquet 07master:69849a2d6ecc: dpxenc: enforce alignment requirement [04:36] <cone-539> ffmpeg.git 03Christophe Gisquet 07master:7cdef77b5036: dpx: warn if encrypted [06:13] <cone-539> ffmpeg.git 03James Almer 07master:dffbac0956ca: lavf/oggparsevp8: use ff_vorbis_stream_comment() [11:23] <sanjose_kid_> using ffmpegs RTP framework how do i get access to the rtp sequence number to measure packet loss? [11:23] <sanjose_kid_> if anyone knows, that would be super helpful info [11:56] <ubitux> J_Darnley: what was this gnu time you talked about? [12:13] <michaelni> sanjose_kid_, access from where ? from a user app i guess there maybe is no way, but sidedata could be used to export it [12:17] <BBB> michaelni: sanjose_kid_: theres indeed no way right now; if you want to expose it, side-data is fine (although you could alternatively but the quality-connection in libavformat also; is it a custom format or something standardized youre using for the QoS connection?) [12:18] <ubitux> ping on motion vectors export [12:18] <ubitux> i'll push soon [12:18] <saste> sanjose_kid_, I think that info is not currently exposed, read above [12:18] <saste> ubitux, good, i'll review it now so I get a grasp of this side-data thing ;-) [12:18] <ubitux> cool, thanks [12:29] <J_Darnley> ubitux: a `time` program that isn't the bash builtin tool [12:29] <J_Darnley> cygwin has one available as a separate package [12:32] <J_Darnley> I only wanted it to measure the run time of ffmpeg [12:36] <J_Darnley> If there isn't an easy to install package, don't bother. [12:37] <J_Darnley> I measured the time spent in the function anyway [12:38] <ubitux> it's not in the package manager [12:38] <ubitux> perf should be available though [12:38] <ubitux> zsh also has a different time if you want [12:43] <cone-524> ffmpeg.git 03Edgar Hucek 07master:ab059f0aa896: vaapi: set the scaling list correctly. [12:43] <cone-524> ffmpeg.git 03Diego Biurrun 07master:e070d0a5ca90: frame: Remove some FF_API_AVFRAME_COLORSPACE leftovers [12:43] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:87988d6569ce: Merge commit 'ab059f0aa896e01e8e4529f5f714fde111f05377' [12:43] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:de7b08cbc1c1: Merge commit 'e070d0a5ca9047192e324a3f87006b316e2a08a7' [12:48] <sanjose_kid_> @michaeilni, were trying to implement packet-loss concealment and in that goal, to identify lost packages which we can then produce redundant packages for [12:49] <cone-524> ffmpeg.git 03Nidhi Makhijani 07master:0528226a05cc: a64: Return correct error code on invalid data stream [12:49] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:8d403d9c89fe: Merge commit '0528226a05cc08b74197547fba0b1939bf68990d' [12:53] <sanjose_kid_> @BBB, Not Custum, were using x.264/Speex or OPUS and using av_read_frame() to get packages [12:53] <sanjose_kid_> and were just reading receiver reports for QoS connection [12:56] <sanjose_kid_> however, the receiver reports don't allow us to initiate a retransmission for redundant packages [13:01] <cone-524> ffmpeg.git 03Nidhi Makhijani 07master:93f29948e4b0: mpeg4video: Fix doxygen comment syntax to document correct struct member [13:01] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:b634c12cb967: Merge commit '93f29948e4b06acfb96e7f82d373ef86d6dc55f7' [13:06] <sanjose_kid_> @michaelni do u think we need to patch ffmpeg to accomplish this? [13:09] <Compn> sanjose_kid_ : you might want to try asking on the mailing list [13:09] <Compn> still, stick around here, but you may get some replies there too [13:11] <sanjose_kid_> Compn - thank you [13:25] <BBB> sanjose_kid_: yes youll need to patch libavformat (one of the libraries in ffmpeg) to accomplish this [13:46] <sanjose_kid_> is it possible to get the RTP header information [13:58] <cone-524> ffmpeg.git 03Christophe Gisquet 07master:4ba45c189cca: dpx: use aligned line starts [14:02] <kurosu> if bumping lavc's micro, is there a document where to report the change (possibly indicating why) ? [14:04] <nevcairiel> micro is really not all that wild [14:04] <kurosu> yeah, and considering it's for dpx... [15:43] <cone-524> ffmpeg.git 03Christophe Gisquet 07master:117bc8e6ffc7: proresenc_kostya: properly account for alpha [15:43] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:6c36b3afe72d: avcodec/lcldec: initialize encoded correctly [16:53] <cone-524> ffmpeg.git 03Christophe Gisquet 07master:58d380f9a7ce: libavcodec: bump micro to reflect dpx changes [16:53] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:5c7899a4834e: avcodec/mjpegdec: Support AV_PIX_FMT_YUV420P16 with upscale_h [17:09] <J_Darnley> I am still looking for someone with an XOP capable CPU to test a patch for me. [17:12] <J_Darnley> Or perhaps someone can tell me whether the XOP looks correct in this patch: https://gitorious.org/ffmpeg/jdarnley-ffmpeg/commit/5c336afd808cf1a1f4e1c35… [17:18] <iive> J_Darnley: what is XOP? [17:20] <ubitux> the amd avx2 [17:22] <J_Darnley> integer versions of the fused multiply-add instructions [17:38] <J_D> Dammit! Who keeps interrupting my connection? [19:31] <cone-524> ffmpeg.git 03Clément BSsch 07master:10d96d8d66b8: avfilter/select: re-align a few comments [19:31] <cone-524> ffmpeg.git 03Clément BSsch 07master:37bfeca78cf3: avfilter/select: larger pixel sad computation [19:53] <cone-524> ffmpeg.git 03Nicolas George 07master:a3aaaec8916b: lavfi/avf_showspectrum: set output frame rate. [19:53] <cone-524> ffmpeg.git 03Nicolas George 07master:65b284a4aef6: lavfi/avf_showspectrum: fix output pts computation. [19:54] <cone-524> ffmpeg.git 03Nicolas George 07master:d4de6d4fadcc: lavfi/avf_showspectrum: do not push the frame at EOF. [19:54] <cone-524> ffmpeg.git 03Nicolas George 07master:ec33df60457d: lavfi/avf_showspectrum: use automatic framing. [19:54] <cone-524> ffmpeg.git 03Nicolas George 07master:7c10e32ae5a1: lavfi/avf_showspectrum: add full frame sliding mode. [19:54] <cone-524> ffmpeg.git 03Nicolas George 07master:638eec2ac34b: lavfi/avf_showspectrum: check RDFT context init. [19:54] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:c6c345ea928b: Merge remote-tracking branch 'cigaes/master' [20:00] <kierank> wm4: you should make ffmpeg merge the dts code [20:00] <kierank> dts-hd [20:00] <kierank> that'll kickstart work on it [20:01] <Daemon404> hah! [20:01] <Daemon404> the problem is the only person who is qualified to review it, cant [20:01] <Daemon404> due to NDA [20:01] <Daemon404> or w/e [20:02] <wm4> last I heard about it, it seemed to work relatively well? [20:02] <Daemon404> its not lossless yet [20:02] <wm4> who really cares about that [20:02] <JEEB> did it now do exact integer decoding of the lossy part? [20:02] <JEEB> so that it matches the XLL extension [20:02] <JEEB> :3 [20:02] <Daemon404> no dts core is still float iirc [20:02] <JEEB> k [20:03] Action: ubitux wonders if he should dump a glib-less pkg-config.c in TOOLS/ for cehoyos [20:03] <ubitux> actually, tools/ this isn't mplayer [20:04] <Daemon404> you should give carl a great big cup of Go Fuck Yourself [20:04] <Daemon404> his argument is like [20:04] <ubitux> that's not exactly polite [20:05] <Daemon404> "we cant have nice things because theyre different than the less nice things" [20:05] <Daemon404> i.e. completely insane [20:05] <Daemon404> [19:04] <@ubitux> that's not exactly polite <--- >implyign carl is ever anything but rude to anyone [20:05] <nevcairiel> JEEB: its relatively close though, 2-3 bits diff in a 24 bit result is beyond audible range [20:05] <JEEB> nice [20:06] <nevcairiel> But it still has other problems [20:06] <ubitux> Daemon404: he isn't right now, he is just politely annoying me ;) [20:07] <Daemon404> i feel like i need to quote that debian thread [20:07] <Daemon404> about how ffmpeg wont accept "people who dont work on X, blockign commits to X" [20:09] <kierank> 7:02 PM <wm4> who really cares about that --> this a million times [20:09] <kierank> wm4: though more like who will really notice [20:10] <wm4> nevcairiel says there are other problems, though, so maybe that's the reason [20:10] <Daemon404> kierank, part of the angry DTS people who yelled at me was "those assholes dont implement DTS right and call it DTS" [20:10] <Daemon404> a large part [20:11] <Daemon404> yes the irony is lost on them [20:11] <kierank> Daemon404: tell them to document it correctly then in the spec [20:11] <kierank> because DCA is an "open standard" [20:11] <Daemon404> you are open to purchase it [20:12] <kierank> no there's an etsi standard [20:12] <kierank> open to purchase is sony bollocks [20:12] <Daemon404> orite [20:15] <nevcairiel> They should penalize companies for putting out incomplete and unimplementable standards [20:17] <kierank> unlicensable too [20:18] <wm4> what is more baffling is that the piracy scene is using it [20:19] <Daemon404> wm4, since when does the scene do anything that isnt buttfuck retarded? [20:19] <cone-524> ffmpeg.git 03James Darnley 07master:7ce6c021dc47: lavc/flacdsp: change lpc_encoder function pointer prototype [20:20] <wm4> Daemon404: true... [20:23] <kierank> wm4: using dts? [20:23] <Daemon404> dts-hd ma [20:23] <Daemon404> even [20:23] <kierank> really? [20:23] <Daemon404> yes [20:23] <Daemon404> though usually they remux the dts core [20:23] <kierank> yeah [20:23] <Daemon404> which is also retarded. [20:23] <jamrial> why? why reencode it? [20:24] <Daemon404> you could say the same in argument for muxing flac or wavpack from lpcm sources [20:24] <Daemon404> it's a collosal waste of bw for no tangible gain. [20:24] <wm4> but they do it for spdif [20:25] <jamrial> dts-core is good. dts-hd ma is probably overkill, yeah [20:25] <Daemon404> dts-core is not 'good' [20:25] <Daemon404> i especially like teh DVDs i have with 2ch dts that has the same bitrate as pcm [20:26] <kierank> dts core is backwards compatible [20:26] <kierank> that's alll [20:26] <Daemon404> persoanlly, i'd just decode the dts-hd ma full properly, and encode it to aac [20:26] <jamrial> that's stupid. dts 6ch is 768kbps to 1.5mbps in almost every case [20:26] <Daemon404> or opus [20:27] <Daemon404> youll end up with something better than teh dts core, veyr likely [20:27] <Daemon404> as well as smaller. [20:28] <kierank> but then you can't spdif [20:28] <kierank> of course opus/aac will be better [20:28] <Daemon404> 0 fucks given about spdif [20:28] <Daemon404> from me [20:28] <Daemon404> for 'consumer' playback of pirated material. [20:50] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:fce8817a01d1: avformat/format: remove unused variable [20:50] <cone-524> ffmpeg.git 03Michael Niedermayer 07master:8b6cbc3c3319: avutil/opt: remove disabled old ABI compatibility code [21:13] <TheFluff> 20:26:53 <@jamrial> that's stupid. dts 6ch is 768kbps to 1.5mbps in almost every case <-- I'm pretty sure I've managed to reencode 6ch dts-core (which is lossy) to flac and actually save space, that's how dumb it is [21:50] <wm4> the old subtitle drama [21:51] <Daemon404> i prefer my solution: dont use i t [22:21] <nevcairiel> Oh pkg-config drama continues [22:22] <Daemon404> Carl Eugen vs The World [22:22] <nevcairiel> Why do we let him block a change again [22:22] <nevcairiel> I forgot [22:22] <Daemon404> [19:07] <@Daemon404> i feel like i need to quote that debian thread [22:22] <Daemon404> [19:07] <@Daemon404> about how ffmpeg wont accept "people who dont work on X, blockign commits to X" [22:23] <wm4> what's with all that useless mp4 "metadata" [22:23] <wm4> how am I suppose to differentiate this from real file tags? [22:24] <nevcairiel> Ask on the ML wth its good for [22:29] <wm4> hm maybe the way to "fix" this would be having multiple metadata fields, one for "tags", one for "additional crap" [22:29] <ubitux> for some reason, "mov m0,[p1]; psadbw m0,[p2]; mov m1,[p1+stride]; psadbw m1,[p2+stride];" is faster than "mov m0,[p1]; mov m1,[p1+stride]; psadbw m0,[p2]; psadbw m1,[p2+stride];" [22:29] <ubitux> i wonder if i misunderstood pairing in the first place [22:33] <ubitux> now i just made 16x16 sse2 from 560 to 500 decicycles, but i still have no idea why 4 8x8 mmxext is still faster (visible on overall time) [22:34] <ubitux> 8x8 is ~240 decicyles so i really wonder what's going on here [22:34] <ubitux> (cache?) [22:51] <ubitux> emms_c() should probably be made public btw [22:53] <Daemon404> wat [23:02] <ubitux> Daemon404: with the pixelutils API, it makes sense because user will need it [23:03] <wm4> that's one fucked up API then [23:03] <ubitux> why? [23:03] <wm4> because it violates standard C? [23:03] <ubitux> you don't want to call emms after each block call [23:03] <ubitux> what? [23:05] <Daemon404> >making the user manually call emms [23:05] <Daemon404> yeah [23:05] <Daemon404> [22:03] < wm4> that's one fucked up API then [23:05] <wm4> ubitux: standard C things cease to work between the API call and emms_c [23:05] <wm4> because the FPU is in "fucked up state" [23:05] <wm4> (wait, why even the FPU...) [23:05] <wm4> god x86 is so fucked up [23:06] <wm4> /rant [23:06] <wm4> wasn't emms_c needed for MMX only? [23:06] <ubitux> probably [23:07] <ubitux> (what do you suggest? calling emms after each block sad even if it will mean nothing thousands of times per frame?) [23:08] <wm4> make the API do a meaningful load of work [23:08] <ubitux> there are tons of sad usages [23:08] <ubitux> i can't guess them all [23:09] <wm4> why does the API even have to export it anyway [23:09] <ubitux> because it's used in libavfilter, and probably will in lavc soon [23:09] <ubitux> (it's actually ported from lavc) [23:09] <wm4> so it's just to uphold the independent libraries lie? [23:10] <ubitux> it's just some common code shared in a clean way [23:10] <ubitux> pixelutils can be useful for users anyway [23:10] <ubitux> i mean, it's not like SAD isn't a common operation [23:11] <wm4> this sounds like it should be really either strictly internal, or a separate library [23:11] <wm4> nothing that belongs into the ffmpeg API [23:11] <ubitux> it won't be compiled if you don't need it anyway [23:11] <ubitux> (but you probably will) [23:11] <wm4> that's not really what I care about [23:12] <ubitux> it can't be "strictly internal" (or maybe avpriv?) [23:12] <ubitux> and it's already in a separate library [23:12] <ubitux> which is called libavutil [23:13] <ubitux> anyway, i want to provide such api for anyone willing to do some image processing [23:13] <ubitux> which sounds like appropriate for a bunch of ffmpeg users [23:13] <wm4> lol [23:13] <wm4> if someone wants to do image processing, they won't use ffmpeg [23:14] <ubitux> now they can [23:14] <ubitux> ;) [23:14] <wm4> if anything, they'll use things like pixman or something [23:14] <wm4> or whatever other specialized image processing libs there are [23:14] <ubitux> well, some are probably already linking on ffmpeg [23:14] <ubitux> for decoding [23:15] <ubitux> so better gives them a bunch of useful tools to do something with that [23:15] <wm4> just because you're developing for ffmpeg it doesn't mean you should dump whatever you find useful into its API [23:15] <Daemon404> ^^^^^^^^^^^^^^^^^^^^^ [23:15] <ubitux> it's some generic pixel utilities, which we need in both libavcodec and libavfilter [23:16] <ubitux> so it actually sounds like something useful for other users [23:16] <ubitux> anyway, this was on the mailing-list a while ago [23:16] <wm4> sounds like dsptuils reloaded... [23:16] <ubitux> you could have commented [23:22] <jamrial> wm4: it's basically a subset of me_cmp from lavc but in lavu [23:52] <wm4> ubitux: hm, since when does libavformat return the correct ASS subtitle format? [23:52] <wm4> ubitux: did it actually get bumped with the recent bump? [00:00] --- Fri Aug 15 2014

1 0

ffmpeg.log.20140814
by burek 15 Aug '14

15 Aug '14

[00:17] <th0rne> If I have an m4a file of bitrate some x and want to convert it to ogg, is there some bitrate cx that I should aim for to get a reasonable balance of not too much extra lossiness and not too much larger file size? What would that value c be? Or am I approaching this wrong? [00:19] <sacarasc> You shouldn't reencode at all. You'll always lose something. [00:30] <th0rne> But assuming I have m4a, but need not m4a and have no way around that. [00:30] <th0rne> Some (additional) loss is acceptable to me. [00:33] <sacarasc> There is no formula for doing it. [00:57] <freeroute> hi, I'm just trying to flip an mp4 upside down but I end up getting errors - http://kpaste.net/9fea11 - I'm really noob at all of this, so could someone please help? [01:00] <c_14> First of all, you're not flipping the mp4 upside down, you're rotating it counterclockwise. Second you're using Libav not FFmpeg; either use FFmpeg or ask #libav for support. [01:02] <freeroute> c_14: should it be "ffmpeg -i <myvid> -vf transpose=0 <myvid-flipped>" ? [01:02] <freeroute> also, how do I get ffmpeg on Ubuntu 12.04? [01:02] <c_14> That or compile it yourself. [01:03] <c_14> freeroute: Either use the hflip or vflip filter. [01:03] <c_14> I never know which one stands for which. [01:03] <c_14> I think it's horizontally across the horizon, so maybe hflip first? [01:06] <freeroute> hmm, it seems to be giving me the same kind of errors. Seems that the format is not supported. I'll try downloading those ffmpeg stuff [01:28] <freeroute> c_14: I downloaded the ffmpeg.static.64bit.latest.tar.gz and extracted it into the folder where I have all the videos [01:28] <freeroute> somehow it doesn't appear to add the +x to it [01:28] <freeroute> no matter what I do, chmod 777 ffmpeg or chmod +x ffmpeg [01:28] <freeroute> is this normal? [01:28] <c_14> Can you pastebin the output of ls -l ffmpeg ? [01:29] <c_14> And the error output as well. [01:30] <freeroute> http://kpaste.net/3c1ae02b [01:30] <freeroute> not getting any error output when I do chmod [01:30] <sacarasc> Is the partition mounted with exec? [01:32] <freeroute> /dev/sda2 on /media/Windows7_OS type fuseblk (rw,nosuid,nodev,allow_other,default_permissions,blksize=4096) [01:32] <freeroute> does this mean it's not? [01:36] <c_14> try something along the lines of `mount -o remount,exec /dev/sda2 /media/Windows7_OS [01:38] <freeroute> c_14: I just did this command from my home Downloads dir - "./ffmpeg -i /media/Windows7_OS/films/galaxy/20140814_003236.mp4 -vf "vflip" /media/Windows7_OS/films/galaxy/20140814_003236_fl.mp4" and it worked [01:38] <freeroute> but I'll keep that in mind next time I get weird permission problems [01:38] <freeroute> I have still lots to learn [01:41] <freeroute> for example, since I don't know how to bulk process the flipped videos I am doing this manually one by one [01:41] <freeroute> I'm sure the command has this "for i=1" or something, but I haven't come to that yet [01:41] <c_14> ffmpeg doesn't have that internally, but any decent shell should [01:46] <freeroute> damn wifi [01:47] <freeroute> c_14: yeah, I haven't had time to learn the shell yet properly, which is a shame really because I'm trying to get on the CLI as much as possible [01:48] <sacarasc> for i in *.mp4; do stuff with "$i"; done [01:54] <freeroute> sacarasc: so how do I say, for every modified file just add "_fl" to it before the extension so that the processed files look like filename_fl.mp4 ? [01:55] <sacarasc> That's where I fail. [01:55] <c_14> for i in *.mp4; do stuff with "$i" and name it "${i#.mp4}_fl.mp4 [01:55] <c_14> something like that [01:55] <c_14> eh, with the trailing "; done of course [01:56] <c_14> Eh '%' instead of '#' [01:58] <c_14> `bash variable mangling' being the golden google search-phrase [02:01] <starPause> i'm able to cut a 30 second (or less) chunk out of a movie, but i want to split an entire movie into several 30 second (or less) chunks with a single ffmpeg command. is that possible or do i need to write a script that calls several ffmpeg commands? [02:02] <c_14> the segment muxer is your friend [02:02] <c_14> (probably) [02:02] <freeroute> c_14: I will study that command and that search term [02:02] <freeroute> one day :p [02:03] <c_14> starPause: `-f segment -segment_time 30' should do it [02:03] <c_14> Read the man page for more fun tidbits. [02:03] <starPause> c_14: thanks for the help will check that [02:34] <sanjose_kid__> i am trying to add the capability to ffmpeg project for packet-loss concealment into ffmpeg's RTP stack [02:34] <sanjose_kid__> am using RTP networking stack for low-latency video-conferencing [02:34] <sanjose_kid__> By packet loss concealment, I refer to whatever sender sends, receiver needs to receive independent of network conditions. [02:34] <sanjose_kid__> does anyone know how I can do this? [02:44] <sanjose_kid__> https://www.ffmpeg.org/ffmpeg-protocols.html [02:44] <sanjose_kid__> this doc doesn't seem to include packet-loss concealment [03:00] <icecube45> hello! I'm trying to stream my webcam via ffmpeg, however, I get the error: Cannot find a proper format for codec 'none' (id 0), pixel format 'none' (id -1) [03:00] <icecube45> Assertion *codec_id != AV_CODEC_ID_NONE failed at /build/buildd/ffmpeg-1.2.6/libavdevice/v4l2.c:868 [03:27] <kode54> for %i in (*.mp4) do ffmpeg -i "%i" -c:a copy -vn "%i.audio.m4a" [03:28] <kode54> oops [03:28] <kode54> didn't realize I was scrolled up [03:37] <freeroute> kode54: your mistake is my learning vehicle :p [03:37] <freeroute> I'll study this command as well [03:49] <kode54> interesting test file [03:49] <kode54> http://www.hydrogenaud.io/forums/index.php?showtopic=106570&hl= [03:49] <kode54> wmal2pcm.exe manages to decode 4 samples from the packet within, disregarding that the ASF header specifies both a duration and a pre-roll of 3 seconds [03:50] <kode54> the stream itself contains a single packet of WMAL data [03:50] <kode54> the decoder eats the first 2 bytes as a stream reset [03:50] <kode54> then eats the other 13KB or so, outputting nothing [04:55] <Zhenya> Hi Everybody. I am piping a mp4 file and always get Error setting option pix_fmt to value -1. [04:55] <Zhenya> I am trying to put -pix_fmt option in front of '-i inputfile', but it seems not work [04:56] <Zhenya> By the way, the ffmpeg version I was using is the latest one from git repo [04:59] <Zhenya> And the mp4 video clip is from a sumsung tablet. [05:01] <Zhenya> The command I use: $ cat b2.mp4 | ./ffmpeg -pix_fmt yuva422p -i pipe:0 a.mp4 [05:02] <Zhenya> The video is at: https://www.dropbox.com/s/oz0e51ggkbpgj0f/b2.mp4 [05:05] <relaxed> Zhenya: why would you piping it? [05:06] <Zhenya> As the video will be a live stream [05:06] <Zhenya> The video will be in record when I have to parse it [05:07] <relaxed> pastebin the output of ffmpeg -i b2.mp4 [05:09] <Zhenya> relaxed: here it is http://pastebin.com/8hNdLeQ2 [05:09] <Zhenya> relaxed: unspecified pixel format... How can I specify one? [05:34] <relaxed> that's not what I asked for. [05:47] <Zhenya> relaxed: sorry, 1 second [05:48] <Zhenya> relaxed: http://pastebin.com/a07HXfJ1 [05:55] <relaxed> did you try? --> cat b2.mp4 | ./ffmpeg -i pipe:0 a.mp4 [05:56] <Zhenya> Yeah, that's exactly what I run [05:57] <relaxed> no, you had -pix_fmt yuva422p in there earlier [05:58] <Zhenya> I added the prefix when the original one not work.. [05:58] <Zhenya> see http://pastebin.com/8hNdLeQ2 [05:58] <Zhenya> relaxed: see http://pastebin.com/8hNdLeQ2 [05:59] <relaxed> cat b2.mp4 | ./ffmpeg -pix_fmt yuv420p -i pipe:0 a.mp4 [05:59] <relaxed> the output also says, "Consider increasing the value for the 'analyzeduration' and 'probesize' options" [06:00] <Zhenya> This is mine: http://pastebin.com/vCwXQJVF [06:00] <Zhenya> Option pixel_format not found. [06:05] <Zhenya> Do you know what's the difference between isomavc1 and isom3gp4? [06:08] <relaxed> cat b2.mp4 | ./ffmpeg -probesize 10M -analyzeduration 10M -i pipe:0 a.mp4 [06:10] <Zhenya> let me try [06:11] <Zhenya> same error.. http://pastebin.com/98Z2etAL [06:12] <Zhenya> you can see in the output: minor_version : 0, I have another file with minor_version : 1. which can be piped [06:15] <relaxed> you should really build it without --disable-yasm, because encoding will be much slower [06:16] <relaxed> build the latest release and see if it still happens [06:16] <Zhenya> ok, got it. I will recompile it later. [06:16] <Zhenya> let me do a git pull [06:17] <Zhenya> relaxed: are you a ffmpeg developer? [06:19] <relaxed> no [06:19] <Zhenya> relaxed: I am recompiling [06:25] <Zhenya> relaxed: no, not work. I think something wrong with the file format [06:27] <relaxed> it's because the moov atom, or index, isn't in the front [06:27] <Zhenya> yeah [06:28] <Zhenya> I was trying to specify them in parameters [06:28] <Zhenya> via parameters [06:28] <relaxed> it works if I pipe it after running it through mp4box [06:29] <Zhenya> -pix_fmt not work for input stream [06:30] <Zhenya> what is mp4box? [06:30] <relaxed> Zhenya: you need the app to record the video in the flv container or something else that pipeable [06:30] <Zhenya> what's your command in shell? [06:30] <Zhenya> is avi pipeable? [06:31] <Zhenya> I am actually building an android app and want to apply subtitle in real time otherwise it may take the time too long [06:32] <Zhenya> so I piping the file in record to ffmpeg to apply subtitle [06:33] <Zhenya> So you know any fast/quick way to apply subtitle [06:33] <Zhenya> hopefully just merge two files into one ball (mp4 file and subtitle file) [06:34] <Zhenya> Do you know* [06:34] <Zhenya> any fast/quick way to apply subtitle? [06:36] <relaxed> hardsubs or soft? [06:37] <Zhenya> i am using ffmpeg subtitle [06:37] <Zhenya> which is really slow [06:37] <Zhenya> to burn subtitle over video [06:37] <Zhenya> if soft is ok? [06:38] <Zhenya> Now I am doing it hard [06:38] <Zhenya> hard subtitle [06:38] <relaxed> libass might work, but I'mnot sure how you would do that on the fly [06:39] <Zhenya> my plan is to use pipe feature to split file into pieces, then apply them one by one, then merge them [06:39] <Zhenya> for example 10s a file [06:40] <Zhenya> something like: tail -f b.mp4 | ffmpeg -ss x:x:x:2.2.2 -i pipe:0 [06:46] <Zhenya> -ss so quick [06:47] <Zhenya> If i can do the pipe.... [09:11] <termos> what exactly do I need so that moov atom is written at the beginning of the file? I keep getting "moov atom not found" [09:12] <relaxed> termos: it probably means the index in your mp4 is absent, which happens when you have an incomplete file [09:13] <termos> I thought there would be a workaround for that, I tried with -movflags +faststart but no success [09:15] <relaxed> are you interrupting the encode? why is it incomplete? [09:19] <termos> yes i'm interrupting it, I want to make sure that the file is playable even if it's interrupted [09:20] <relaxed> try using "killall -INT ffmpeg" instead of ctrl+c or "q" [09:22] <relaxed> I think with the other ways you're stopping ffmpeg before it has a chance to use -movflags +faststart [09:29] <termos> hmm I see [09:30] <termos> I can't really decide how to end the program, it's just that I'm transcoding really big files and I want to make sure that the file is playable even if the program crashes [09:30] <termos> without running it through a second pass or some qt-faststart utility [09:31] <termos> similar to how it works if I write flv to file, then it can end at any time [09:55] <relaxed> Is there some reason you need mp4? If not, go with matroska [10:19] <termos> not really any reason, mkv seems to work a lot bette [10:19] <termos> better [10:38] <Zhenya> relaxed: Do you know how to set a subtitle stream to be on by default? [10:40] <Mavrik> termos, MOOV atom cannot be generated before full file (segment) is written [10:40] <Mavrik> so what you're asking cannot be done [10:40] <Mavrik> do qt-faststart as a second pass. [11:10] <ghospich> Anyone faced with this issue? https://trac.ffmpeg.org/ticket/3852#ticket [11:17] <ubitux> ghospich: http://trac.ffmpeg.org/ticket/2067 [11:26] <ghospich> thanks, i'll try [13:41] <baidoc2> I`m trying to get the frames of a video frame by frame, it works fine when I write the frames as files but now I want to redirect the frames to pipe [13:41] <baidoc2> now how can I know when a frame is complete, to step to the next frame? [13:41] <baidoc2> because atm what i see is a bytestream [13:53] <Mavrik> baidoc2, well you know the frame size [13:54] <baidoc2> how can I know it? [13:55] <Mavrik> width * size * bits_per_pixel [16:11] <Nopik> hi, I'm trying to change pixel format, I'm doing ffplay -i file.mpg -pixel_format rgba, but is says that "Option pixel_format not found.", I tried several approaches, none of which seemed to work, any ideas how I should format the argument? [16:13] <Keshl> I use "-pix_fmt", oÉo. [16:15] <Nopik> hm, and when i tried that on ffmpeg, it worked first time. Well, 'worked'. No error thrown, but the resulting file is of wrong format [16:57] <sfan5> !pc Nopik [16:58] <cbenhagen> hi, i am trying to decode a raw frame from a phantom flex 4k camera. i am somehow unable to set the correct codec_tag on the commandline. rawvideo (BIT[0] / 0x544942) is what i need (http://git.videolan.org/?p=ffmpeg.git;a=commit;h=97bb0076c5bb1b30a33b911f8b…) but rawvideo ([186]GB[16] / 0x104247BA) is what gets set. how do i set this codec_tag? [17:02] <cbenhagen> http://pastebin.com/vyhgAHYC [17:04] <cbenhagen> this is what i get from the same frame in cine format. eg with a cine header http://pastebin.com/udvm0Qu9 [17:07] <cbenhagen> actually that was the output with cine header. sorry for the confusion: http://pastebin.com/L2d7s91V [17:24] <Phlarp> Can anyone assist me in consolidating these commands together? I feel like I am (unnecessarily) encoding things multiple times. http://pastebin.com/2k9ykEze [17:34] <sfan5> Phlarp: without figuring everything out you could "concat" the commands [17:34] <sfan5> like this: [17:36] <sfan5> ffmpeg <args> -c:v rawvideo -c:a pcm_s16le -f matroska - | ffmpeg -f matroska -i - <more args> final.mp4 [17:37] <c_14> Phlarp: just take all those filterchains and append them to each other making one nice large complex filtergraph [17:39] <Phlarp> c_14: How do I do this? I understand the idea, but I'm having trouble with the syntax, nothing I try seems to work [17:42] <c_14> I don't feel like doing all the writing so I'm going to be shortening a lot: -filter_complex '[0:v][1:v]overlay[..][o1];[o1]drawtext[..][o2];[o2][1:v]overlay[..][o3];[0:a]amix[..][a]' -map '[a]' -map '[o3]' finished.mp4 [17:46] <j53> hello, can someone help me [17:47] <j53> im trying to encode 1080i 25fps source video to 720p at 50fps [17:47] <j53> i have two issues, if i add cropping to my command line it works, but the video is 25fps [17:48] <j53> if i remove cropping from the command line, it is 50fps, but there is no crop [17:48] <j53> ok [17:50] <j53> http://pastebin.com/gQTUfgqB this is my script im working on [17:50] <j53> i will run it on a test video for console output, just a se [17:50] <Phlarp> c_14: I got them strung together. Thank you for the example, this makes much more sense now. [17:51] <c_14> j53: you can't have more than one -vf in a single command line, put both filters in the same -vf and separate with commas [18:43] <allengates> what the best audio Noise reduction library do you guys know? [18:55] <Keshl> I'm not sure what it's called, but whichever one Audacity uses is _really_ good, provided you have roughly 15 seconds of nothing but ambient noise. [19:03] <allengates> ok, I will try. [19:05] <rp__> Hi, anybody knows how to record live streams with current timestamps (as metadata) and then seek stream by passing timestamp as start position? [19:05] <rp__> I'm trying to set up streaming server with archive playback functionality in it. [19:08] <rp__> According to ffserver docs (https://www.ffmpeg.org/ffserver.html) it should be easy configurable but I've been stuck on this for a while. ("ffserver is a streaming server for both audio and video. It supports several live feeds, streaming from files and time shifting on live feeds. You can seek to positions in the past on each live feed, provided you specify a big enough feed storage.") [19:19] <tomjscott> Is there licensing information covering the distribution and use of the pre-built ffplay.exe app? I only see licensing regarding compiling the source into another app. [20:02] <Jaxan> I have a question about memory managment. If I understand correctly, when decoding, the AVFrame has a buffer. This buffer is allocated by calling avcodec_decode_video2, but AVCodec own it and destroys it for example on avcodec_close. Right? [20:02] <Jaxan> And when AVCodecContext.refcounted_frames I am the owner. Why was AVCodec designed to be the owner? performance? [20:16] <Jaxan> on a sidenote, is there any high-level documentation about these kind of things. I am particularly interested to see a list of 'open/close' or 'alloc/free' pairs, or some other lifetime-related documentation [20:45] <fajung> is there a way to add cover image to a output.mp4 ? [22:05] <Nopik> what should be the syntax to read file, change its pixel format to bgra, run that through filter, then output mpeg? I'm using ffmpeg -i file.mov -vcodec rawvideo -pix_fmt bgra -vf nopik out.avi -> works fine, produces bgra avi uncompressed. But the moment I change out.avi to out.mpg ffmpeg can't encode mpg anymore, getting me lots of buffer underflow errors [22:06] <Nopik> it works fine if i drop pix_fmt, but in such case the filter does receive yuv420 [22:06] <smo_> is it possible to do for exemple [22:07] <smo_> ffmpeg -i http://mirrorblender.top-ix.org/peach/bigbuckbunny_movies/big_buck_bunny_72… -sn -c:v libx264 -c:a libvorbis -f matroska bbunny.mp4 [22:07] <smo_> then start vlc immediately to read the bbunny.mp4 AND get the original video length [22:07] <smo_> when i do things like that my players always report NAN:NAN or 00:00 as length [22:07] Last message repeated 1 time(s). [22:07] <Nopik> ah, nevermind, I've been also using -vcodec rawvideo, when I dropped it and I dropped -pix_fmt it worked, since my filter already negotiated rgba [22:07] <smo_> so no seek nothing :( [22:07] <sfan5> smo_: why are you putting matroska into a .mp4 file? [22:08] <smo_> normaly i use liopus for audio not vorbis [22:08] <smo_> libopus [22:08] <sfan5> thats not what I mean [22:08] <sfan5> you are putting data using the matroska file format into an .mp4 [22:09] <smo_> what must i use [22:09] <sfan5> either put mp4 into mp4 [22:09] <sfan5> or use an mkv container [22:09] <sfan5> just omit the -f option [22:09] <sfan5> ffmpeg can auto-select the format [22:10] <smo_> will fix my NAN:NAN problem ? [22:10] <sfan5> proably not [22:38] <Nopik> how to play 2 streams simultaneously with single ffplay? e.g. I have in1.mpg and in2.mpg I would like to play them synchronously with single command. any visualization will do, e.g. 2 ffplay windows, or one window bigger, whereas those 2 videos are next to each other [22:49] <c_14> overlay filter [22:49] <Nopik> yeah, got it, thanks [22:50] <Nopik> can i force mpg/avi encoder to use something else than yuv420p? it seems to ignore all my tries to do so [22:50] <c_14> pixel formats usually depend on the codec not the container [22:54] <nr08> Quick question on audio conversion. I am running the following command: "ffmpeg.exe -i audio.mp3 -acodec pcm_s16le audio.wav". The initial audio.mp3 file is roughly 10mbs with a 128kbs bitrate. The output does not change the bitrate to the expected 16kbs. Is there something I am missing? NOTE: I have also attempted adding "-ab 16k" and "-b:a 16k [22:54] <nr08> " options and this does not solve the issue either. [22:55] <Nopik> c_14: thanks [22:58] <c_14> nr08: Going by my understanding of pcm, the bitrate should be 689kb/s [22:58] <c_14> Assuming a sampling rate off 44100Hz [22:59] <nr08> Executing "ffmpeg.exe -formats" displays the "pcm_s16le" format as a PCM signed 16-bit little-endian. Thats specified within the specific codec I am using to encode the new file. [23:01] <c_14> PCM 16 bits per sample times 44100 samples per second [23:01] <c_14> Though it could be 15 bits per sample with one bit being reserved for the sign [23:02] <c_14> In both cases you should get around 600+ kbps [23:04] <nr08> In attempt to create the desired output (8000k 1 channel 16kb/s), I also tried "ffmpeg.exe -i audio.mp3 -ac 1 -ar 8000 -ab 16k audio.wav" and the bitrate did not change. [23:05] <nr08> If this helps any, I am attempting to convert any mp3 files to a .wav format supported by asterisk. The specification in my previous message is what asterisk wants, in a PCM format. [23:06] <c_14> Like I just said, the bitrate of pcm audio depends directly and _only_ on the sampling rate and bit depth. For 16kbps audio you need with a bit depth of 16 bps, you need a sampling rate of 1024 [23:06] <c_14> Hz [23:07] <c_14> s/audio you need/audio [23:10] <nr08> I see. Thank you for your patience. [23:54] <ghospich> Yay, finally found nice workaround for subtitles in fast seek mode! [23:55] <ghospich> -vf setpts=PTS+60/TB,subtitles=sub.srt,setpts=PTS-STARTPTS [23:57] <ghospich> Crazy, but working. Workaround which provided in #2067 has some quality issues for me, but this one is working nice. [00:00] --- Fri Aug 15 2014

1 0

ffmpeg-devel.log.20140813
by burek 14 Aug '14

14 Aug '14

[00:11] <BBB> ubitux: ok fixed [00:15] <BBB> Compn: is ffmpeg's goal to be the fastest and support the most formats and feathres? is libav's goal to have clean code without hacks and to fix old api because of design restraints and limitations? - actually a fair point, I dont think its about goal, as much as about priority; in the end, both projects want both [00:15] <BBB> Compn: but not any less true obviously [00:30] <Compn> BBB : i dont think ffmpeg actually has goals or stuff like that [00:30] <Compn> :P [00:33] <Compn> but yes what you say makes sense. [00:41] <jamrial> libav has removed decoders they considered useless, like snow, so their definition of "most formats" is not very broad [00:42] <jamrial> they also removed libav* only features like mp3 header de/compress bitstream filter [00:42] <jamrial> meaning that a file you created with one version of libav you then couldn't read with the next [00:43] <j-b> jamrial: come on, snow... [00:43] <jamrial> yes, i agree it's not exaclty useful. just using it as an example of them not really bothering with supporting every format possible [00:46] <jamrial> the codec was there. they simply removed it to reduce maintenance burden. shows they really care about clean code *a lot* more than features [00:48] <Compn> hmm i am missing mails. [00:48] <Compn> whers the one michael replied to [00:56] <wm4> Compn: the cross-posting is getting a little out of hand [00:57] <wm4> ok who wants to maintain ffmpeg [00:58] Action: wm4 looks at j-b [00:58] <Compn> i asked him before, he said he didnt want to [00:58] <Compn> poor michael, my mail trolled too hard [00:58] <Compn> michaelni : my mail was meant to troll discussion between devels... [00:59] <Compn> not to make an insult on anyone [00:59] <Compn> i think you have done a great job with ffmpeg . a lot of the problems in the past are gone, not that they were your problems , but the project has grown a lot. [01:01] <j-b> wm4: do a linus way [01:01] <Compn> maybe i should have cc'd libav-devel :P [01:01] <j-b> subsystem submaintainers [01:01] <j-b> that merge patches [01:02] <wm4> so each lib gets a maintainer? that must be funny [01:02] <j-b> that get merge upstream to the true-upstream [01:02] <j-b> wm4: why so? [01:02] <wm4> e.g. lavc and lavf are closely coupled [01:03] <Compn> wm4 : per file maintainer ? [01:03] <wm4> I just can't imagine how that'd work [01:03] <Compn> which is pretty much what we have now... [01:03] <Compn> >MAINTAINERS [01:03] <j-b> wm4: so, how is that different from the kernel? [01:03] <wm4> Compn: not really, it implies maintainers for larger parts of the code, which merge from "lower" devs [01:04] <wm4> j-b: I guess the bottom line is that you need capable people as the "top most" maintainers [01:04] <j-b> agree. [01:04] <wm4> who will that be? [01:04] <j-b> depends for which part [01:04] <j-b> but for libav/subtitles, ubitux would be logical [01:05] <j-b> for libswscale, michaelni would be obivous [01:05] <wm4> not to say there are no capable people (at least technically), but everyone seems to mess with everything [01:05] <j-b> for libavresample/libswr are easy too [01:05] <j-b> libavformat/avio would be wbs [01:05] <j-b> libavutil people like elenril, or another german-mind [01:06] <j-b> for libavcodec/xiph- , it's not hard to find someone [01:06] <j-b> then you need one libavformat merger and one libavcodec merger [01:07] <j-b> for libavcodec/vp* BBB [01:07] <j-b> complex? sure. [01:07] <cone-847> ffmpeg.git 03James Almer 07master:7cafdfe9c799: lavfi: duplicate ff_log2_tab [01:08] <wm4> <j-b> then you need one libavformat merger and one libavcodec merger <- this seems to be the interesting part [01:09] <kierank> Well I would say filtering is more controversial [01:11] <BBB> I dont want to maintain anything [01:12] <Compn> hah two great mails from michael on debian-devel [01:13] <Compn> the 24h notice on mphq still burns on me, i can work with libav devs, just i cannot work on any server admin/hosted by libav... [01:14] <Compn> ah well [01:14] <Compn> cant think about split [01:14] <Compn> need to do something productive [01:14] Action: Compn afk [01:30] Action: kierank wonders whether to hit the send button in that thread [01:30] <J_Darnley> What harm can a little more fire do? [01:30] <Compn> oh man dont do it [01:31] <Compn> :P [01:31] Action: Compn rips keyboard in half [01:31] <wm4> Compn: weren't you doing productive things? [01:42] <BBB> kierank: do it! :) [01:44] <j-b> wm4: or just you fork, and we follow you [01:44] <cone-847> ffmpeg.git 03James Darnley 07master:0081a14e7d1e: lavc/flacenc: add sse4 version of the 16-bit lpc encoder [01:44] <wm4> j-b: you have more experience with open source, you should do it [01:54] <kierank> there [01:54] <kierank> i said it [01:54] <kierank> that is the elephant in the room [01:57] <wm4> so the problem is that michaelni isn't showing up on any conference? [01:57] <kierank> every attempt at vdd to move towards discussion ends with that [01:57] <kierank> Compn was there [01:57] <kierank> and others [01:58] <wm4> so why not host a conference at mini's home? [01:58] <kierank> google tried [01:58] <kierank> *this is not a joke* [01:58] <wm4> it's funny anyway [01:59] <Daemon404> at [01:59] <Daemon404> wat [01:59] <cone-847> ffmpeg.git 03James Almer 07master:a8592db9bb78: avcodec/idctdsp: make add/put_pixels_clamped_c internal functions [01:59] <kierank> Daemon404: you were at the meeting too [02:00] <Daemon404> i dont remember that part [02:00] <wm4> but I admit I don't understand at all why mini's _physical_ presence should be needed to resolve this [02:00] <kierank> because this is a social problem [02:00] <kierank> and you can't solve social problems hidden away [02:00] <kierank> almost by definition [02:15] <Compn> pfft [02:15] <Compn> kierank : you think being in person will change anything ? never. [02:15] <Compn> oh lol @ sending it to debian-dev [02:15] <Compn> ehe [02:15] <kierank> sent it to whatever [02:16] <kierank> i don't give a shit [02:16] <kierank> send it to the pope and he can help too [02:16] <kierank> and obama [02:16] <Compn> i mean you sent it there [02:16] <Compn> debian-dev needs all the flames it can take [02:16] Action: Compn bored now [02:16] <iive> is that invitation? [02:18] <wm4> there's a giant flamewar about the inclusion of ffmpeg in debian [02:18] <Compn> (also i dont think ffmpeg in debian will change anything) [02:18] <iive> i'm aware. package stuck in ftpqueue for over 3 months [02:18] <Compn> aside from not hurting users... [02:18] <iive> Compn: it would. [02:19] <iive> not for ffmpeg, but for libav. [02:19] <Compn> what would it do for libav ? [02:19] <iive> it would cut its water supply and let it try in the sun. [02:19] <iive> it [02:21] <iive> right now all projects that want to use multimedia in debian are forced to work with libav [02:26] <kierank> wm4: can't claim to be a leader and not appear ever [02:26] <kierank> that's anothe point [02:27] <Compn> lol [03:11] <BBB> kierank: nah, no angry emails, well said, showing up at conferences is useful [03:12] <BBB> michaelni: you should go to vdd this year, for real. ask carl eugen or so to shadow you if youre somewhat scared of the open world, its not that bad [03:13] <kierank> thierry will also be there (i assume) [03:54] <Timothy_Gu> jamrial: what's wrong with simply making log2_tab avpriv_? [04:01] <cone-847> ffmpeg.git 03James Darnley 07master:54a51d384055: lavc/flacenc: partially unroll loop in flac_enc_lpc_16 [04:03] <jamrial> it was duplicated for all libs that need it some time ago, commit 930c9d43. the commit message explains why [05:02] <cone-847> ffmpeg.git 03James Darnley 07master:0a52bbdc3161: cvcodec/flacdsp_lpc_template: osmetic fix [06:08] <cone-847> ffmpeg.git 03James Darnley 07fatal: ambiguous argument 'refs/tags/n0.10.15': unknown revision or path not in the working tree. [06:08] <cone-847> Use '--' to separate paths from revisions [06:08] <cone-847> refs/tags/n0.10.15:HEAD: cvcodec/flacdsp_lpc_template: osmetic fix [07:08] <jamrial> holy crap yasm 1.3.0 [09:19] <ubitux> kierank: i doubt michael going to a vdd or something will actually make any positive change [09:20] <ubitux> i'm sure you'd love to have an oral fight, but it makes very little sense [09:21] <ubitux> libav people are very used to such thing (they are basically the people who represented ffmpeg in the past in real-life meetings) [09:21] <ubitux> and even if you have the best neutral mediator in the fight, it makes no difference [09:21] <ubitux> people not comfortable with "the oral medium" don't stand a chance, at all [09:22] <ubitux> and even if it's the case, absolutely no additional information will be transmitted [09:22] <ubitux> and only "feelings" will be remembered from that talk [09:22] <ubitux> like "lol <X> totally clashed <Y>", "it was an awesome discussion" or stuff like that [09:23] <ubitux> on a personnal note, real life meetings are boring as hell, and IMO never actually helped much [09:24] <ubitux> the only benefits i see to such things are: put faces on nicknames, and communicate with corp who are too clumsy (or whatever reason) to use IRC or mails [09:24] <nevcairiel> most people that insist on them are those that know that they are more experienced and comfortable with them, and hope to win the argument that way [09:25] <ubitux> and used it as an argument later "we made monty admit ogg sucks IRL!!" [09:25] <ubitux> without any actual trace of anything [09:26] <kurosu_> Michael attending would be a good thing, but he would already have if he hadn't a major issue with it [09:27] <kurosu_> What if he is agoraphobic or anything that would make such a public event an actual pain to him ? [09:28] <ubitux> you don't need to have social anxiety for seeing only threat in going to such place [09:28] <kurosu_> He may also *not* want to speak about it [09:28] <ubitux> real life is the land of hypocrisy ;) [09:29] <kurosu_> yeah, I'm just stating that people are trying to get him to come, while his absence is not a ploy/laziness but a much more serious issue [09:29] <kurosu_> *may not be [09:31] <kurosu_> kierank, BBB: not blaming you at all for suggesting this, but please consider that other viewpoint (potentially rhetorical) [09:32] <ubitux> libav banned him from irc & mails to avoid the most they can any communication with him [09:33] <nevcairiel> i smiled at the comment in the mail thread about chatting happily until a certain libav developer came around [09:33] <ubitux> it's obvious that requesting for a real-life meeting is interpreted as a ploy [09:33] <ubitux> nevcairiel: hahaha i actually was there at that time [09:33] <ubitux> i didn't really understood why people suddenly stopped talking first, but then as i really love awkward situation like this i made sure to stay silent [09:35] <kurosu_> handling conflict face-to-face is difficult [09:36] <ubitux> what kierank suggest is probably in the spirit of smoothing relationship with random talks [09:36] <ubitux> but it's really boring as hell and makes absolutely no difference on the technical matters [09:37] <ubitux> and for people who only care about that, it's an insane waste of time [10:02] <ubitux> /usr/include/fribidi/fribidi-common.h:35:5: warning: "DONT_HAVE_FRIBIDI_CONFIG_H" is not defined [-Wundef] [10:02] <ubitux> #if DONT_HAVE_FRIBIDI_CONFIG_H+0 [10:02] <ubitux> ^ [10:03] <ubitux> grumbl wtf is this [10:30] <ubitux> ok, related to -Wundef [10:33] <ubitux> http://pastie.org/pastes/9469187/text sadness. [10:48] <J_Darnley> > DONT_HAVE_FRIBIDI_CONFIG_H [10:48] <J_Darnley> Oh my [10:49] <ubitux> (fyi https://bugs.freedesktop.org/show_bug.cgi?id=82548) [10:49] <J_Darnley> That must be confusung whenever it is zero. [10:50] <ubitux> i wonder what workaround this +0 is [10:56] <nevcairiel> to make it eval to 0 instead of empty if its unset? i dunno [11:45] <ubitux> any comment on the MV patch? [11:57] <kriegero1> hi, i'm in hurry now so i'll just notify you there: commit b4d4f8f8266e6407f21a62110905b27d3d3d71da breaks digest HTTP authentication. Will come back in few hours and try to fix or file a bugreport. [11:59] <ubitux> ah! i was wondering [11:59] <ubitux> i was testing the icecast thing again but it wasn't working [12:00] <ubitux> i wondered what was going on but didn't investigate, thx [12:00] <kriegero1> pls highlight me if anybody comes up with some patch in few hours [12:01] <ubitux> kriegero1: is it a regression from the merge or the commit itself? [12:04] <nevcairiel> there dont appear to be changes in the merge itself, so must be the commit [12:46] <BBB> ubitux: its not clumsy, its legal [12:46] <BBB> (bigcorp crap) [12:47] <ubitux> yes sure, whatever reason as i said :p [12:47] <BBB> :) [12:47] <ubitux> point is, michael isn't a bigcorp [13:27] <cone-539> ffmpeg.git 03James Almer 07master:9f61d6d8fb65: lavc/dnxhd: ff_dnxhd_cid_table is not exported [14:52] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:1e519b9d407f: avutil: turn arm setend into a cpuflag [16:10] <J_Darnley> Is there anyone here with XOP or AVX capable CPUs that is willing to test a couple of patches for me? [16:11] <ubitux> avx or avx2? [16:12] <J_Darnley> uh, I'm not sure [16:12] <J_Darnley> whichever has integer ops for ymm registers [16:13] <ubitux> avx2 then, ok [16:13] <michaelni> ubitux, "[FFmpeg-devel] [PATCH 2/2] lavfi/avf_showspectrum: fix output pts computation." is missing a review (its just a 1 line patch) [16:13] <ubitux> J_Darnley: do you want a ssh on a box so you can test it yourself? [16:13] <J_Darnley> That would be nice, yes. [16:13] <ubitux> J_Darnley: can you send me an ssh public key [16:13] <ubitux> ? [16:14] <J_Darnley> Yes, do you have a key size requirement? [16:14] <ubitux> michaelni: didn't i answered in that thread? [16:14] <ubitux> J_Darnley: not really [16:15] <J_Darnley> (where did I put them?) [16:15] <cone-539> ffmpeg.git 03Paul B Mahol 07master:e3212bfd1493: avfitler/vf_perspective: support slice threading [16:15] <michaelni> ubitux, you did but it wasnt applied, so maybe you should reply to each patch individually or just push yourself [16:24] <J_Darnley> ubitux: sent via email [16:25] <ubitux> perfect, give me a few minutes [16:48] <ubitux> michaelni: i'd better let nicolas do it [16:48] <ubitux> michaelni: i'll try to find time to test [16:48] <ubitux> if you insist :P [17:03] <J_D> Wow! Look how fast configure is! [17:05] <Compn> ? [17:06] <J_D> Wrong window (sort of) [17:07] <J_D> I'm connected to ubitux's laptop and am amazed at how much fater configure is on not windows [17:07] <J_D> *faster [17:08] <ubitux> J_D: can be even faster if you dash ./configure [17:08] <ubitux> J_D: ah, forgot to tell you: --cc='ccache cc' is your friend [17:12] <Compn> oh yes, windows configure is terrible. [17:31] <J_Darnley> Oh, is that my problem? [17:32] <J_Darnley> Why doesn't the SPLATD macro set every dword in a ymm reg to the value it should be? [17:32] <J_Darnley> Is this the silly "split register" thing? [17:35] <ubitux> J_Darnley: your problem? i'm just suggesting ccache to avoid wasting time when recompiling [17:35] <J_Darnley> No sorry [17:36] <J_Darnley> That's me debugging [17:36] <J_Darnley> I am vaguely aware of ccache [17:46] <kurosu> J_Darnley, pass --cc="ccache gcc" to configure and be done with it [18:01] <J_Darnley> ah ha! I need vbroadcastss [18:18] <cone-539> ffmpeg.git 03Stefano Sabatini 07master:a539f1cbbd8f: doc/filters: correct confusing statements about *showinfo shown values [18:33] <cone-539> ffmpeg.git 03Martin Storsjö 07master:8bf3bf69ad73: http: Stop reading after receiving the whole file for non-chunked transfers [18:33] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:9f7a2ecb2926: Merge commit '8bf3bf69ad7333bf0c45f4d2797fc2c61bc8922f' [18:45] <cone-539> ffmpeg.git 03Martin Storsjö 07master:428b0578c642: w32threads: Use newer thread synchronization functions when targeting Vista [18:45] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:a0f3db8f60e0: Merge commit '428b0578c64241fc677fed7083cc8fe65e10f32e' [19:02] <cone-539> ffmpeg.git 03Martin Storsjö 07master:72199e1041b4: configure: Check for nanosleep in headers as well, not only in libs [19:02] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:cf24e10a0920: Merge commit '72199e1041b4052544e181ad90b431ca6704da12' [19:37] <cone-539> ffmpeg.git 03Clément BSsch 07master:c85e0ca68a3d: avfilter/dctdnoiz: add slice threading [19:52] <ubitux> Daemon404: if you want to retry dctdnoiz, it's now usable [19:52] <ubitux> (from a speed PoV) [19:52] <ubitux> Daemon404: also, it has 8x8 and 16x16 block sizes, so feel free to compare [19:53] <Daemon404> is there any degredation when using slices [19:53] <Daemon404> i guess there wouldnt be [19:56] <ubitux> no, same quality [19:57] <ubitux> each slice recomputes a few blocks above and after for its context [20:02] <cone-539> ffmpeg.git 03Clément BSsch 07master:8fcd9386ec79: Changelog: notify about the recent performance boost in dctdnoiz filter [20:05] <michaelni> j-b, did you find any sub system maintainer volunteers? That is should i add someone to MAINTAINERs for libavformat/codec/... ? [21:00] <cone-539> ffmpeg.git 03Ronald S. Bultje 07master:14e302551812: vp9: ignore reference segmentation map if error_resilience flag is set. [21:04] <cone-539> ffmpeg.git 03Clément BSsch 07master:ec14f53dbd02: reset and complete RELEASE_NOTES for the next version [21:09] <j-b> a1m [21:09] <j-b> oops [21:10] <j-b> michaelni: that was just an idea. [21:10] <j-b> michaelni: but that requires important changes that I doubt you are ready for. [21:11] <durandal_1707> like what? [22:00] <J_Darnley> How does the log file for that ticket end up being 500 MB? [22:11] <kurosu> The input file is a 850MB audio file, and the log level is at its max [22:30] <J_Darnley> oh yeah, some formats print for every packet [22:31] <cone-539> ffmpeg.git 03Andrew Stone 07master:afbd4b7e093a: lavf: add AVFormatContext/AVStream fields for signaling to the user when events happen. [22:31] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:ed8e6d535cf4: Merge commit 'afbd4b7e093adf6d7a830b32759ca3ba8500363d' [22:32] <jamrial> he says the audio is not written to the file until conversion is complete, and kept in memory instead [22:32] <jamrial> if the output file has both audio and cover art, there's a posibility that the mp3 muxer buffering the audio packets waiting for the cover art stream is causing this [22:47] <cone-539> ffmpeg.git 03Andrew Stone 07master:a8c104a511f9: AVOption: add support for dictionary types. [22:47] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:2ba45778095c: Merge commit 'a8c104a511f97e4ea617df73b31737e28a8a5126' [22:53] <cone-539> ffmpeg.git 03Andrew Stone 07master:7601f9412a2d: http: export icecast metadata as an option with name "metadata". [22:53] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:69a5cd9feac5: Merge commit '7601f9412a2d3387617a45966b65b452a632c27a' [22:59] <cone-539> ffmpeg.git 03Andrew Stone 07master:7e38903b5c86: http: enable icy metadata by default. [22:59] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:313d75cd4300: Merge commit '7e38903b5c86a759549e70647ae42bb22d353b14' [23:05] <J_Darnley> ubitux: I don't know if there's an easy to install package for you but can I request that you install the GNU time utility? [23:07] <cone-539> ffmpeg.git 03Andrew Stone 07master:019d3fccc4dc: Set protocol-level metadata in AVFormatContext any time a packet is read. [23:07] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:f01a1281ee82: Merge commit '019d3fccc4dcf5c8379112f697ce9eb08edee9b9' [23:19] <wm4> so, should the old metadata update API be deprecated? [23:36] <cone-539> ffmpeg.git 03Andrew Stone 07master:93c04e095dc3: Expose metadata found in onCuePoint events in .flv files. [23:36] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:9f97f7c54638: Merge commit '93c04e095dc37ebdab22174e88cfa91e24940866' [23:43] <cone-539> ffmpeg.git 03Andrew Stone 07master:0f789322efa7: flvdec: update AVFormatContext.event_flags with METADATA_UPDATED whenever metadata changes. [23:44] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:314c198413bf: Merge commit '0f789322efa78a672e4c3027e5cc12b8a947043a' [23:50] <cone-539> ffmpeg.git 03Andrew Stone 07master:fa3a5dd4dea3: nutdec: update AVFormatContext.event_flags with STREAM_/METADATA_UPDATED whenever metadata changes. [23:50] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:1110a030c732: Merge commit 'fa3a5dd4dea34baa6bb2f7fe6006fc4b2888f2aa' [23:56] <cone-539> ffmpeg.git 03Andrew Stone 07master:cc3e88a2b9e7: mov: update AVFormatContext.event_flags with METADATA_UPDATED whenever metadata changes. [23:56] <cone-539> ffmpeg.git 03Michael Niedermayer 07master:a90364d001eb: Merge commit 'cc3e88a2b9e7ecf62e4ea1c41ce1623cea67ee96' [00:00] --- Thu Aug 14 2014

1 0

ffmpeg.log.20140813
by burek 14 Aug '14

14 Aug '14

[00:52] <c_14> fajung: try removing `-flags +aic+mv4' from the commandline [00:57] <fajung> is there any way to make the output the same the same bitrate as the input? so I don't have to keep changing it -b:v xxxxk [00:59] <c_14> no ffmpeg-internal way that I know of and if the input uses variable bitrate encoding, no good way unless the encoder writes pertinent settings into the metadata [02:50] <kline> is there some sort of political issue with ffmpeg and debian? im seeing in various debian places notes saying that ffmpeg is deprecated, but the site shows recent blog entries [02:52] <kline> oh gg [02:52] <kline> cant help but think you guys got a little shafted because the package maintainer fell on the other side [02:56] <sacarasc> IIRC, Debian is getting ffmpeg back soon. [02:56] <c_14> If the flamewar ever ends, that is. [05:20] <jrgill> Is there any common issue where the native AAC encoder renders excessive artifacts for stereo tracks? The output is unusable at 256 Kbps stereo while 192 Kbps mono is fine. [05:21] <c_14> The encoder is experimental, there might be issues. What version of FFmpeg are you running? [05:22] <jrgill> ffmpeg version N-65366-g1c59419 Copyright (c) 2000-2014 the FFmpeg developers [05:25] <jrgill> I'll give libvo_aacenc a try. Working on Windows so that's my only other option right now. [05:26] <jrgill> Will update back here soon. [05:28] <jrgill> I checked this a while ago but not in full. Maybe it's related. https://trac.ffmpeg.org/ticket/2686 [05:34] <jrgill> Wow, apparently this is a big issue. Just noticed the scrollbar, heh. [06:09] <jrgill> Well, at least libvo_aacenc consistently sounds like crap. Will try 192 Kbps stereo with native encoder... [07:11] <jrgill> Does libmp3lame support -r:a 48k? [07:14] <relaxed> jrgill: yes, see "ffmpeg -h encoder=libmp3lame" [07:15] <jrgill> Thanks, checking now. [07:20] <jrgill> Is any of this invalid? -c:a libmp3lame -r:a 48k -b:a 320k -af pan="stereo|c0=c0|c1=-1*c1" [07:20] <jrgill> AAC-LC 320 Kbps input [07:21] <jrgill> 48k [07:21] <relaxed> why are you encoding it? [07:22] <jrgill> I need to invert the right channel and the included AAC encoders suck. [07:23] <jrgill> Other times it will be PCM, so just this one time re-encoding. [07:24] <relaxed> you probably want single quotes around the whole filter [07:34] <jrgill> I'm not seeing options for CBR and VBR. http://privatepaste.com/download/59de9f1057 [07:35] <relaxed> -q:a 0 would be libmp3lame's highest quality vbr [07:35] <relaxed> cbr would be -b:a $value [07:37] <jrgill> Well, let's see what happens with -q:a 0. [07:45] <towski> I think the recent debian ffmpeg changes broke a script I run [07:46] <towski> I maybe figured out some of the changes, but I still get [07:46] <towski> Cannot find a matching stream for unlabeled input pad 1 on filter Parsed_overlay_0 [07:46] <towski> this is the command I'm running [07:46] <towski> http://pastebin.com/xPE4ggru [07:47] <towski> also please let me know if I need to be asking in avconv, I'm not really clear on the different [07:54] <relaxed> towski: what is that filter supposed to be doing? [07:55] <relaxed> -q:v 3 does nothing there [07:59] <towski> relaxed: it used to be -qscale 3 [07:59] <towski> but that threw an error that it was ambiguous [08:00] <towski> here is the command that used to work: http://pastebin.com/Yn0x7tqz [08:00] <towski> I changed -f pulse since that didn't seem to work anymore to -f alsa -i pule [08:00] <towski> pulse* [08:00] <relaxed> towski: ask in #libav [08:03] <towski> relaxed: thanks. do you have any info on what changed recently in debian testing? I saw an email about a big change and assume that's what broke this [08:03] <towski> it could just be a newer version [09:36] <jrgill> I guess M4V can't contain MP3 audio. Figures... [10:02] <polysics> hello! cna I use the concat filter so that I can concatenate portions of different files? [10:03] <polysics> example: I have two 10 second videos. I want the full first video and 5 seconds of the second, concatenated [10:03] <polysics> I was hoping that could be done in one command [10:08] <termos> How can I set the volume per channel using ffmpeg audio filters? I looked at the volume filter and it doesn't seem to be doing exactly this [10:25] <polysics> the concat demuxer is also behaving weirdly: it stops the stream at the longest clip, ie. if I use a 5sec clip and a 10sec clip, it plays the full first clip and 5 seconds of the second [10:27] <polysics> I also think I am using the duration directive in there wrong - it does not truncate files, does it? [10:56] <polysics> hmm [10:57] <polysics> https://gist.github.com/polysics/31ad6235f1219cdfb7a1 [10:57] <polysics> gets me [10:57] <polysics> [Parsed_trim_1 @ 0x7fb959500560] Media type mismatch between the 'Parsed_trim_1' filter output pad 0 (video) and the 'Parsed_concat_3' filter input pad 1 (audio) [10:57] <polysics> [AVFilterGraph @ 0x7fb95942c1e0] Cannot create the link trim:0 -> concat:1 [10:58] <polysics> I am not sure what I am doing wrong, trying to add audio to those files and join them [11:03] <polysics> or maybe concat does not do what I think it does? :) [11:23] <polysics> shouldn't this work now? https://gist.github.com/polysics/37582ec1c7707fc3b448 [11:30] <polysics> oh, Atrim [11:30] <polysics> great, now one last thing: I would like to encode this in very high quality h264 [11:30] <polysics> any recommended settings? [11:32] <DopeLabs> -c:v libx264 [11:32] <relaxed> polysics: https://trac.ffmpeg.org/wiki/Encode/H.264 [11:33] <polysics> resulting files look like they are a bit too big- but maybe I could encode to 720p or even 480p [11:33] <polysics> 1080p is large no matter what, I assume [11:34] <DopeLabs> depends on the bitrate, fps, etc [11:34] <sfan5> I'd suggest keeping the input res. [11:35] <sfan5> Higher res. is usually better [11:37] <slowguy> http://pastebin.com/FVYjBtrd i am giving this command but ffmpeg says No such filter: '"crop' [11:38] <polysics> 4s file at 1080p is 20Mb [11:39] <Diogo> hi i need to convert a video file .avi into hls file(h264 aac) in shorter time possible? like run multiple instance of ffmpeg ? [11:47] <fling> How to create a hyperlapse? [11:48] <DopeLabs> i think thats some new hotness developed by microsoft [11:49] <ubitux> patch welcome, but please something testable (aka a filter that doesn't take hours to process 10 seconds of video) [11:49] <fling> DopeLabs: I see old hyperlapse videos on youtube [11:50] <DopeLabs> ah mkay [12:00] <AurelPop> Hi, I'm writing a small library in C, which is supposed to take frames from rtmp stream. Sometimes I get "Stream discovered after head already parsed" and the process just hangs... Can somebody explain to me exactly why? and how can I solve this? .... ome timeout and try again?... [12:13] <t4nk460> Hi [12:27] <AurelPop> Hi, I'm writing a small library in C, which is supposed to take frames from rtmp stream. Sometimes I get "Stream discovered after head already parsed" and the process just hangs... Can somebody explain to me exactly why? and how can I solve this? .... ome timeout and try again?... [13:10] <fl0bbe> why hallo there :) [13:19] <Fyr> how to convert an audio file into raw format? [13:20] <Fyr> I have an m4a and a flac, I want to compare their raw audio. [13:20] <spaam> ffmpeg -i superDuperFile.m4a SuperDUper_sacarasc.wav ? [13:22] <Fyr> their md5sums have a difference. =( [13:23] <AurelPop> How can I get the behavoir of "stimeout" from code? [13:23] <sacarasc> Fyr: They will, because AAC is lossy, so some of the information is lost when encoded. [13:23] <Fyr> ALAC [13:23] <Fyr> m4a was encoded with ALAC. [13:24] <Fyr> how can I compare the quality of two files? [13:25] <Fyr> I've already converted FLAC-ALAC-FLAC and found that their md5-sums coincide. [13:25] <spaam> but are they converted from the same source? [13:26] <Fyr> it's what I need to find out. [13:32] <Fyr> they maybe converted from the same source, but one file can be louder. [14:13] <fajung> is it possible to force as default subtitle.srt ? like mkvmerge does [14:14] <fajung> I've this code:"-i '~/input.avi' -sub_charenc UTF-8 -i '~/input.srt' -c:s:0 mov_text -metadata:s:s:0 language=spa -metadata:s:s:0 default-flag=0 -acodec libfdk_aac -metadata title="title...." -vcodec libx264 -b:v 950k '~/out.mp4'" [16:38] <kode54> is anyone interested in looking at a WMA Lossless file that breaks in the decoder? [16:38] <kode54> it's four samples long, which should at least produce one packet of audio [16:39] <kode54> but the decoder hits the first real packet of data (13KB) and simply returns without producing any samples [16:39] <kode54> and since the duration is not actually 1ms or more, the file's headers report a duration of 0 [16:58] <nicoit> Can anybody give me any example(c++ code) how to use ffmpeg to decode video using dxva2? [17:15] <SkoobyNoob> does anyone know how to turn a group of mp4 files into mp3s with ffmpeg or do i have to use a script [17:16] <sacarasc> A 1 line script. [17:16] <Chaz6> for i in *.mp4; do ffmpeg -i $i $i.mp3; done [17:16] <BammBamm> wow, freaky, this was my question almost exactly [17:16] <SkoobyNoob> ok [17:17] <BammBamm> how can i use this one ( ffmpeg -i video.mp4 -vn -acodec copy audio.m4a) in WinFF? [17:17] <Mavrik> of course that command line will murder your audio quality :P [17:17] <Chaz6> yep choose your options wisely [17:18] <SkoobyNoob> i tried ffmpeh -i *.mp4 -f mp3 -ab 192000 -vn '%(mp3)s' [17:18] <BammBamm> how can I set up WinFF so that it rips LOSSLESSLY the m4a outta the mp4s? [17:18] <SkoobyNoob> but that made my laptop freeze [17:18] <relaxed> BammBamm: we don't support WinFF here [17:18] <BammBamm> crap [17:19] <sacarasc> BammBamm: for i in *.mp4; do ffmpeg -i "$i" -c:a copy -vn "$i".audio.m4a; done [17:19] <relaxed> SkoobyNoob: on windows? [17:19] <sacarasc> Play around with more bash to get it done properly. [17:19] <SkoobyNoob> no on ubuntu [17:19] <SkoobyNoob> i got ubuntu 14.4 [17:20] <relaxed> for i in *mp4; do ffmpeg -i "$i" -q:a 0 "${i%.*}".mp3; done [17:20] <SkoobyNoob> hmm [17:20] <relaxed> er, wait! [17:20] <SkoobyNoob> yes that makes more sense [17:20] <relaxed> for i in *mp4; do ffmpeg -i "$i" -map 0:a -q:a 0 "${i%.*}".mp3; done [17:20] <SkoobyNoob> could you explain what each option does please? [17:20] <SkoobyNoob> i know -i [17:21] <relaxed> bourne shell for loop, -q:a 0 == high quality vbr [17:22] <SkoobyNoob> cool [17:22] <SkoobyNoob> what about -map? [17:22] <relaxed> map only the audio (which wasn't really needed) [17:22] <SkoobyNoob> since its already turning it into a mp3? [17:23] <relaxed> correct [17:23] <SkoobyNoob> so i dont have to suppress with -vn? [17:24] <SkoobyNoob> ok is working [17:24] <SkoobyNoob> thanks i have been looking for this answer for days [17:24] <SkoobyNoob> ok another dumb question [17:24] <SkoobyNoob> how can i keep the same thumbnail [17:25] <SkoobyNoob> from the video on the mp3? [17:25] <BammBamm> sacarasc: thanks. But when I tried that, it says "i unexpected at this moment" [17:25] <SkoobyNoob> you are using WINFF BammBamm ? [17:25] <BammBamm> yes [17:26] <SkoobyNoob> man you should just come to the dark side [17:26] <BammBamm> SkoobyNoob: no [17:26] <SkoobyNoob> we have cookies [17:26] <sacarasc> Then you'd also get more support here. :p [17:26] <BammBamm> that command sacarasc gave i tried in a dosbox via ffmpeg own batchfile [17:27] <SkoobyNoob> hm [17:27] <SkoobyNoob> hmmmm [17:27] <SkoobyNoob> hmmmmmmmmm [17:27] <BammBamm> ff-prompt bat [17:27] <sacarasc> BammBamm: It was a bash line, not a CMD line. [17:27] <BammBamm> aaah [17:27] <BammBamm> bash = linux, yes? [17:27] <SkoobyNoob> you gotta use the bourne again friend [17:27] <sacarasc> You can get bash on Windows, BSDs, others, too. [17:28] <BammBamm> i used this succesfully in dosbox : ffmpeg -i video.mp4 -vn -acodec copy audio.m4a [17:28] <BammBamm> but how do i make that for a bunch of files [17:28] <relaxed> google windows batch scripting [17:29] <BammBamm> relaxed: i did, and this irc-channel popped up .... ;) [17:30] <sacarasc> BammBamm: http://stackoverflow.com/questions/2591758/batch-script-loop [17:31] <sacarasc> Or you could try powershell! [17:32] <fajung> if I add a subtitle.srt into an output.mp4, is there any way to force as default, like in the mkvmerge does ? [17:34] <BammBamm> sacarasc: that's a bit above my pay grade. my point is, now, WinFF accepts FFmpeg commands, so, how would I go about lossless audio extraction? [17:34] <sacarasc> WinFF is not supported here, and you already had it. [17:35] <sacarasc> [16:17:21] <BammBamm> how can i use this one ( ffmpeg -i video.mp4 -vn -acodec copy audio.m4a) in WinFF? [17:35] <sacarasc> That would be lossless. [17:35] <BammBamm> sacarasc: that bash thing you gave me at first, can you convert it to something that would work in a dosbox [17:35] <BammBamm> sacarasc: OK sorry bout that [17:36] <sacarasc> I have never done any batch scripting, and if it's above your paygrade, how much do I get paid to learn? :D [17:36] <BammBamm> kisses [17:36] <BammBamm> ;) [17:39] <BammBamm> so, this should work? [17:39] <BammBamm> for /l %x in (1, 1, 100) do ( [17:39] <BammBamm> echo %x [17:39] <BammBamm> ffmpeg -i *.mp4 -vn -acodec copy audio.m4a [17:39] <BammBamm> ) [17:40] <BammBamm> ??? [17:40] <BammBamm> sacarasc: ? [17:40] <sacarasc> [16:36:19] <sacarasc> I have never done any batch scripting, [17:42] Action: BammBamm thinks she was counting on sacarscs natural intelligence and chivalry, although not necessarily in that order ..... [17:42] <BammBamm> ;( [17:42] <BammBamm> ;) [17:42] <sacarasc> You missed my natural laziness and aversion to learning. [17:43] <BammBamm> sacarasc: I soooooooooo picked up on that, [17:44] <BammBamm> i just am unrealistically optimistic [17:52] <SkoobyNoob> well BammBamm i thought u were a dude [17:52] <SkoobyNoob> so that is were the Chivalry cuts off for me [17:52] <BammBamm> 95% chance you were right [17:52] <BammBamm> so i don't bl [17:52] <SkoobyNoob> 95%? interesting [17:53] <BammBamm> this stupid batch keeps giving me "x is unexpected at this time" etc [17:53] <SkoobyNoob> i wonder if women or users with female nicks have better luck [18:12] <nicoit> did anybody see my question? [18:12] <nicoit> "(17:58:15) nicoit: Can anybody give me any example(c++ code) how to use ffmpeg to decode video using dxva2?" [18:21] <saste> nicoit, http://trac.ffmpeg.org/wiki/HWAccelIntro [18:21] <saste> also, did you see ffmpeg_dxva2.c in the source code? [18:24] <BammBamm> so, can FFmpeg do batch lossless audio ripping/extraction? [18:24] <BammBamm> if yes, how? [18:25] <Fjorgynn> I use EAC for that on Windows [18:26] <sacarasc> BammBamm: FFmpeg itself can't, but you can write a script to do it. [18:28] <slowguy> i have a video which has dvb_teletext based subtitles..ffmpeg would say decoder dvb_teletext not found [18:28] <slowguy> but i want just audio and video in output streams [18:28] <slowguy> can't i tell ffmpeg to ignore the teletext stream? [18:29] <sacarasc> -sn [18:29] <slowguy> in all output streams? [18:29] <slowguy> okay got it thank you [18:48] <RenatoCRON> to com dor [18:48] Action: RenatoCRON sorry, wrong window! [18:54] <fajung> is there any way to force a subtitle.srt as default track, like in the mkvmerge does ? [19:00] <BammBamm> sacarasc: HOW would i go about writing scripts for that? is there a ..... website? ?? [19:01] <sacarasc> There are probably many that can explain a for loop for batch or maybe a foreach loop for powershell. [19:06] <circ-user-w3x6R> I wish to post & subscribe to ffmpeg-user(a)ffmpeg.org see https://lists.ffmpeg.org/mailman/listinfo/ffmpeg-user/, but apparently something goes wrong what. "To post a message to all the list members, send email to ffmpeg-user(a)ffmpeg.org" I did, but my post doesn't appears in https://ffmpeg.org/pipermail/ffmpeg-user/2014-August/thread.html [19:07] <circ-user-w3x6R> Same issue with subscription, send email, but nothing [19:36] <pmac_> Is there any way to have FFMPEG get itself all setup, spesifically for connecting to an RTMP output, and hold in that configruation until some signal is sent to start it? [20:46] <Phlarp> http://pastebin.com/MMpPwtL1 Say I had this command, and wanted to incorporate an adelay option so the mixed audio stream doesn't start for 3 seconds, how would I add that? [22:25] <c_14> Either just add a ,adelay=[foobar] to the end of the filter_complex, or create a silent stream with aevalsrc of the length you want and then concat them [00:00] --- Thu Aug 14 2014

1 0

ffmpeg-devel.log.20140812
by burek 13 Aug '14

13 Aug '14

[00:09] <kurosu> michaelni, the prores patchset is also under review in libav - so you may want to wait before applying if a valid change is needed and requested [00:09] <kurosu> the prores patchset was tested in ffmpeg too (there's a small conflict) [00:11] <kurosu> patches 2/3 could be merged, but are somewhat different problems [00:12] <kurosu> patch 2 could be considered incomplete but the additional fix would be canceled by patch 3 anyway [00:12] <mark4o> perhaps that prores patch fixes https://trac.ffmpeg.org/ticket/2760 [00:12] <kurosu> mark4o, that's the one being fixed ;) [00:12] <mark4o> ah nice [00:13] <kurosu> see "[PATCH 3/4] proresenc_kostya: realloc if buffer too small" [00:15] <mark4o> I filed that over a year ago and thought that surely it had been forgotten [00:15] <kurosu> it had been [00:16] <kurosu> I just took a liking in checking the bugs and it was one of the simplest [00:32] <mark4o> kurosu: It looks like warn is never changed from its initial value of 0. Although Im not sure why it requests a sample when there is already one in the ticket. [01:08] <ross__> Hi all. I have had FFMPEG crash and opst an error in syslog "kernel: [584702.632810] ffmpeg[22864]: segfault at 0 ip 0000000000c5ff75 sp 00007fff6389edd8 error 6 in ffmpeg[400000+c1e000]" running command line "screen -d -m -S RT ffmpeg -i "udp://@224.2.9.4:1234?fifo_size=10000&overrun_nonfatal=1" -c:v libx264 -me_method epzs -threads 16 -r 25 -g 45 -bf 2 -trellis 2 -cmp 2 -subcmp 2 -s 720x576 -preset faster -filter:v yadif -b:v 800k -bt [01:08] <ross__> 200k -acodec mp3 -ac 2 -ab 128k -ar 44100 -async 1 -y -hls_time 10 -hls_list_size 4000 -hls_wrap 4000 -start_number 0 -f hls /mnt/ramdisk/rt.m3u8". What other information do I need to create a meaningful bug report?? [01:34] <michaelni> ross__, see https://www.ffmpeg.org/bugreports.html [01:34] <michaelni> basically we need a reproduceable testcase [02:40] <cone-620> ffmpeg.git 03Christophe Gisquet 07master:bf10f09bccdc: proresenc_kostya: remove unneeded parameters [02:40] <cone-620> ffmpeg.git 03Christophe Gisquet 07master:52b81ff4635c: proresenc_kostya: report buffer overflow [02:40] <BBB> ubitux: looks like a bug in vp9_ipred_hd_32x32 [02:40] <BBB> ubitux: havent looked closely, maybe over the weekend [02:40] <BBB> or maybe someone else wants to have a look [02:41] <BBB> ubitux: there may be more bugs, if I disable that function, I still see mismatches further down [02:43] <BBB> (and they seem not simd related) [03:16] <Youka> Has GCC (not MinGW) pthread_time.h too? [03:45] <cone-620> ffmpeg.git 03Daniel Oberhoff 07master:9f617a14a008: avfilter: ported lenscorrection filter from frei0r [04:02] <cone-620> ffmpeg.git 03James Almer 07master:744f15b6b380: lavu: rename ff_opencl_set_parameter() to avpriv_opencl_set_parameter() [04:16] <hotwings> wow, doing an ffmpeg clone was so slow it actually ended with with fatal: read error: Connection reset by peer [04:26] <jamrial> try the github mirror instead [04:30] <hotwings> i wonder how github managed to mirror it considering how slow it is [04:30] <hotwings> yeah, github mirror breezed along at 6MB/sec. much better than 6k/sec [04:32] <Compn> bring up any hosting problems with j-b [04:36] <hotwings> ok, will msg him [04:45] <cone-620> ffmpeg.git 03James Almer 07master:d6711ee648f8: lavu: stop exporting internal functions [04:53] <cone-620> ffmpeg.git 03Christophe Gisquet 07release/2.2:7740b111dd9a: proresenc_kostya: remove unneeded parameters [04:53] <cone-620> ffmpeg.git 03Christophe Gisquet 07release/2.2:1ad1723c24cd: proresenc_kostya: report buffer overflow [05:18] <Youka> I have a small but important patch: http://pastebin.com/6viJFsgc [05:18] <Youka> How should i submit him (without going through the complex and long process on the developer page). [05:23] <cone-620> ffmpeg.git 03Michael Niedermayer 07release/2.2:49fa398858df: Changelog: add entry for proresenc [05:26] <michaelni> Youka, post it to the ffmpeg-devel mailing list [05:31] <michaelni> Youka, not sure you saw my reply as you did "Ping timeout", but submit the patch to the ffmpeg-devel mailing list please [05:31] <Youka> ok, thx michaelni [05:32] <cone-620> ffmpeg.git 03James Almer 07fatal: ambiguous argument 'refs/tags/n2.2.7': unknown revision or path not in the working tree. [05:32] <cone-620> Use '--' to separate paths from revisions [05:32] <cone-620> refs/tags/n2.2.7:HEAD: lavu: stop exporting internal functions [05:42] <Compn> Youka : just mail it to ffmpeg-devel(a)ffmpeg.org , if you cant figure out how to subscribe , i can manually approve your message to the list :P [05:42] <Youka> Already subscriped, posting it next. Thanks for help. [05:45] <Compn> no problem :) [05:45] Action: Compn falls aslwepep [07:43] <kurosu> mark4o, for warn, indeed (debug code removal accident) and anyway I was asked to move it to the header [07:45] <kurosu> for the sample request, whenever that condition is hit (ie irrespective of the specific cause in your bug report), the project wants to know about it and fix the allocation in light of the new troublesome input [11:13] <nevcairiel> ohh, after ages a new yasm release [11:22] <cone-847> ffmpeg.git 03Carl Eugen Hoyos 07master:fddea3f074d9: Check for hevc startcode when muxing into mpeg-ts. [11:22] <cone-847> ffmpeg.git 03Carl Eugen Hoyos 07master:99867fc0c42a: Fix warning if https protocol was requested but isn't available. [11:36] <cone-847> ffmpeg.git 03Martin Storsjö 07master:223c374a9d97: configure: Check for nanosleep in headers as well, not only in libs [12:38] <cone-847> ffmpeg.git 03James Almer 07master:e20ff251a688: x86/ttadsp: remove an unnecessary mova [13:08] <ubitux> BtbN: you were working on nvenc at some point right? [13:09] <cone-847> ffmpeg.git 03James Almer 07master:c97870d1a10c: x86/dca: remove unused header [13:18] <BBB> ubitux: if youre interested in the other bug, its frame 5, col 12, row 7 (decoding 6), 16x16 block, tx8x8 block bottomleft; I dont think its prediction, since the rest of the block is fine (but worth testing anyway), but it looks like its the transform (both c and simd would have a bug if thats correct) [13:18] <ubitux> i was surprised at your patch since the bug was also affecting -cpuflags none [13:18] <ubitux> but i guess you still have glitch [13:31] <ubitux> BBB: i don't think i'll have time to look into it today [13:38] <BBB> ubitux: yeah, looks like theres 2 bugs at least [13:38] <BBB> still looking into the other one [13:40] <ubitux> durandal_1707: still have a monocore cpu? [13:40] <durandal_1707> atm i'm on monocore [13:40] <ubitux> durandal_1707: it makes me sad for you you know [13:48] Action: michaelni is in favor of ffmpeg-spi funding modern hw for ffmpeg developers who dont have modern hw [13:49] <ubitux> we can buy 2 or at best 3 decent laptop [13:49] <ubitux> afaict [13:50] <ubitux> maybe 5-6 for basic ones [13:51] <michaelni> what do you call "decent laptop" ? [13:54] <ubitux> dunno, something like one of the T serie @ lenovo (though, i really hate what they did recently) [13:55] <ubitux> BBB was talking about an apple hw with avx2, but that's really expensive [13:55] <ubitux> a "basic one" can be enough for most of the stuff ofc [13:55] <ubitux> like, i got an avx2 laptop for < 1k¬ [14:01] <michaelni> AFAIK we have 4kusd in the SPI account and theres a 4k donation that should reach the account so we can afford more than 2-3 decent notebooks if they are <1k [14:03] <ubitux> i really wouldn't mind using spi money to offer a multicore notebook to durandal_1707 :)) [14:03] <ubitux> not sure if other developers are in such needs [14:03] Action: michaelni agrees [14:04] <ubitux> btw, jamrial and plepere are sharing an shh access on my avx2 notebook; i can probably grant another access if someone needs [14:06] <michaelni> someone, preferrably durandal_1707 needs to write a request for a SPI payed notebook as mail to ffmpeg-devel (this is required for all use of FFmpeg-SPI money) [14:09] <BBB> why dont you buy one for durandal_1707 then [14:09] <BBB> Ill buy my own [14:09] <BBB> I dont like basic ones, they have these complicated oses that eat half of your time [14:10] <BBB> and you have to get into distro trollwars (like on the ml just now :) ) [14:10] <kierank> wow a patch from realnetworks [14:11] <wm5> against an ancient ffmpeg? [14:12] <cone-847> ffmpeg.git 03Ronald S. Bultje 07master:45bed0ab3006: vp9/x86: fix bug in intra_pred_hd_32x32. [14:14] <relaxed> it was probably a recent version of ffmpeg at the time, but due to buffering... [14:14] <kierank> gcc 4.1 [14:14] <kierank> lol [14:19] <nevcairiel> it just takes a while for mail to get here [14:19] <ubitux> haha [14:20] <BtbN> ubitux, yep [14:21] <ubitux> From 4cf471903c9285f3e7dd035f0ef9cf153ad28535 Mon Sep 17 00:00:00 2001 [14:21] <ubitux> From: user <user(a)localhost.localdomain> [14:21] <ubitux> blast from the past [14:21] <BtbN> it's basicaly finished and works, on windows [14:21] <BtbN> on linux the driver does weird things [14:21] <BtbN> and nvidia was not cooperative in resolving them [14:21] <ubitux> BtbN: ah, cool; what's your opinion vs x264? any benefit with simultaneous encodes? [14:22] <BtbN> The quality is bad, but it's fast [14:22] <BtbN> https://github.com/BtbN/FFmpeg/tree/nvenc [14:23] <BtbN> nvidias own tools manage to get a better quality at the same bitrate, but i have no idea how to improve on that [14:23] <BtbN> the documentation isn't the best [14:23] <nevcairiel> sounds like the configuration of the encoder just isnt good [14:24] <ubitux> BtbN: you can't submit it because of the license thing? [14:25] <BtbN> It would be useless if i submit it, nobody would use it, because no distributor would compile it in [14:26] <ubitux> distributors are not the only users of ffmpeg :p [14:26] <BtbN> The nvenc sdk has no standard way to install it [14:26] <BtbN> And it's only about the header [14:28] <BtbN> nevcairiel, yes, of course, but i have no idea how to improve it. I played around with all options, there is some CBR 2 Pass mode that looks promising, but it's not documented how to use it [14:29] <nevcairiel> i have been meaning to figure out the configuration a bit more so that I can improve the quality of OBS, but i'm just way lazy [14:29] <BtbN> I'm the one who wrote the nvenc encoder in OBS. [14:30] <wm5> what's OBS? [14:31] <BtbN> I spent quite some time trying to figure out how to improve the quality [14:31] <BtbN> I even reverse engineered parts of nvidias shadowplay software, but it's using an entirely diffrent api internaly [14:31] <nevcairiel> wm5: an app to grab and encode your screen content and directly stream to youtube/twitch/whatever [15:18] Action: Daemon404 wishes he never read this debain thread [15:18] <Compn> depressing ? [15:18] <Compn> for the projects or for debian ? [15:19] <Daemon404> i forgot how... special... some people on the debian ML are [15:19] <Compn> kik [15:19] <wm5> come on it's fun [15:19] <Compn> er lol [15:19] <wm5> contribute some trolling [15:20] <Daemon404> the thread isnt even about its topic anymore [15:20] <Daemon404> just API/ABI complaining [15:42] <kierank> Daemon404: yeah i did warn you [15:52] <cone-847> ffmpeg.git 03Carl Eugen Hoyos 07master:b70332205966: Remove useless configure option "--enable-sram". [16:05] <cone-847> ffmpeg.git 03Christophe Gisquet 07master:d1b1c3bb5ef1: hevc: reorder loops [16:05] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:2278146a0ca0: Merge commit 'd1b1c3bb5ef1bee409c06c3bcacfb8674e1cb574' [16:17] <cone-847> ffmpeg.git 03Christophe Gisquet 07master:09182b322437: hevc: simplify rounding [16:17] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:e6ffe7bcfc4a: Merge commit '09182b3224370e1d636631e77f201a8388766bd6' [16:36] <cone-847> ffmpeg.git 03Christophe Gisquet 07master:3505b19652ea: hevc: derive partially amvp list [16:36] <cone-847> ffmpeg.git 03Christophe Gisquet 07master:ed53cc217f6a: hevc: derive partial merge list [16:36] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:cde2de4a05dc: Merge commit '3505b19652ea34089a4cd08d4d3358fcc0a3db8b' [16:36] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:5195512addc1: Merge commit 'ed53cc217f6a6e4ddb35e3f01a79496091dc82dc' [16:47] <cone-847> ffmpeg.git 03Christophe Gisquet 07master:cf6090dc6252: hevc: use intreadwrite [16:47] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:17cbc6457b7a: Merge commit 'cf6090dc6252f2b276aa4133e3d73a89f4c6046c' [16:59] <cone-847> ffmpeg.git 03Christophe Gisquet 07master:14ca0aa69b5d: hevc: wait proper position for tmvp [16:59] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:980aa2ae355d: Merge commit '14ca0aa69b5d954ce56d0826ab9f1f70e535c92a' [17:06] <cone-847> ffmpeg.git 03Christophe Gisquet 07master:3187fa14a326: hevc_mvs: avoid deriving tmvp in amvp [17:06] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:b097d1765b38: Merge commit '3187fa14a326908f9471a038e3b5b24c0eaf655e' [17:27] <ubitux> "Kick the non-cooperating people off both projects." great suggestion [17:28] <ubitux> that's not at all what lead to the fork [17:29] <wm4> :) [17:30] <Daemon404> debian people bro. [17:38] <ubitux> so basically the only interesting feature of yasm 1.3.0 is "Use a larger hash table size in NASM macro handling."? :) [17:41] <nevcairiel> a bunch of fixed instructions, and a few new ones, although the new ones arent all that interesting for us [17:41] <nevcairiel> maybe aes [17:41] <ubitux> > Fast and efficient virtual multi-pass optimizer that automatically generates much smaller code for jumps and immediates. [17:41] <ubitux> what about this? sounds weird [17:42] <nevcairiel> thats not new [17:42] <ubitux> ah yeah, i was too greedy [17:42] <nevcairiel> they kinda put all features there, its a bit misleading [18:23] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:da7d839a0d3e: ffv1dec: check that global parameters do not change in version 0/1 [18:23] <cone-847> ffmpeg.git 03Michael Niedermayer 07master:113229922ae2: Merge commit 'da7d839a0d3ec40423a665dc85e0cfaed3f92eb8' [18:56] <cone-847> ffmpeg.git 03Anton Khirnov 07release/0.10:3aebdffb010d: cdgraphics: switch to bytestream2 [18:56] <cone-847> ffmpeg.git 03Anton Khirnov 07release/0.10:cf6b2a0ad2b0: cdgraphics: do not return 0 from the decode function [18:56] <cone-847> ffmpeg.git 03Anton Khirnov 07release/0.10:9d5f4f025304: svq1: do not modify the input packet [18:56] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:a457115f0f09: Merge commit '3aebdffb010df025728d6c2af89642f9634aa806' into release/0.10 [18:56] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:6596f5c3aa9a: Merge commit 'cf6b2a0ad2b06aabf04bec4c7b19e78a560cd904' into release/0.10 [18:56] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:aed8f48545a4: Merge commit '9d5f4f025304ac7c69775179044e6f69f370441a' into release/0.10 [18:56] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:a2c6cb260fda: avcodec/svq1dec: Fix multiple bugs from "svq1: do not modify the input packet" [19:07] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:c79cf0129eda: error_concealment: avoid using the picture if not fully setup [19:08] <cone-847> ffmpeg.git 03Reinhard Tartler 07release/0.10:4709baecc9cc: Prepare for 0.8.14 Release [19:08] <cone-847> ffmpeg.git 03Reinhard Tartler 07release/0.10:a79e58cdc6e4: Update Changelog for v0.8.14 [19:08] <cone-847> ffmpeg.git 03Luca Barbato 07release/0.10:d86df7dd497e: mpegts: Define the section length with a constant [19:08] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:688ab2660474: Merge commit 'c79cf0129edafc388ba1c47cd7b6a620557e48de' into release/0.10 [19:08] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:a94fe0e15148: Merge commit 'a79e58cdc6e46c62346d13e7ddd4da2008714200' into release/0.10 [19:08] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:da19bfb2876d: Merge commit 'd86df7dd497ead1132bd95df7c4c18d91fbe3def' into release/0.10 [19:22] <cone-847> ffmpeg.git 03Luca Barbato 07release/0.10:ebe2292eafa7: mpegts: Do not try to write a PMT larger than SECTION_SIZE [19:22] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:dcc68de942f0: vp3: Copy all 3 frames for thread updates [19:22] <cone-847> ffmpeg.git 03Reinhard Tartler 07release/0.10:042c25f54bd2: Update Changelog for v0.8.14 [19:22] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:437a0ae1a105: Merge commit 'ebe2292eafa7cac71dcdddb865e18c05635fe117' into release/0.10 [19:22] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:a5a2f8ac5ab4: Merge commit 'dcc68de942f06d358192a57057adb133f7c40dd7' into release/0.10 [19:22] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:acf7c5b8ee63: Merge commit '042c25f54bd25b52d2936b822be026450971a82d' into release/0.10 [19:23] <j-b> 'morning [20:00] <cone-847> ffmpeg.git 03Luca Barbato 07release/0.10:0ab76ddf313e: avcodec: Introduce ff_get_buffer [20:00] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:bb015b519e29: Merge commit '0ab76ddf313eeab70d06619ae0376fd7dd40761b' into release/0.10 [20:09] <michaelni> evening [20:20] <beastd> OMG you can terribly fat finger "input" to "inout"-- Damn those keyboard layouts ;) [20:24] <cone-847> ffmpeg.git 03Martin Storsjö 07release/0.10:c98d164a6a2c: configure: Check for -Werror parameters on clang [20:25] <cone-847> ffmpeg.git 03Luca Barbato 07release/0.10:e24d1cbc4e13: lavf: Fix leftovers from the ff_get_buffer patch [20:25] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:5d790cdb9ee2: Merge commit 'c98d164a6a2c3d93bfb10d44c946bc3ed56f14e7' into release/0.10 [20:25] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:854fa0d1ab35: Merge commit 'e24d1cbc4e134e0b4448b3bc3a80f3197271b347' into release/0.10 [20:25] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:80a77c0c5bef: avcodec/libspeexdec: fix missing header includes [20:35] <cone-847> ffmpeg.git 03Reinhard Tartler 07release/0.10:5a2d1913a96b: Prepare for 0.8.15 Release [20:35] <cone-847> ffmpeg.git 03Reinhard Tartler 07release/0.10:452e343295e2: avcodec: Add more missing #includes for ff_get_buffer() [20:35] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:b5d7b80a7e43: ffv1dec: check that global parameters do not change in version 0/1 [20:35] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:257188acea8d: Merge commit '5a2d1913a96bb0e029b424a4dd5b414cfb91c708' into release/0.10 [20:35] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:8044ab94e485: Merge commit '452e343295e25f7b2cbc8529c66e9386e2ea6f55' into release/0.10 [20:35] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:497d66126242: Merge commit 'b5d7b80a7e43779ca2962ba56442579c2a7e927d' into release/0.10 [20:42] <cone-847> ffmpeg.git 03Mans Rullgard 07release/0.10:394f4ee61e3b: pcmenc: set correct bitrate value [20:52] <cone-847> ffmpeg.git 03Michael Niedermayer 07release/0.10:c13b19a5bcc2: Update for 0.10.15 [20:59] <cone-847> ffmpeg.git 03James Almer 07master:9ffac3d00d2f: lsws: duplicate ff_log2_tab [21:43] <michaelni> ubitux, is the xvid idct patch ok without changing formating ? [21:44] <ubitux> michaelni: sure yes, i doubt there will be much changes to merge but yeah [21:44] <ubitux> go ahead, thx [21:44] <michaelni> ok thx [21:59] <ubitux> wm4: it's to allow future realloc of that space (re: av_strndup) [22:00] <wm4> you know you can post on libav-devel [22:00] <ubitux> no, they will misinterpret it [22:00] <ubitux> every time i'm talking there they believe i'm trying to show everyone ffmpeg is better or whatever [22:01] <ubitux> so i'm just an observer [22:01] <ubitux> wm4: anyway, that's the same with av_strdup(); this typically happens in drawtext where the avoption text string is expended [22:02] <cone-847> ffmpeg.git 03Pascal Massimino 07master:298b3b6c1f8f: avcodec: add C xvid IDCT support [22:02] <wm4> makes sense [23:37] <jamrial> ffplay and libavfilter are also using ff_log2_tab from lavu [23:38] <jamrial> just sent a patch to deal with libavfilter the same as every other library, but not sure about ffplay [23:39] <Daemon404> ... there is no reason ffplay needs that table [23:39] <jamrial> av_log2() [23:41] <Daemon404> oh right because the stupid function is in a header [23:41] <Daemon404> lovely terrible design. [23:44] <wm4> trollol [23:47] <J_Darnley> Oh fuck. Intel's docs used to be accessible by number. Now they've got titles. [00:00] --- Wed Aug 13 2014

1 0

ffmpeg.log.20140812
by burek 13 Aug '14

13 Aug '14

[06:41] <hotwings> any idea why i would get "libswresample.so.1: cannot open shared object file: No such file or directory" when /usr/local/lib/libswresample.so.1 exists? [06:42] <klaxa> try running ldconfig as root [06:43] <klaxa> basically, it's not indexed by the linker [06:43] <klaxa> as far as i understand [06:43] <klaxa> might be wrong on the terminology [06:43] <hotwings> that fixed it, thanks for the help klaxa. [06:44] <klaxa> glad i could help :) [06:44] <hotwings> i didnt have that problem before but i did update my gcc packages earlier, and then grabbed a new git [06:44] <hotwings> ffmpeg git that is [07:00] <surge> anyone get ffmpeg working on CentOS 7 yet? [08:40] <hunterp> I would like to delay only ONE of the two input files. But, even with a value of 5, itsoffset is not working....why?? ffmpeg -itsoffset 5 -i one.m4a -i five.m4a -strict -2 -filter_complex "[0:a][1:a]amerge,pan=stereo:c0<c0+c2:c1<c1+c3[aout]" -map "[aout]" out.m4a [08:50] <relaxed> hunterp: place ii in front of the input you want delayed, not both inputs. [08:50] <relaxed> it* [08:51] <hunterp> i only show one usage, im not sure what you mean [08:52] <relaxed> ffmpeg -i first_input -itsoffset 5 -i delayed_input ... [08:52] <hunterp> thats what I showed you,except reversed. i dont see how that makes any meaningful difference [08:53] <relaxed> you have it before both inputs. [08:53] <hunterp> no i dont. not in the buffer [08:59] <termos> I keep getting Operation not permitted on av_interleaved_write_frame but I have write access to the harddisk and the RTMP stream. What could be the issue here? [08:59] <relaxed> hunterp: you're wrong, try it. [09:00] <Mavrik> termos, where are you writing your output to? and in which format? [09:00] <relaxed> That's like saying in the string "1 2 3", "1" doesn't come before "3". [09:02] <termos> I'm writing it to a cdn and and a an s3fuse mount, I keep getting the message for both of them though. I run multiple threads doing the exact same thing and some of them keep getting Operation not permitted [09:39] <termos> Okey after more testing it seems to be only related to my s3fuse mount, I need to check if the problem is with the mount or something. [11:08] <valder> wondering if anybody could help me with this code. I'm trying to extract audio via ffmpeg from a video. I couldn't find a Java solution otherwise I would have done that. but any help would be appreciated. The output file can't be played by any player I can find. I think it may have to be the encoding part of the process but I'm not sure. This isn't my original code but something I found on SO, I've tweaked it to fit my situati [11:32] <gabriel_> Hi all! Anybody familiar with building ffmpeg for Nemo (Mer project)? Is it actually possible? [12:44] <termos> I keep getting "Current profile doesn't provide more RBSP data in PPS, skipping", should I be worried? [13:04] <phil4000n> http://ffmpeg.gusari.org/static Why the static builds for linux do not contain as for Windows all codecs such as libOpus, nor ffplay and ffprobe? Licence issue??? [13:12] <zloy> hello guys, I want to capture my screen with audio using the following command: ffmpeg -f alsa -ac 2 -i pulse -f x11grab -framerate 10 -s 1920x1200 -i :0.0 -acodec libmp3lame -vcodec libx264 -b:v 500k -bufsize 1024k -crf 10 -b:a 92k -preset ultrafast -threads 0 -y Test.mkv [13:12] <zloy> everything works find, but I have the audio sync issue [13:12] <zloy> it's not sycnronized with video [13:14] <phil4000n> try to add -async 1 [13:16] <zloy> phil4000n: while capturing the screen? [13:16] <phil4000n> try [13:18] <zloy> hm, ok [13:24] <zloy> phil4000n: unfortunately, it didn't help :-( [13:26] <phil4000n> maybe it's too heavy x264 @1920x1200 for your CPU, try to capture as mpeg4 to see if it helps [13:49] <relaxed> zloy: try changing -framerate to 60 and put "-r 25" after "-i :0.0" [14:14] <zloy> relaxed: thank you [14:40] <vladtv> hello whether there is a network buffering in ffmpeg for incoming stream? [14:43] <vladtv> in vlc - network-caching [16:31] <Katharsis> i'm using `for name in *.ogg; do ffmpeg -i "$name" -ab 128k -map_metadata 0:0,s0 "${name/.ogg/.mp3}"; done;` for convering all mp3 to ogg file [16:32] <Katharsis> the problem is that the ogg lost some information like timestamp [16:32] <Katharsis> any ideas how to keep the timestamp? [16:38] <Katharsis> ok, i used `ffmpeg -i in.ogg -acodec libvorbis -aq 6 out.ogg` and it works [16:42] <nicktrandafl> Hi. I followed this howto https://trac.ffmpeg.org/wiki/CompilationGuide/MSVC and everything went ok. I used this configuration "./configure --toolchain=msvc --arch=x86 --enable-hwaccels --enable-dxva2 --enable-shared --disable-static --prefix=/c/ffmpeg_build". I got ffmpeg 2.3 binaries. I searched information how to use dxva2 api, but i didn't find any up to date information. I tried use this function avcodec_find_decoder_by_na [16:42] <nicktrandafl> Quastions: [16:42] <nicktrandafl> 1) how to check if I have builded ffmpeg correct(if my ffmpeg binaries support dxva2)? [16:42] <nicktrandafl> 2) please, give me an example of usage of dxva2 api? [16:43] <nicktrandafl> sorry for my english [16:51] <brontosaurusrex> where are nightly gusari builds? [16:53] <nicktrandafl1> Is this quastion for my? [16:54] <brontosaurusrex> nah, i see this are linux only [16:54] <brontosaurusrex> I need a fresh mac ffmpeg build [17:17] <relaxed> brontosaurusrex: http://www.evermeet.cx/ffmpeg/ [17:18] <relaxed> ^^ listed on the download page [17:45] <brontosaurusrex> well, just build my own [17:45] <brontosaurusrex> but thanks relaxed [17:45] <brontosaurusrex> i have a problem thought with adobe premiere, 1st frame of my x264 encode is missing, any cures? [17:45] <brontosaurusrex> x264 encoding settings doesn't seem to matter either [17:49] <relaxed> pastebin.com the command and all output [17:50] <brontosaurusrex> any command, also tested ffmpeg from 2013 and current build from git [17:51] <brontosaurusrex> relaxed, so your using adobe and have not seen anything like that? [17:52] <relaxed> I do not use premiere. [17:54] <JohnBergqvist> Hi, i'm trying to create a screenshot from a video, at the resolution the video is encoded at, however when taking the screenshot, it appears to ignore the Aspect ratios, so that the screenshot appears stretched slightly. is there any way to compensate for this (ideally in a manner that's not specific to each individual video)? [17:54] <JohnBergqvist> The command i'm using is: "fffmpeg -ss 00:10:00 -t 1 -i Video.ts -f mjpeg 101.jpg" and the video's properties are: "Stream #0:0[0x1011]: Video: h264 (High) ([27][0][0][0] / 0x001B), yuv420p, 720x480 [SAR 8:9 DAR 4:3], 23.98 fps, 23.98 tbr, 90k tbn, 47.95 tbc" [17:54] <JohnBergqvist> the aspect ratio of the screenshot I want is 4:3 [17:55] <JohnBergqvist> when taking a screenshot with VLC (which gives me the correct aspect ratio) the resolution is: 720x540, yet the screenshot i'm getting from ffmpeg is 720x480 [17:59] <relaxed> so use, ffmpeg -ss 00:10:00 -i Video.ts -vf scale=w=720:h=540 -f mjpeg 101.jpg [17:59] <JohnBergqvist> yes, but "in a manner that's not specific to each individual video". [18:00] <JohnBergqvist> I might not always know the ideal resolution it should be. [18:01] <JohnBergqvist> Also, trying that still gives me the same stretched screenshot :/ [18:03] <relaxed> -vf scale=w=720:h=540,setsar=1:1 [18:04] <JohnBergqvist> no, I can't specify the fixed resolution by hand [18:05] <JohnBergqvist> is there no way to calculate it from the resolution that ffmpeg's picked up (720x480)? [18:06] <relaxed> the math is --> bc -l <<<"480*(9/8)" [18:07] <relaxed> anamorphic video is stored as 720x480 but played back differently on a computer because the pixels aren't square. [18:07] <JohnBergqvist> yeah. [18:08] <JohnBergqvist> is there no way for ffmpeg to compensate for that difference though? [18:08] <JohnBergqvist> without hard-coding any final resolution? [18:08] <relaxed> try -vf scale=w=720:h=-1 [18:09] <JohnBergqvist> again, i can't use actual numbers. [18:09] <JohnBergqvist> is there a way to specify the width that ffmpeg has reported instead of manually using 720? [18:09] <relaxed> iw [18:10] <JohnBergqvist> ok [18:10] <relaxed> read about the scale filter in the manual [18:10] <JohnBergqvist> using -vf scale=iw:-1 does't have any effect [18:11] <relaxed> you can probably do it using the varible names [18:11] <relaxed> with math including the sar [18:12] <JohnBergqvist> im crap at maths :( [18:15] <JohnBergqvist> if I use "-vf scale=iw*sar:ih" I get a resolution of 640x480 which corrects the aspect ratio, but that doesn't give me the max 720x540 though. [18:16] <relaxed> try -filter:v "scale=iw:ih*ow/sar/iw" [18:17] <JohnBergqvist> Oooh, that's done it! lovely, thanks ^_^ [18:17] <JohnBergqvist> lemme just test it with other videos [18:17] <relaxed> \o/ [18:17] <JohnBergqvist> would it mess up a non-anamorphic video? [18:17] <JohnBergqvist> *screenshot [18:19] <JohnBergqvist> Ahh, it appears it is :( [18:20] <JohnBergqvist> have a 16:9 file this time which is 720x480 according to ffmpeg, yet 853x480 according to vlc [18:21] <JohnBergqvist> ffmpeg's screenshot gives me 720x405 [18:21] <kaotiko> hi [18:22] <relaxed> write a script that checks the sar [18:23] <JohnBergqvist> ffmpeg gives me the sar: [SAR 32:27 DAR 16:9] can I not use that directly in a scale command? [18:24] <JEEB> yes [18:24] <JEEB> they are available as variables [18:24] <JohnBergqvist> ok [18:24] <JEEB> it's all in the documentation :) [18:24] <JohnBergqvist> what were you going to say after that relaxed? [18:24] <JEEB> you can get width/height/sar/dar etc [18:26] <JohnBergqvist> im confused, whats the difference between SAR and DAR? and how do they relate to the actual video size reported by FFMPEG (720x480) vs the size that's being played for the file (16:9 - 853x480)? [18:26] <JEEB> SAR is the aspect ratio of a single sample [18:26] <JohnBergqvist> I don't really understand it from what i've read [18:26] <JEEB> DAR is the aspect ratio of the whole final picture [18:26] <JohnBergqvist> ok [18:26] <JohnBergqvist> so how do they relate to those 2 resolutions? [18:27] <JEEB> 720x480 is the actual video frame coded [18:27] <JEEB> then it has information "this picture has a SAR like this, show it like that2 [18:27] <JEEB> and the player then scales when playing [18:27] <JohnBergqvist> so what about this 853x480 resolution vlc gives me? how does it get that? [18:27] <JEEB> it applies SAR to the actual video size [18:28] <JohnBergqvist> OK. Can I do this in ffmpeg then? and therefore get that 853x480 pic myself? [18:28] <JEEB> you scale it so that you get 1:1 SAR [18:29] <JohnBergqvist> wait [18:29] <JohnBergqvist> I think i've found a solution [18:30] <JohnBergqvist> oh no. [18:30] <JohnBergqvist> someone suggested I try: -vf scale=w=(sar*iw):h=ih but that gets me the correct resolution for the 16:9 vid, but for the 4:3 vid earlier, it gives me 640x480, not 720x540 [18:30] <JohnBergqvist> OK, so how do I scale it to get 1:1 SAR then? [18:30] <JEEB> uhh [18:31] <JEEB> that thing you have there only modifies width [18:31] <JohnBergqvist> yeah [18:31] <JohnBergqvist> ignore what i typed then [18:31] <JEEB> so no wonder it scales it down :P [18:31] <JEEB> you basically need to make it a bit less simple, and actually check which side has to be scaled in order to not downscale anything [18:31] <relaxed> JohnBergqvist: the math is $(bc -l <<<"(16/9)*480") [18:31] <JEEB> I think I had something similar scripted once [18:32] <JEEB> but I'm not going to go through my archives of lulzy crap for it :P [18:32] <JohnBergqvist> ok [18:32] <JohnBergqvist> so in a non-specific form, that would be (bc -l <<<"(DAR)*iw") then? [18:33] <relaxed> No. Because the answer is 853.333 [18:33] <JohnBergqvist> i've also been told that apparently that 720x540 VLC is giving me for the 4:3 vid is wrong, and that the "correct" size should be 640x480? [18:33] <JEEB> uhh [18:33] <JEEB> what is correct depends on what you want [18:33] <JohnBergqvist> god this is confusing :( [18:34] <JEEB> if you want to only modify width then it's "correct" to do so, if you never want to downscale either width or height, then it's not "correct" [18:34] <JEEB> some players do it the latter way, some the former way [18:34] <JEEB> the main thing is that the aspect ratio is applied [18:34] <JohnBergqvist> What do you think I should go on? bearing in mind this needs to apply to any video of any type, resolution & aspect? [18:35] <JohnBergqvist> should I do it via width or height? [18:35] <JEEB> I generally don't convert anamorphic (SAR is not 1:1) content to non-anamorphic at all [18:35] <JEEB> if I have to (for plastic boxes and such) [18:36] <JohnBergqvist> so what does that mean? :$ [18:36] <JEEB> then I do it so that I check which side I have to scale up to match the aspect ratio [18:36] <JEEB> it means that in most cases I keep anamorphic stuff anamorphic :P [18:36] <JEEB> and just keep the aspect ratio flag [18:36] <JohnBergqvist> so which of those two scalings i've suggested (doing it via width or height) does that apply to? [18:37] <JohnBergqvist> bearing in mind there, these images are meant to be displayed in a web browser, if that helps? [18:38] <JEEB> if whatever you are aiming at is a dumb shit that cannot handle anamorphic, then as I said I usually do it so that I scale either one up depending on the picture ratio and the final aspect ratio [18:38] <JEEB> I remember actually forcing flash to a specific aspect ratio years ago :P although that probably didn't work for full screen [18:39] <JohnBergqvist> argh [18:40] <JEEB> also you cannot encode 4:2:0 YCbCr with non-mod2 resolutions, so you will probably want to account the horizontal/vertical subsampling in your resize :P [18:40] <JohnBergqvist> which method does your one apply to then? [18:40] <JEEB> what? [18:40] <JohnBergqvist> "scale either one up depending on the picture ratio" is that: -vf scale="w=(sar*iw):h=ih"? [18:41] <JEEB> it means that you do one or the other depending on the input aspect ratio and the final aspect ratio [18:41] <JEEB> you can do simple if-else stuff [18:43] <JohnBergqvist> ah ok [18:51] <JohnBergqvist> well I don't mind settling for 640x480 for the 4:3 stuff for now, so i'll keep it to: -vf scale="w=(sar*iw):h=ih" [18:51] <JohnBergqvist> as that gives me the max resolution for 16:9 stuff [18:53] <JEEB> scale=trunc(if(gte(a\,873/480)\,720\,873/a)/hsub)*hsub:trunc(if(gte(a\,873/480)\,480/(a/(873/480))\,480)/vsub)*vsub [18:53] <JEEB> modify this according to your needs :P [18:54] <JEEB> (or actually, just take it as an example of using if structures and other stuff :P) [18:55] <JEEB> also another line I had commented in the same file was scale=trunc(sar*iw/(ih/oh)/hsub)*hsub:trunc(min(ih\,272)/vsub)*vsub [18:55] <JEEB> just read the fine documentation and make your own line ÖP [18:56] <JEEB> :P [18:57] <JohnBergqvist> Thanks very much, but i'll stick to what i've got [18:57] <JEEB> yes [18:57] <JEEB> take what is spoonfed to you and don't even try to learn things [18:57] <JohnBergqvist> No need to act like that [18:57] <JohnBergqvist> im already confused enough as it is. [19:00] <JEEB> just wait until you learn how SD aspect ratios can be "fun" (as in, not what most applications tell you). Compared to trying to understand that mess, this is simple :P [19:01] <JohnBergqvist> yes... [19:01] <JohnBergqvist> well what i'm using this for is a local media manager application [19:02] <JohnBergqvist> so god knows what crazily-resolution/aspected files the user could load in... [19:03] <JEEB> well, for example DVDs [19:03] <JEEB> most applications will give you the general "flag" those MPEG-2 streams have [19:03] <JEEB> which is supposed to mean "16:9", for example [19:03] <JEEB> yet, for NTSC DVDs (as an example), that "16:9" is actually 40:33 SAR [19:04] <JEEB> see if you can calculate what that ends up with when the picture size is 720x480 [19:05] <JohnBergqvist> I can't calculate anything. I'm 22 and I can still barely add up [19:05] <JohnBergqvist> Trust me on this. I have tried many times to improve :P [19:05] <JEEB> 720 * 40 / 33 [19:05] <Mavrik> *grumble* [19:05] <Mavrik> we should just go back in time and shoot the person that made anamorphic video [19:06] <JEEB> anyways, what I've learned with this stuff is that you can't just automate that crap unless it's a container or video stream that contains the exact aspect ratio [19:06] <JohnBergqvist> ok thanks [19:06] <JohnBergqvist> and yes, yes we should. and PAL vs NTSC... [19:06] <Mavrik> JEEB, not caring about exact ratio helps as well [19:07] <JEEB> Mavrik, or you just handle DVDs and SD DTV according to the spec :D [19:07] <JEEB> and don't care if that was actually how it was mastered [19:07] <JEEB> which means applying 40:33 SAR instead of 16:9 DAR [19:07] <Mavrik> I've mostly dealt with random crap being uploaded from cameras, etc. [19:07] <JEEB> (with 16:9, NTSC) [19:08] <Mavrik> also internets, so knowing what came from DVD wasn't as easy [19:08] <JEEB> naturally [19:08] <JEEB> if it's a vob, that's kind of obvious, or MPEG-2 video in TS or PS [19:08] <Mavrik> mhm [19:08] <JEEB> (in theory, as in you should be able to follow the spec) [19:08] <JEEB> but H.264 already has specific aspect ratio capabilities [19:09] <JEEB> so lörs lärä [19:09] <Mavrik> but I just had the honor of parsing AFD flags from SDI which was also very fun [19:09] <Mavrik> especially since local stations don't use it for widescreen signaling properly :P [19:09] <JEEB> yes, broken shit everywhere [19:09] <JohnBergqvist> Also i'm glad that blu-rays are finally killing off PAL speedup [19:10] <JohnBergqvist> although admittedly that's not always the case.... :( [19:10] <Mavrik> PAL speedup? [19:10] <JEEB> yes, since 25i is still available there are some mastering companies still doing it [19:10] <JEEB> Mavrik, simply speeding up 24(/1.001) content to 25 [19:10] <JEEB> most common way of doing PAL releases of NTSC stuff [19:11] <Mavrik> hmmm [19:11] <JEEB> PAL telecine is also done but much less frequent [19:11] <Mavrik> yeah, that sucked ass [19:11] <Mavrik> half of the movies here looked terrible [19:11] <Mavrik> (I'm from PAL area :P ) [19:15] <JohnBergqvist> pretty much any every PAL DVD of an NTSC tv show or film is PAL sped-up [19:15] <JohnBergqvist> so pitch & speed is up half a semi-tone essentially [19:15] <JohnBergqvist> *a semi-tone, pitch wise [19:16] <JohnBergqvist> although you can correct it so that both a PAL & NTSC DVD will match, if you bother, but it's quite time consuming, so most people only do it for music DVDs? [19:17] <JohnBergqvist> it does't affect video quality, just speed. [19:17] <JohnBergqvist> it's only noticable once you watch a lot of it at the original speed [19:17] <JohnBergqvist> for instance, in the UK, sideshow bob's voice never sounds like Kelsey Grammer to me, cos it's sped up a little :/ v. weird [19:18] <vladtv> hi, my command: avconv -threads 8 -i http://127.0.0.1:8888/udp/238.0.0.1:3001 -i ls.png -i /media/F/ad.png -filter_complex 'overlay=1:main_h-overlay_h-1,overlay=1:1,yadif=1:0' -metadata service_name=LIVESPORT.WS -metadata service_provider=vladimir29 -s 1280x720 -vcodec libx264 -b 3500k -maxrate 3700k -b-pyramid 1 -keyint_min 1 -subq 1 -refs 1 -pass 1 -fastfirstpass 1 -g 25 -bufsize 1M -level [19:18] <vladtv> 32 -acodec libmp3lame -ab 96000 -ar 44100 -f mpegts udp://238.0.0.2:3001 but more errors [h264 @ 0x90b6240] mmco: unref short failure [19:18] <vladtv> [mpegts @ 0x84a2d20] Continuity check failed for pid 17 expected 6 got 7 [19:18] <vladtv> Continuity check failed for pid 301 expected 5 got 148 bitrate=3751.1kbits/s [19:18] <vladtv> [mpegts @ 0x84a2d20] Continuity check failed for pid 401 expected 5 got 10 [19:18] <vladtv> [mpegts @ 0x84a2d20] Continuity check failed for pid 0 expected 12 got 15 [19:18] <vladtv> [mpegts @ 0x84a2d20] PES packet size mismatch [19:18] <vladtv> [h264 @ 0x9013ae0] number of reference frames (0+6) exceeds max (4; probably corrupt input), discarding one [19:18] <vladtv> Last message repeated 1 times [19:18] <vladtv> [mp2 @ 0x84ac600] Header missing [19:18] <vladtv> Error while decoding stream #0:1 [19:18] <vladtv> what is problem? [19:20] <vladtv> this stream is online now http://46.105.179.221:8090 [19:22] <JohnBergqvist> Does h264 video not have a bitrate? [19:22] <JohnBergqvist> because ffmpeg doesn't seem to be reporting one fo rit [19:22] <JohnBergqvist> *for it [19:24] <vladtv> help me please... [20:20] <cbsrobot-> vladtv: as you are using avconv, does it work with ffmpeg ? [23:16] <ConnorM> the ffmpeg.org website is impressive :D [23:27] <fajung> i've compile ffmpeg according the ffmpeg ubuntu's guide, but now I'm getting "Internal error, IDCT permutation not set" [23:31] <fajung> this: http://pastie.org/9468199 [23:32] <fajung> do u want to see the log? [23:35] <sacarasc> The complete output, yeah. [23:36] <sacarasc> But you should be using x264, not h264. [23:44] <fajung> in pastebin.com id= tp2x7QC6 [23:49] <fajung> if I use x264, Unknown encoder 'x264', with libx164 the same msg [00:00] --- Wed Aug 13 2014

1 0

ffmpeg-devel.log.20140811
by burek 12 Aug '14

12 Aug '14

[00:16] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:49456a95ebcb: avformat/movenc: Fix printf type [00:46] <Daemon404> interesting blog post... "noodles" [00:49] <J_Darnley> Where did ~100 new fate tests come from 4 days ago? [01:27] <iive> + [01:51] <michaelni> J_Darnley, some fate test spliting [01:59] <cone-492> ffmpeg.git 03Luca Barbato 07master:c4ac48c5a146: swresample: document the need to configure the context using AVOptions [01:59] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:f4e814f787d0: swresample: check av_opt_set for failure in swr_alloc_set_opts() [02:26] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:6801eb0a0981: avcodec/vc1dec: do not crash when flushing without an allocated frame [02:33] <cone-492> ffmpeg.git 03Ronald S. Bultje 07master:bfffce4d081c: hevc: add missing comma in log message. [03:06] <cone-492> ffmpeg.git 03Christophe Gisquet 07master:5ec85c9750f0: hevc: do generic validation of bitstream [03:38] <viperfx_> Should I submit a ticket regarding this error: [tls @ 0x7fb64af13f20] error:00000000:lib(0):func(0):reason(0)0/0? [04:00] <BBB> viperfx_: probably yes [04:17] <viperfx_> BBB: is there a workaround I can try? [04:17] <viperfx_> Maybe another SSL library instead of OpenSSL? [04:17] <Compn> polarssl ? [04:19] <viperfx_> Is there one that is known to work well in armv7 [04:20] <Compn> gnutls ? [04:24] <viperfx_> OK I will try that [04:28] <kurosu> "Liked-by: BBB" <- welcome to FBmpeg [04:29] <kurosu> next step: "Plussed-by" [05:00] <jamrial> something broke fate-aliaspix in a lot of systems [09:29] <ubitux> http://research.microsoft.com/en-us/um/redmond/projects/hyperlapse/ [09:29] <ubitux> nice filter [10:45] <viperfx_> Could someone please test this bug on HEAD? https://trac.ffmpeg.org/ticket/3845 [11:09] <J_Darnley> Can't you? [11:11] <J_Darnley> Weren't you talking about some API functions the other day? [11:11] <J_Darnley> That implies your compiling something. [11:13] <viperfx_> J_Darnley: Yep I can, I can do it on iOS. But he is asking for console output of the flags and version [11:13] <viperfx_> That is why I was asking if someone has the version of HEAD already [11:13] <viperfx_> to check the bug [11:14] <viperfx_> I am compiling for armv7 now [11:15] Action: J_Darnley is greatly ashamed by his use of "your" above. [11:18] <J_Darnley> I don't see why you can't compile the git head and post the output [11:21] <viperfx_> No worries, I am doing it on my Desktop now [12:47] <BBB> kurosu: youre a first! http://git.videolan.org/?p=ffmpeg.git&a=search&h=HEAD&st=commit&s=liked-by [12:48] <BBB> theres 5 mentions of bbb though: http://git.videolan.org/?p=ffmpeg.git&a=search&h=HEAD&st=commit&s=BBB [12:52] <kurosu> BBB: frist like! [12:52] <kurosu> unfortunately, +1 already claimed: http://git.videolan.org/?p=ffmpeg.git&a=search&h=HEAD&st=commit&s=Plus-one-… [12:55] <kurosu> the correlation between Trolled-By and Daemon404 is high [12:55] <viperfx_> Guys I have compiled ffmpeg with HEAD on my desktop and I cannot reproduce the issue while encoding to another format. The error does occur on ffplay though [12:55] <viperfx_> I also compiled the latest snapshot in http://ffmpeg.org/releases/ for iOS and I still get the same issue [12:56] <viperfx_> As the end of playback of a HTTP source in the app, I get the tls error [13:04] <viperfx_> I think I am gonna try compiling with the latest openssl [13:27] <BBB> viperfx_: is sdl linked to an older ssl/tsl? [13:27] <viperfx_> BBB: SDL? [13:28] <BBB> it is my understanding ffplay uses sdl for its UI [13:28] <BBB> sdl is one of those antique libraries that may hardlink against the wrong ssl [13:28] <BBB> anyway if latest ffmpeg doesnt reproduce it then I guess its not really a bug... [13:28] <ubitux> sdl linking to ssl? oO [13:28] <Daemon404> kurosu, damn straight [13:33] <BBB> ubitux: you know know :D [13:34] <ubitux> it's not linking to any ssl here on my distro [13:34] <ubitux> (fortunately) [13:35] <viperfx_> Compiling FFmpeg (HEAD) with the latest openssl (1.0.1i) seems to have fixed it for me on iOS. [13:35] <viperfx_> yay \o/ [13:36] <ubitux> BBB: i don't see anything in the source, fortunately :) [13:45] <viperfx_> Damn it! Its back :( [13:45] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:9eda4e8bd759: avcodec/idctdsp: move add/put_pixels_clamped_c to header [13:45] <cone-460> ffmpeg.git 03James Almer 07master:8f2634f970b2: lavc: stop exporting internal functions and tables [13:45] <viperfx_> For an audio track of >5m it seems that error pops up [14:00] <Youka> Tested building ffmpeg 2.3 with MinGW a minute ago and there's still the bug with wrongly detected HAVE_NANOSLEEP on configure :/ [14:11] <cone-460> ffmpeg.git 03James Almer 07master:59ecd4882df9: lavc/raw: remove obsolete ff_raw_pix_fmt_tags cruft [14:11] <cone-460> ffmpeg.git 03James Almer 07master:31b7ab9f0699: lavf/mpegts: remove obsolete ff_mpegts_parse_* cruft [14:13] <Youka> That bug stops/breaks 'make'/compilation unless people set HAVE_NANOSLEEP to 0 in config.h after 'configure'. I remember the discussion with fault found by detection of nanosleep function with symbol finding in binaries instead of consider declarations in header files. [14:16] <Youka> (In a strange way, MinGW has nanosleep anywhere in binaries, but nowhere in headers. Nanosleep was an experiment and shouldn't be used.) [14:20] <nevcairiel> send a patch if you care, everyone that does care about mingw uses mingw-w64 where this is not a problem [14:22] <Youka> ok, thx for feedback [14:30] <cone-460> ffmpeg.git 03James Almer 07master:7b3de03c8fa7: lavf: stop exporting internal functions [14:47] <ubitux> question about MB export [14:48] <ubitux> should i have a MV specific side data [14:48] <ubitux> or should i try to come up with a MB information side data with all kind of info in it [14:48] <ubitux> like quant, intra, .. [14:48] <ubitux> i was currently changing the "MBInfo" into a MVInfo and just add a u64 flags for later flags [14:50] <ubitux> so basically, source ( ), w/h, absolute src[xy], absolute dst[xy], and 64-bits for flags [14:50] <ubitux> if we need intra information or quantization, we would need another side data type [14:52] <wm4> what is this even for [14:52] <wm4> and why [14:53] <ubitux> it's exported only on demand, for app that would need them [14:53] <ubitux> this is typically useful to get motion information for free [14:54] <ubitux> -flags2 +export_mvs, for codecs that support it [15:32] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:11512d70facf: avcodec/get_bits: add BITS_LEFT() for finding the bits left with an opened reader [15:32] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:ba47d519e537: avcodec/huffyuvdec: fix overread checks [15:45] <cone-460> ffmpeg.git 03Michael Niedermayer 07master:9a162146ca6c: avcodec/snow: fix null pointer dereference in cleanup after allocation failure [16:10] <wm4> ubitux: maybe the sami-broken motion vector drawing code could be made into a filter, then [16:11] <ubitux> yes that's the goal [16:11] <ubitux> but the other information should be exported as well [16:59] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.3:a1fe3b415073: avcodec/vc1dec: do not crash when flushing without an allocated frame [17:00] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.3:6badd558ce19: avcodec/get_bits: add BITS_LEFT() for finding the bits left with an opened reader [17:00] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.3:dbf5d7e5cd16: avcodec/huffyuvdec: fix overread checks [17:00] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.3:11420649d021: avcodec/snow: fix null pointer dereference in cleanup after allocation failure [17:31] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.3:b88de7b31a4a: Changelog: update for 2.3.2 [17:35] <kurosu> michaelni, regarding BITS_LEFT for huffyuvdec, was there an issue reported ? [17:36] <michaelni> yeah to my inbox IIRC [17:44] <michaelni> no, it was to ffmpeg-security@ its "just" a out of array read though [17:56] <cone-460> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n2.3.2': unknown revision or path not in the working tree. [17:56] <cone-460> Use '--' to separate paths from revisions [17:56] <cone-460> refs/tags/n2.3.2:HEAD: avcodec/snow: fix null pointer dereference in cleanup after allocation failure [18:35] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:6be5a3c0451e: wmalosslessdec: fix mclms_coeffs* array size [18:35] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:bea14966e2a3: mmvideo: check horizontal coordinate too [18:35] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:aa943bd31fad: huffyuvdec: check width size for yuv422p [18:35] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:6419569a9d10: Merge commit '6be5a3c0451e8f199ef1da09961aa76c08c87afd' into release/2.2 [18:35] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:97cbad3d2c90: Merge commit 'bea14966e2a37019cb4e38420868c5bb0542d487' into release/2.2 [18:35] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:45a529d805cf: Merge commit 'aa943bd31fada23db5cb9611215656ab9ebe5b94' into release/2.2 [18:40] <viperfx_> Does anyone know how I can use the API to print out what ffmpeg -v does? [18:41] <JEEB> use lavf and lavc and heighten the verbosity level [18:48] <cone-460> ffmpeg.git 03Vittorio Giovara 07release/2.2:a5992a274ff5: stereo3d: add missing include guards [18:48] <cone-460> ffmpeg.git 03Vittorio Giovara 07release/2.2:6598aaea1ad2: jpeg2000: enable 4 component pixel formats [18:48] <cone-460> ffmpeg.git 03Anton Khirnov 07release/2.2:5bf5a35fb5d4: cdgraphics: switch to bytestream2 [18:48] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:90a1c5e95c3f: Merge commit 'a5992a274ff5f6c4bec3445cb410da0adce8ef70' into release/2.2 [18:48] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:7b67ce9ade01: Merge commit '6598aaea1ad2cf82d40abb191ac26a5e4e5147ba' into release/2.2 [18:48] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:3301b248b033: Merge commit '5bf5a35fb5d452ea4b30cd7b853d92df6705d250' into release/2.2 [18:49] <viperfx_> JEEB: using av_log_set_level() ? [18:52] <Daemon404> yes [18:54] <viperfx_> I used AV_LOG_VERBOSE and I still cant see the compile flags that I would normally see in command line [18:55] <Daemon404> thats not part of the log [18:58] <cone-460> ffmpeg.git 03Anton Khirnov 07release/2.2:18f48e05a22a: cdgraphics: do not return 0 from the decode function [18:59] <cone-460> ffmpeg.git 03Anton Khirnov 07release/2.2:d513c6a0ee58: svq1: do not modify the input packet [18:59] <cone-460> ffmpeg.git 03Felix Abecassis 07release/2.2:67134ad31f1f: h264: fix interpretation of interleaved stereo modes [18:59] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:c11b3010c21a: Merge commit '18f48e05a22a73a389fb3ab4b3eaf78903bab5ef' into release/2.2 [18:59] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:963514ea1a6e: Merge commit 'd513c6a0ee582d22b6e793286774abbde01f6680' into release/2.2 [18:59] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:723512ac7171: avcodec/svq1dec: Fix multiple bugs from "svq1: do not modify the input packet" [18:59] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:0dc5868f141c: Merge commit '67134ad31f1f3bc1515eae129e4368401f7c3342' into release/2.2 [18:59] <kurosu> michaelni, ok [19:00] <kurosu> michaelni, I think I fixed #2760 but I plan to discuss with kostya first [19:02] <michaelni> kurosu, of course, its kostyas code, btw kostya is welcome on ffmpeg-devel IRC channel as well as ML [19:03] <kurosu> I don't think he cares nor wants [19:03] <michaelni> yes :( [19:12] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:588e7226edb8: ffmpeg_opt: Use av_guess_codec() instead of AVOutputFormat->*codec [19:12] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:ffc66ac0d641: avutil/cpu: add aarch64 entries to 2nd table [19:12] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:0397d434054a: avcodec/iff: check pixfmt for rgb8 / rgbn [19:12] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:bb7f236c7fef: avcodec/snow: fix null pointer dereference in cleanup after allocation failure [19:55] <Chillance> so, what is going on with https://trac.ffmpeg.org/ticket/3831 ? [19:57] <wm4> I guess nobody cares about the gif encoder [19:58] <Chillance> seems I might have to resort to something else then [19:58] <wm4> there are better tools anyway [19:58] <wm4> since gif probably needs some sort of global palette generation... [20:06] <Chillance> you got suggestions? [20:06] <Chillance> qgifer doesnt work.. I suppose its using ffmpeg actually [20:07] <wm4> not sure... photoshop? [20:08] <wm4> last I tried with imagemagick, the result was ugly [20:09] <Chillance> hmm [20:11] <michaelni> Chillance, i do care about the gif encoder, i care about everything i just dont have enough time to work on everything [20:21] <cone-460> ffmpeg.git 03Michael Niedermayer 07release/2.2:c0ad5f9333f4: Update for 2.2.7 [20:43] <ubitux> BBB ^ [20:43] <ubitux> wm4: it's not gif related [20:43] <ubitux> at least it doesn't look like so, i'd say sws [20:44] <wm4> even worse [21:17] <cone-460> ffmpeg.git 03Reimar Döffinger 07master:2c0454cd208d: Add missing initialization for AVProbeData. [21:20] <jamrial> ubitux: could you try aplying the libavutil.v patch i sent a couple days ago in your "--enable-shared + libs" fate system and see if it complains about some other lavu function missing aside from the opencl one? [21:20] <jamrial> i have less than a fourth of those libraries installed here [21:21] <ubitux> what's the title of the patch? [21:21] <jamrial> "[PATCH 4/4] lavu: stop exporting internal functions" [21:25] <ubitux> so without opencl, right? [21:28] <jamrial> yeah, i already know lavfi with opencl uses a lavu ff_ function, so i only need to know about other libraries [21:29] <ubitux> ok, just a moment [21:36] <ubitux> jamrial: seems to pass here but i haven't opencl to compare the failure [21:37] <jamrial> if some lavu ff_ function was used outside lavu it would fail at linking [21:37] <jamrial> with an error like "ffmpeg/src/libavfilter/unsharp_opencl.c:184: undefined reference to `ff_opencl_set_parameter'" [21:39] <wm4> heh why was such a change even allowed to make it in [21:40] <ubitux> jamrial: http://pastie.org/pastes/9462874/text [21:41] <jamrial> if compilation succeeded then it's good [21:41] <jamrial> thanks [21:43] <jamrial> wm4: probably nobody that reviewed that code noticed it since lavu has been exporting ff_ functions all this time [21:47] <wm4> (I don't really understand why ff_ is evil but avpriv_ isn't, anyway) [21:47] <Chillance> michaelni, any eta on looking at my ticket? [21:48] <michaelni> Chillance, why is it so important to you ? [21:49] <Chillance> Im trying to create some animated gifs for my crowdfounding project [21:49] <Chillance> and online converters creates a larger file [21:52] <michaelni> well, if theres lots of money involved you could offer a bounty, maybe someone would look into it sooner then [21:57] <jamrial> wm4: ABI afaik. you can (in theory) make, modify and undo ff_* functions at will since they are internal to one library. avpriv_* can be introduced but not removed or their prototypes altered (until you bump major) without the risk of trouble [21:57] <jamrial> exporting ff_* functions kill the entire point and benefit of making that distinction [21:58] <wm4> I see [23:39] <hotwings> git clone of ffmpeg is going at a blazing 6k/sec.. maybe it will be finished by xmas [23:40] <hotwings> any idea when this is going to be fixed? [23:46] <wm4> hotwings: what URL are you using? [23:53] <hotwings> git://git.videolan.org/ffmpeg.git [23:54] <wm4> the official git is git://source.ffmpeg.org/ffmpeg.git [23:54] <wm4> I'm not sure what server it maps to, or if it's different from the videolan one [23:54] <wm4> there's also a github mirror [23:55] <J_Darnley> Are you not the first person to report speed issues? [23:56] <hotwings> i used to use git://source.ffmpeg.org/ffmpeg.git but was told it just forwards to videolan and at some point in the future would be dead. thats why i use videolan [23:56] <hotwings> J_Darnley - if youre asking me, i have no clue if others have reported speed issues [00:00] --- Tue Aug 12 2014

1 0

ffmpeg.log.20140811
by burek 12 Aug '14

12 Aug '14

[01:18] <GRMrGecko> I'm trying to copy this encode with ffmpeg, https://p.gecko.im/rj can someone help me? [08:16] <slowguy> i am transcoding a file using ffmpeg but i want the output file to be directly uploaded to the server [08:16] <slowguy> i tried ftp but could not make it work [08:17] <slowguy> any other method which works easily? [08:24] <grosso> hi [08:26] <grosso> is there any way to create a codec context with the settings for fastest video encoding, no matter what codec is used? [08:27] <termos> does there exist a the ffmpeg functions that are thread safe? [08:28] <grosso> I mean: is there any codec_context setting that can be set for enabling fast encoding in any codec? [08:31] <grosso> I know there exist: av_opt_set (ctx->priv_data,"preset","ultrafast",0); for x264... is there somethig similar for mpeg1, mpeg2, etc.? [08:38] <grosso> is there any documentation that I can read on these subjects? [09:13] <DonGnom> slowguy: i don't know if that would work but i think netcat can get input from a pipe and output it on a different system so when ffmpeg is able to output to stdout you can pipe it into netcat and transfer it that way. [10:04] <Mavrik> morning. [12:51] <slowguy> hello [12:52] <slowguy> when i use ftp ouput then ffmpeg gives very low fps [12:53] <slowguy> but my internet upload speed is much faster [12:53] <slowguy> what is the best way to store the ouput of ffmpeg on a rwmote server [12:53] <slowguy> please tell me [12:53] <slowguy> n [12:54] <sacarasc> As has been suggested before, some kind of remote FS. [12:55] <spaam> i want cloud output. [12:55] Action: sacarasc evaporates spaam. [12:56] <slowguy> sacarasc: i tried mountimg s [12:56] <slowguy> samba server too [12:56] <slowguy> bbut that is even more slow [12:57] <slowguy> ffmpeg is doing 2fps :( [12:57] <spaam> and its fast when you save it to your localdrive + [12:58] <slowguy> yes at local disk it is aroung 150 fps [12:59] <slowguy> but my internet connection is 60Mbps uplink easily [12:59] <sacarasc> Why not encode locally, then push it? [13:00] <slowguy> yes that is where i am stuck at [13:01] <slowguy> say ffmpeg is dumping a .ts locally [13:01] <slowguy> then hiw do i push it simultaneously [13:01] <sacarasc> I can't see why you'd not just do it locally so you can make sure it's not broken before pushing. [13:02] <slowguy> oh no my client wants the file on server as soon as serial finishes airing on tv [13:03] <slowguy> if i start uploading after the serial finishes then it will be too late [13:14] <Max__> I'm trying to pull the pcm audio streams out of an avi file. It runs through about 5 minutes of the stream and then stops with an "Unrecognized audio sample rate index (5)" error. [13:14] <Max__> How can I fix this, or work around it. [13:15] <slowguy> sacarasc: can ffserver be used for this purpose [13:16] <sacarasc> I don't know. [13:20] <Max__Elswhere> http://pastebin.com/BLq0R5zA [13:20] <Max__> sacarasc: ^ that's my paste [13:21] <sacarasc> Why not use -c:a copy ? [13:23] <Max__> No reason. But that also fails the same way. [13:29] <Max__> I don't mind losing the first 5 minutes, but when I try I get the same result. [13:29] <Max__Elswhere> http://pastebin.com/YqErbASh [13:43] <Mavrik> g'day [13:51] <heftig-z> can't the composition time offset be negative? [13:51] <heftig-z> i'm getting [flv @ 0x7f5d280008c0] Invalid timestamps stream=0, pts=80, dts=160, size=215 [14:22] <vladtv> hello, Can I add a transparent logo in ffmpeg? [14:25] <vladtv> i use ... -i /media/F/ad720.png -filter_complex 'overlay=1:main_h-overlay_h-1' ...but is not transparent [14:31] <sacarasc> Is it transparent already? [14:32] <vladtv> it is png file [14:32] <sacarasc> PNG can be transparent if you make them, I believe. [14:33] <vladtv> yes but I need to make transparency in ffmpeg [14:34] <vladtv> 50-60% [14:34] <sacarasc> Check the video filters for alpha stuff, I've never done it, not sure if it can. [15:50] <sekon> Hello, [15:51] <sekon> When fmpeg is called from the shell with certain arguments it creates a valid file [15:51] <sekon> when the same command is called from python's os.system it creates invalid files [15:52] <sekon> The command line used is ffmpeg -loglevel debug -y -i ~/companies/rajneesh/services/ocaatv/src/recording_sample/teststream1/sample_gfVz184qNhVlEzSNWujMuCbYF.ts -ss 45 -t 15 -acodec copy -bsf:a aac_adtstoasc -vcodec copy -movflags +faststart /tmp/test.mp4 >& /tmp/ffmpeg.debug.removeme [15:52] <sekon> the files /tmp/ffmpeeg.debug.removeme is not showing anything that indicates something is wrong [15:53] <sekon> The created file cant be played/probed because of the error [15:53] <sekon> moov atom not found [15:53] <sekon> I am converting ts to mp4 [16:07] <c_14> Does the command finish correctly _before_ you try playing/probing the file? [16:23] <blight> hi [16:23] <blight> does anybody know how to disable sdl when compiling on osx? [16:23] <c_14> --disable-sdl ? [16:24] <blight> nope [16:24] <blight> ffmpeg-2.3.1 btw [16:25] <c_14> --disable-outdev=sdl [16:25] <c_14> --disable-ffplay [16:25] <blight> compiling with --disable-outdevs atm to see if that fixes, but configure still says "SDL enabled: yes" or something [16:25] <blight> ahh ffplay, gonna try that too, thanks [16:25] <blight> first i'll wait for the current compile [16:27] <sekon> c_14: will check [16:31] <sekon> c_14: it is returning 0 both on the command line and from python [16:31] <blight> c_14: configure still says "sdl support: yes" :( [16:31] <sekon> but when i do ffprobe for the file created from shell it is valid [16:32] <sekon> but the snippet created from python comes out invalid [16:32] <c_14> sekon: you are executing both ffprobes after the processes finish? [16:32] <c_14> If yes, please pastebin the output from the python encoding. [16:38] <c_14> blight: It appears that that says yes even with --disable-all, it _might_ just be a check to see if SDL is installed on your system and have nothing to do with ffmpeg being built with it... [16:39] <sekon> c_14: yes [16:39] <sekon> c_14: please look at http://pastebin.com/Hg3siS3W [16:40] <c_14> blight: looking at the configure script, it appears sdl is always enabled if present. [16:42] <c_14> sekon: How short is this video? [16:44] <blight> c_14: i tried to link sdl-config to "false" but it still says sdl support enabled [16:45] <blight> c_14: when i look at the libs created i see they link against the sdl lib which i dont want [16:46] <sekon> c_14: c_14 The input file for the command (according to ffprobe) is Duration: 00:06:30.76 [16:47] <blight> ahh t uses pkgconfig anyway :( [16:50] <c_14> sekon: Because the output only has 2 frames according to that command. [16:54] <blight> c_14: overriding sdl-config and pkg-config to fail seems to make configure say "sdl support: no" [16:56] <sekon> c_14: ok .. is there any sane explanation as to why the same command spawned from python is showing 2 frames while spawned from shell is giving 325 frames ? [16:57] <c_14> Try comparing commandline output, but I can't think of anything obvious. [17:36] <sekon> c_14: i did .. i dont realise what is wrong [17:36] <sekon> or why this is happening [17:43] <c_14> try using the -report option in the python script and checking the commandline it reports there to see if it's the same as the one you use in the shell in case python is changing something somewhere, but other than that, I'm not sure what could be going wrong [17:59] <sekon> c_14: googling for python and -report gave me no real links [17:59] <sekon> i am using flask web framework [17:59] <sekon> if that helps [19:29] <blight> hmm [19:30] <blight> how do i compile 64 bit version of ffmpeg on osx? (10.6) [19:38] <blight> oops seems like the mac i use is 32 bit so i would have to cross compile [20:37] <c_14> sekon: -report as an ffmpeg option [22:18] <brontosaurusrex> nice new page. [23:14] <hotwings> anyone know why cloning the ffmpeg git is so massively slow lately? [23:15] <sacarasc> New site is nuking the git machine. [23:20] <hotwings> hopefully they get it sorted out soon cuz this absolutely sucks [00:00] --- Tue Aug 12 2014

1 0

ffmpeg-devel.log.20140810
by burek 11 Aug '14

11 Aug '14

[00:10] <kierank> ah finally [00:10] <kierank> http://pastebin.com/LV7G554T [00:16] <kierank> no change with assert level 2 [00:20] <kierank> how do you use asan with ffmpeg [00:21] <j-b> ubitux: sorry, ridiculous [00:22] <j-b> nevcairiel: ok [00:24] <michaelni> kierank, iam sure theres a easier way but: --extra-cflags='-faddress-sanitizer -O1 -fno-omit-frame-pointer' --extra-ldflags=-faddress-sanitizer [00:25] <J_Darnley> Wasn't a --toolchain=gcc-asan option added? [00:25] <michaelni> also output would need to be send through asan_symbolize.py [00:25] <michaelni> J_Darnley, probably [00:26] <michaelni> i just copy and pasted what i had in a stored configure line for asas testing [00:26] <michaelni> that predates --toochain=... [00:27] <J_Darnley> Yes, pardon the 4-line paste: [00:27] <J_Darnley> gcc-asan) [00:27] <J_Darnley> cc_default="gcc" [00:27] <J_Darnley> add_cflags -fsanitize=address [00:27] <J_Darnley> add_ldflags -fsanitize=address [00:29] <kierank> what is the minimum gcc [00:29] <kierank> oh 4.8 [00:29] <kierank> i'm on 4.6 [00:29] <J_Darnley> Not sure [00:30] <J_Darnley> ... oh [00:39] <kierank> do I still have to do this asan_smybolize thing? [00:41] <viperfx_> Does anyone know how to disable indexing for mov demuxer [00:44] <smarter> kierank: only if you don't have llvm-symbolizer in your PATH [00:46] <viperfx_> Is there any way to improve the speed of the avformat_open_input() call for mp4a DASH file through http? Does FFmpeg need to read all of it before it can proceed? [00:47] <J_Darnley> I don't know what dash is but is the moov atom at the front? [00:48] <kierank> symbol lookup error: /usr/lib/libswresample.so.0: undefined symbol: __asan_init_v1 [00:49] <cone-860> ffmpeg.git 03Anton Khirnov 07master:1985c2e75c60: Bump major versions of all libraries. [00:49] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:889cb3ae2375: avcodec: remove deprecated use of FF_IDCT_XVIDMMX [00:49] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:c89afc9acfc7: delay removial of FF_API_CODEC_ID, FF_API_OLD_AVOPTIONS and FF_API_OLD_ENCODE_VIDEO [00:49] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:5e1832c29c2d: avutil: delay removial of FF_API_GET_CHANNEL_LAYOUT_COMPAT [00:49] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:70b8668fb532: drop LLS1, rename LLS2 to LLS [00:49] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:6fac4826249d: avcodec/flacdec: fix handling of av_samples_fill_arrays() return value [00:49] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:80c1cb522c97: avcodec/version: leave old audio API [00:49] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:a57ece0cb48b: Merge commit '1985c2e75c607ac51bfd8dc87d2957a5edf2b6f8' [00:49] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:65cb02e87a8f: avcodec/version: delay removial of FF_API_AUDIO_CONVERT/FF_API_AVCODEC_RESAMPLE [00:50] <michaelni> kierank, asan symbolize is for clang asan i dont know about gcc asan [00:54] <kierank> michaelni: anything else apart from asan that's useful [00:54] <kierank> it doesn't work at all [00:56] <mark4o> kierank: valgrind? [00:56] <kierank> not on a realtime stream [00:56] <kierank> i'll have to figure out how to disable video [00:57] <michaelni> cant you dont the stream to a file for easier debuging ? [00:57] <michaelni> duMP [00:57] <kierank> not easily [00:57] <mark4o> did you add -fsanitize=address to your link flags? [00:57] <kierank> yes [00:57] <kierank> but it broke everything [01:01] <viperfx_> J_Darnley: is that the header info? How do I detect that? [01:02] <kierank> great and now moving to 4.8 broke the entire build [01:03] <J_Darnley> Yeah. A demuxer has to read that to read everything else. Usually it would just seek. [01:05] <viperfx_> It seems to take 4-6s to complete that call. That time suggests that it is reading the whole file. [01:07] <kierank> put the moov atom at the front [01:08] <kierank> probably it's seeking to the end [01:12] <Paranoialmaniac> dash always starts moov atom immediately if the file is the first segment [01:13] <Paranoialmaniac> and moov atom shall have no sample table for fast start [01:19] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:d9e7dc3e1664: ffmpeg: do not use SWR_CH_MAX [01:19] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:ced765ceb323: avfilter/af_pan: check if the number of channels where sucessfully set [01:19] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:ce1bb4b75d52: avfilter/af_pan: Do not use SWR_CH_MAX [01:19] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:36bce999f5e7: libavfilter/af_amerge: Do not depend on SWR_CH_MAX [01:19] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:74be0f82a775: swresample-test: make it independant of the internal SWR_CH_MAX [01:19] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:97f8c7a03edd: bump libpostproc and libswresample [01:25] <viperfx_> The file is on a server I have no control over [01:28] <kierank> oh bugger it's because of this abi break [01:29] <Paranoialmaniac> viperfx_: you have no control and the file has no moov atom before any mdat and moof atom, then you cannot do anything [01:31] <viperfx_> Hmm isn't DASH meant to be for adapting streaming? How do browsers play the audio without downloading it all first [01:31] <Paranoialmaniac> if the moov atom is not preceeded by any moof and mdat atom, the file is not a segment file for dash [01:32] <Paranoialmaniac> * does not preceed [01:32] <Paranoialmaniac> sorry [01:32] <viperfx_> I don't quite understand the terms you are using. Should I use http://atomicparsley.sourceforge.net/ to check the file [01:33] <Paranoialmaniac> yes. something like this [01:34] <cone-860> ffmpeg.git 03Anton Khirnov 07master:472f9ed312ec: Remove obsolete FF_API_AVFRAME_COLORSPACE cruft. [01:34] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:287602f9d80f: Merge commit '472f9ed312ec784f7c72876b77163f4741880d05' [01:34] <viperfx_> Ah damn, it says :AtomicParsley error: unsupported MPEG-4 file brand found 'dash' [01:35] <Paranoialmaniac> viperfx_: use boxdumper https://onedrive.live.com/?cid=6bdd4375ac8933c6&id=6BDD4375AC8933C6!404 [01:36] <viperfx_> I am afraid I am on OSX [01:36] <jamrial> michaelni: /usr/lib/gcc/x86_64-pc-linux-gnu/4.7.3/../../../../x86_64-pc-linux-gnu/bin/ld: libavcodec/libavcodec.so.56: version node not found for symbol av_fast_malloc@LIBAVCODEC_55 [01:36] <jamrial> ~/ffmpeg $ grep -r 'LIBAVCODEC_55' * [01:36] <viperfx_> I can send you the URL if you like? [01:36] <jamrial> libavcodec/utils.c:# define LIBNAME "LIBAVCODEC_55" [01:37] <Paranoialmaniac> viperfx_: you can easily build l-smash i think [01:37] <mark4o> viperfx_: brew install l-smash (if you have homebrew installed) [01:38] <Paranoialmaniac> the latest l-smash in homebrew is somewhat broken though [01:38] <Paranoialmaniac> enough to dump the structure [01:40] <viperfx_> Yep got it. Should I show you a pastebin? [01:40] <Paranoialmaniac> show me please [01:41] <viperfx_> It is quite large. Is the --box cmd enough? [01:42] <cone-860> ffmpeg.git 03Nicolas George 07master:f87db4468539: lavf/avio: do not include bprint.h. [01:42] <cone-860> ffmpeg.git 03Anton Khirnov 07master:24e87f7f425a: Remove obsolete FF_API_PROBE_MIME cruft. [01:42] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:0f7ca2d77893: Merge remote-tracking branch 'cigaes/master' [01:42] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:c218d8218d95: Merge commit '24e87f7f425a52b1e69661dcb2fbe0555a76f30b' [01:42] <Paranoialmaniac> viperfx_: yes. enough [01:43] <viperfx_> Great. http://tny.cz/571eb4af [01:44] <kierank> doesn't seem to crash in valgrind [01:44] <kierank> but v quickly in gdb [01:44] <Paranoialmaniac> this structure is youtube's one [01:44] <viperfx_> Yep [01:45] <viperfx_> Its youtube's m4a dash audio file [01:46] <Paranoialmaniac> at least moov atom is placed much earlier. dunno why so take much time to start [01:47] <viperfx_> Could this be a bug in the demuxer? [01:48] <Paranoialmaniac> dunno. ask the author of the demuxer [01:48] <viperfx_> the moov atom is the one placed just after ftyp? [01:48] <Paranoialmaniac> yes [01:48] <viperfx_> Urm could I be using the wrong demuxer? [01:48] <Paranoialmaniac> moov atom contains initialization info for codecs [01:49] <viperfx_> Is there a particular one that works well with this? [01:53] <Paranoialmaniac> dunno [01:53] <Paranoialmaniac> though i want to add the feature into l-smash [01:54] <Paranoialmaniac> btw i'm an author of l-smash [01:55] <michaelni> viperfx_, how can the issue be reproduced = [01:55] <michaelni> ? [01:56] <viperfx_> michaelni: Well first you would have to obtain the stream link through youtube-dl [01:56] <viperfx_> youtube-dl -f 140 -g YOUTUBE_URL [01:57] <viperfx_> Then you should be able to see the effect on any ffmpeg programs such as ffplay [01:58] <viperfx_> Or even ffmpeg -i URL [01:58] <viperfx_> michaelni: youtube-dl can be obtained from here : https://github.com/rg3/youtube-dl [01:59] <michaelni> and what is YOUTUBE_URL ? [01:59] <viperfx_> The url of any music video on youtube you like :) [02:01] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:a973e01501bc: avcodec: disable FF_API_FAST_MALLOC [02:07] <cone-860> ffmpeg.git 03Anton Khirnov 07master:24c788f48787: Remove obsolete FF_API_REFERENCE_DTS cruft. [02:07] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:329832d779e2: Merge commit '24c788f4878729a42150581d7ed7e9a11daf874d' [02:14] <cone-860> ffmpeg.git 03Anton Khirnov 07master:9325d88eba80: lavc: remove obsolete and disabled avpriv functions [02:14] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:a71fcfad892c: Merge commit '9325d88eba8038b3e2a4485e473a018410379e2d' [02:15] <llogan> i'll be back in a week. going on a fishing/shrimping/drinking trip. [02:17] <viperfx_> michaelni: have you noticed the delay? [02:18] <michaelni> viperfx_, didnt yet had time to look but i made a note [02:18] <viperfx_> ok cheers [02:25] <kierank> possibly these crashes were a gdb bug [02:25] <kierank> stops crashing with gdb 7.8 [02:25] <kierank> oh wait [02:25] <kierank> crashed [02:26] <kierank> doesn't appear to crash in valgrind [02:39] <cone-860> ffmpeg.git 03Luca Barbato 07master:8c6f43029137: mpeg: Suppress a compiler warning on callback type [02:40] <cone-860> ffmpeg.git 03Michael Niedermayer 07master:e18d9d90f1d3: Merge commit '8c6f430291374cf6d2cfb85cdbb809803b5a7d83' [02:45] <kierank> crash is ff_vector_fmac_scalar_avx [02:45] <kierank> as far as I can tell [02:53] <michaelni> viperfx_, -f 130 results in "ERROR: requested format not available" [02:53] <michaelni> without f 130 i dont see a long delay [02:53] <viperfx_> michaelni: its -f 140 [03:06] <jamrial> kierank: is the data 32-byte aligned? [03:06] <kierank> don't know exactly where the crash is [03:06] <kierank> it's stopped crashing now so i guess it's in a particular opus function [03:06] <kierank> depending on the content [03:13] <michaelni> kierank, also is the len parameter to ff_vector_fmac_scalar_avx a multiple of 16, >= 16 and not too large ? [03:13] <kierank> len parameter is 960 [03:13] <michaelni> always ? [03:13] <michaelni> or thats the last before the crash ? [03:13] <kierank> last before crash [03:13] <michaelni> ok [03:14] <kierank> but my stream is only 20ms frames [03:15] <kierank> crash is here [03:15] <kierank> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/opusdec.c;h=b28ed… [03:28] <kierank> e405f130 e4004180 are the two addresses [03:29] <kierank> jamrial: ^ [03:36] <kierank> sigh ffmpeg decides opus audio is mp3 [03:38] <kierank> good to see it ignores the pmt [03:45] <jamrial> e405f130 doesn't seem 32-byte aligned [03:47] <kierank> the input is very rarely 32 byte aligned [03:47] <kierank> out_tmp[i] [03:47] <kierank> sorry the output [03:48] <jamrial> both need to be 32-byte aligned. it's a constrain mentioned in float_dsp.h [03:49] <kierank> is fast_malloc guaranteed to be 32-byte aligned [03:55] <kierank> both the buffers are totally fucked [03:56] <kierank> michaelni: anyway it's reproducable now [03:57] <kierank> as long as I get rid of the aac probe code [03:57] <kierank> because ffmpeg likes to think opus in ts is aac [03:58] <Compn> you added opus in ts demux without adding an encoder, and then expect us to test it using ... magic ? [03:59] <kierank> no but i expect a demux not to override what i tell it to [03:59] <Compn> change the opus probe score ? [03:59] <kierank> the spec definitively says it's opus but ffmpeg decides to change it somewhere [03:59] <kierank> why probe? [03:59] <kierank> the mapping definitively says it is opus in ts [03:59] <Compn> you say aac probe is winning [03:59] <kierank> it has a string Opus [03:59] <kierank> yes but why is aac probe even being called [03:59] <kierank> that's the ridiculous part [03:59] <Compn> ok, i see the problem [04:00] <kierank> it would have been merged from libav with bugs [04:00] <kierank> which would have been fun [04:00] <Compn> missed a codec mapping then? [04:00] <Compn> its in mpegts.c ... wheres the other one [04:01] <Compn> theres riff and isom but , probably the bug is in mpegts code [04:01] <kierank> it's mapped correctly [04:01] <Compn> are you sure you built against local ffmpeg and are running local ffmpeg not a system lavf... ? [04:01] <kierank> local ffmpeg yes [04:01] <kierank> because as soon as I remove the aac probe code it works [04:01] <Compn> paste full ffmpeg output to pastebin [04:02] <Compn> with the aac code taking over [04:02] <kierank> that's not the bug [04:02] <michaelni> if its mapped correctly why does it not work ? also how can i reproduce that probe issue =? [04:02] <kierank> the bug is opus crashes [04:02] <kierank> michaelni: because there is some magic that decides it's aac [04:02] <kierank> possibly in ffmpeg.c [04:02] <kierank> who knows [04:02] <kierank> michaelni: the bigger issue is the crash [04:02] <Compn> how can opus crash if its aac decoder ? [04:02] <Compn> heh [04:02] <kierank> any why none of the addreses in opusdec.c are 32-bit aligned [04:02] <kierank> Compn: because i use my own demux [04:02] <Compn> ah [04:03] <Compn> the truth! [04:03] <kierank> specifically because of silly hacks like that [04:03] <kierank> but that is orthogonal to the crash [04:03] <jamrial> kierank: av_fast_malloc should be in most cases aligned to 32 bytes as long as you compile ffmpeg with avx support. it nonetheless depends on system and build-time config [04:03] <Compn> problem is your buggy demuxer then :P [04:03] <kierank> Compn: seriously? [04:03] <Compn> its just a guess... [04:03] <kierank> no [04:04] <kierank> the addresses are still broken in opusdec with ffmpeg [04:04] <kierank> the demux works fine [04:04] <Compn> is it the autodetect of LATM that causes aac probe ? [04:04] <Compn> * so auto-detect LOAS/LATM instead of hardcoding it. */ [04:06] <michaelni> kierank, please dont forget to provide some usefull info about the opus misdetection once you are done with the more imporatnt things [04:07] <michaelni> i do want to fix it [04:07] <kierank> michaelni: well in principle if I say something is opus in mpegts nothing should change it [04:08] <michaelni> yes, but you want that [04:08] <michaelni> you only have to tell me how to reproduce it [04:08] <michaelni> and ill fix it [04:09] <michaelni> or rather ill try [04:09] <kierank> i'm not concerned about this specific issue [04:09] <kierank> it's more why ffmpeg decides to override the demux [04:10] <Compn> kierank : you didnt specify opus in mpegts.c [04:10] <kierank> the whole point of going through this process and getting mozilla to pay $2500 registration is so that we could uniqely identify it [04:10] <michaelni> kierank, i really dont know what or why should override the demuxer [04:10] <Compn> i mean, doesnt it have to have a ISO type in this table in mpegts.c ? static const StreamType ISO_types[] = { [04:10] <kierank> yes and there is a patch on the ML that does that [04:10] <Compn> your patch doesnt specify it. [04:10] <kierank> it's not an iso type [04:10] <kierank> it's a registered type [04:11] <kierank> static const StreamType REGD_types[] = { [04:11] <Compn> i guess it doesnt have to be both... [04:11] <kierank> can't be both [04:11] <kierank> by definition [04:12] <kierank> either you register it with the registration authority or it's an ISO type [04:12] <Compn> ah [04:12] <Compn> are you going to ifdef all the opus stuff in mpegts ? [04:13] <Compn> depending if opus decoder is enabled or not... [04:13] <Compn> strange though, i wonder where the aac probe is jumping in on [04:14] <Compn> kierank : in ffmpeg debug log does it say 'probing ' ? [04:14] <Compn> if (pes->st->codec->codec_id == AV_CODEC_ID_NONE && !pes->st->request_probe) { [04:15] <Compn> is it a pes header in your mpegts ? [04:15] <kierank> yes [04:15] <michaelni> kierank, make sure you set st->request_probe = 0 [04:15] <michaelni> and not 1 [04:16] <kierank> i don't do any probing [04:16] <kierank> what i think happens is ffmpeg ignores pmt [04:16] <kierank> and probes first [04:16] <Compn> ok then its activating the pes header detection stuff which detects it as CODEC_ID_NONE, then the aac probe code jumps in [04:16] <Compn> maybe [04:18] <michaelni> kierank, if nothing enables probing then streams shouldnt be probed [04:18] <Compn> around line 990 mpegts.c [04:18] <kierank> michaelni: yes but what probably happens is a pes arrives first [04:18] <kierank> and the demux uses that instead of following the spec and using pmt [04:19] <Compn> so we should change the demuxer to check for pmt before using pes header code ? [04:19] <kierank> yes [04:19] <kierank> i'm sure it did at some point [04:19] <Compn> i remember someone complaining about pes headers in mpegts before... few months ago [04:19] <kierank> but there was some file that required breaking the spec [04:20] <Compn> theres always a file :) [04:20] <kierank> ok so opusdec [04:20] <kierank> libav is also bugged [04:21] <kierank> but they don't seem to have unrolled [04:21] <kierank> so possibly the bug is less likely to occur [04:21] <kierank> less likely to segfault [04:22] <Compn> if you can share sample michael would like to fix it kierank [04:22] <kierank> give me a minute [04:22] <kierank> https://dl.dropboxusercontent.com/u/2701213/alignment.opus [04:23] <kierank> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/opusdec.c;h=b28ed… [04:23] <kierank> if you printf the addresses here they are not 32-byte aligned [04:23] <Compn> what compiler to reproduce ? [04:25] <kierank> 4.8.1 x86-64 [04:28] <michaelni> kierank, they are 32byte aligned here [04:28] <michaelni> av_log(0,0, "XXX %d %d\n", ((int)out_tmp[i]) & 31, ((int)s->celt_output[i]) & 31); [04:31] <michaelni> anyway i need t o go to bed soon, ill look at the log tomorrow [04:44] <kierank> michaelni: [av] XXX 16 0 [04:45] <michaelni> its always 0 0 here [04:45] <kierank> not on the same file [04:45] <kierank> seems that file is ok [04:45] <kierank> but on the live stream it crashes but i'll have to try and catch a bit [04:45] <kierank> that causes the crash [04:46] <kierank> might depend on API vs ffmpeg [04:48] <michaelni> are you getting misaligned pointers from a *malloc function or they get offseted to that later ? [04:49] <kierank> could be the calling application as well [04:49] <kierank> i think [04:52] <kierank> I think the API has to say the output data from get_buffer MUST be 32 byte aligned [05:02] <cone-860> ffmpeg.git 03James Almer 07master:ddce2b265444: tools/fourcc2pixfmt: use avpriv_get_raw_pix_fmt_tags() [05:02] <cone-860> ffmpeg.git 03James Almer 07master:6985ef781397: lavfi: stop exporting internal functions [05:21] <mark4o> kierank: your alignment.opus file is invalid; the reference decoder doesnt accept it at all [05:21] <mark4o> it is because you have a channel mapping for mapping family 0, which is not valid [05:30] <kierank> it was the output when remuxing without any checks for validity [05:31] <kierank> but the crash is unrelated [05:36] <mark4o> remuxing from what source? [05:44] <kierank> from ts [05:44] <kierank> without checking the ts extradata is suitable for going into ogg [05:45] <kierank> (because in the past the only extradata type was ogg) [05:51] <Compn> mark4o : we still dont want decoder to crash, even if invalid file (especially if invalid file) [05:52] <mark4o> kierank: ah ok, so it is the extradata that you are generating in mpegts.c; I think it is correct except for the length, which you are always setting to sizeof(opus_default_extradata) [05:52] <mark4o> ffmpeg ignores extra junk at the end but the reference decoder doesnt [05:53] <mark4o> Compn: yes, just another issue not related to crash [05:53] <kierank> I have sent a patch for the crash [05:54] <mark4o> kierank: the patch that only changes a comment? [05:54] <kierank> yes [05:55] <kierank> if you make your buffers 32-byte aligned you don't have a crash [05:55] <mark4o> ok [11:38] <ubitux> kierank: is it me or there is a weird sound jump around 64 seconds? [11:38] <ubitux> (ffplay -ss 64 alignment.opus) [11:43] <ubitux> mmh yeah it seems there is a large ts jump [11:43] <ubitux> [Parsed_ashowinfo_0 @ 0x7fccd0004f40] n:149 pts:3086400 pts_time:64.3 pos:236672 fmt:fltp channels:2 chlayout:stereo rate:48000 nb_samples:960 checksum:11086BF6 plane_checksums: [ 5F43425D 04692999 ] [11:43] <ubitux> [Parsed_ashowinfo_0 @ 0x7fccd0004f40] n:150 pts:3121920 pts_time:65.04 pos:236672 fmt:fltp channels:2 chlayout:stereo rate:48000 nb_samples:960 checksum:D7F9FF6C plane_checksums: [ 761B7391 7D448BDB ] [12:00] <ubitux> http://fate.ffmpeg.org/report.cgi?time=20140810044909&slot=x86_64-archlinux… [12:06] <viperfx_> michaelni: Should I submit a bug report regarding the youtube mp4a dash format issue? [12:15] <michaelni> viperfx_, yes [12:15] <viperfx_> Did you get a chance to confirm the issue? michaelni [12:15] <viperfx_> it would be good to know its not just me :) [12:16] <michaelni> viperfx_, i saw a ~4sec delay on startup [12:17] <michaelni> would be ideal if we could avoid that [12:17] <viperfx_> Yep okay, and that is abnormal for a file with the moov atom at the start [12:17] <viperfx_> Is there any other details I should include? That could be helpful to the devs [12:19] <viperfx_> michaelni: is this something fairly trivial to fix ? [12:20] <viperfx_> Is there a chance that a newcomer could tackle this? [12:37] <cone-672> ffmpeg.git 03Christophe Gisquet 07master:7117547298b1: hevc: fix incorrect sao buffer size [12:48] <viperfx_> michaelni: Done! https://trac.ffmpeg.org/ticket/3842 [12:54] <wm4> viperfx_: have you found the option that makes it work? [12:54] <viperfx_> wm4: What do you mean? [12:54] <wm4> disabling indexing or so [12:55] <wm4> should make it load instantly [12:55] <wm4> you should also write that it downloads the whole file [12:56] <wm4> ah, found it: fflags=+ignidx [12:57] <wm4> that actually disables loading the index, rather than disabling indexing [12:59] <viperfx_> wm4: on the ticket? [12:59] <viperfx_> If I was using ffmpeg using the iibav libraries [12:59] <wm4> yes [12:59] <viperfx_> where do I set this flag [13:00] <wm4> av_set_opt [13:00] <wm4> err [13:00] <wm4> av_opt_set [13:00] <wm4> use "fflags" and "+ignidx" as params [13:03] <viperfx_> wm4: That func does not seem available to me [13:04] <wm4> pretty sure it's available... [13:04] <viperfx_> Is it global or part of a class [13:05] <viperfx_> AVOptions? [13:05] <wm4> uh, ffmpeg is fully in C [13:05] <wm4> did you include libavutil/opt.h? [13:05] <viperfx_> No, I did not :/ [13:05] <wm4> pass the lavf context as first argument [13:06] <viperfx_> AVFormatContext ? [13:07] <wm4> yes [13:09] <viperfx_> wm4: What do i set as the search flag? [13:10] <wm4> 0 should work in this case [13:13] <viperfx_> wm4: Yep your right, its much faster. However, it does not detect the duration. [13:13] <viperfx_> But its a great improvement! [13:14] <viperfx_> I suppose that is kinda what disabling indexing does? :) [13:14] <wm4> depends entirely on the file format [13:15] <wm4> and in this case depends on details how mov works [13:16] <viperfx_> Ah okay, well it seems to detect the duration of 10s. But then carries on playing til the end :) [13:18] <Paranoialmaniac> you cannot know whole duration unless there is mehd atom or edts atom basically [13:18] <Paranoialmaniac> at startup [13:19] <Paranoialmaniac> youtube sets whole duration of a track to tkhd duration but this is out-of-spec [13:20] <wm4> what makes libavformat read the whole file? [13:25] <Paranoialmaniac> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavformat/mov.c;h=b2ef482c… [13:29] <Paranoialmaniac> hmm? the demuxer doesn't start immediately the first mdat atom is so large? [13:30] <wm4> that's a large condition [13:30] <wm4> so I guess it's searching for the mdat atom? [13:31] <Paranoialmaniac> even if fragmented, the demuxer can't skip the first mdat when minimal indexing [13:31] <ubitux> wm4: note that we can't really modify libavr much or we would have to change its name [13:31] <Paranoialmaniac> this behavior is weird and prevent fast-start [13:32] <wm4> ubitux: how so? you're doing that with the other libs too [13:32] <ubitux> yes but we didn't fork them from libav [13:33] <ubitux> typically if we happened to modify avr, we should either do it in libav, or change its name [13:33] <wm4> you could just enable it by default [13:33] <wm4> but then you'd be afraid that swr falls into disuse [13:33] <Compn> to enable avr by default means we would have to test it [13:33] <ubitux> ^ [13:33] <Compn> do you have a simple way to test it wm4 ? [13:33] <ubitux> and we don't test it [13:34] <wm4> libav has tests [13:34] <ubitux> yes, but for libav [13:34] <Compn> wouldnt we have to modify ffmpeg to make switching of resample possible ? [13:34] <Compn> patch welcome [13:34] <ubitux> it is maintained on libav side, but we don't support a proper integration in ffmpeg [13:34] <ubitux> Compn: i don't think such patch would be welcome actually [13:34] <Compn> its a lib, that we dont have support for in our program, i guess [13:34] <wm4> Compn: no, lavfi has 2 reampler filters [13:35] <wm4> one for swr and one for lavr [13:35] <Compn> ubitux : lol, well 'patch welcome' is usually a troll because no one ever sends patch [13:35] <viperfx_> Guys could someone confirm the issue is reproducable with the current FFmpeg git head? [13:35] <ubitux> but it's dishonest to say it here [13:35] <wm4> but seriously, what the heck are you doing, why are you making people write code for 2 APIs? [13:35] <wm4> viperfx_: just say yes [13:35] <michaelni> wm4, ubitux i cannot maintain avr in ffmpeg, if you guys want that swr fork instead of swr i wont stop you but you must maintain it long term [13:35] <viperfx_> urm alright [13:35] <michaelni> incuding security support [13:36] <wm4> viperfx_: I'm pretty sure nothing about it changed recently [13:36] <Compn> wm4 : write a wrapper lib for either one so they can use the same api ? [13:36] <ubitux> i just think that modifying avr to a certain extent means that we should rename it, and it defeats the purpose wm4 is looking for [13:36] <wm4> well, debian is enabling it by default, so maybe I can drop my swr code [13:36] <ubitux> we can't rely on libav to maintain issue we would have on our side [13:36] <Compn> wm4 : or would that be ok with you ? [13:36] <wm4> (problem solved) [13:37] <ubitux> wm4: following that reasoning, debian might include ffmpeg, so maybe you can drop your libav code [13:37] <wm4> true [13:37] <Compn> somehow i doubt debian will reverse and put mplayer back in :P [13:37] <wm4> that's a much larger step, though, and there are still libav distros [13:38] <ubitux> such as? [13:38] <viperfx_> So in its current state. I can disable indexing and wont be able to seek or get the correct duration. [13:38] <viperfx_> and not get the current duration* [13:38] <wm4> viperfx_: seeking should still work, I think [13:39] <michaelni> ubitux, wm4, also theres another aspect i truely dont get, what is the whole avr/swr fuzz about ? why dont we work together to improve swr and use it [13:40] <ubitux> because it's not available in debian [13:40] <ubitux> that's the only reason [13:40] <wm4> wrong [13:40] <michaelni> ubitux, thats fixable [13:40] <michaelni> ubitux, any reasons against ffmpeg dont apply to swr [13:41] <ubitux> wm4: so what else? [13:41] <Compn> michaelni : can you make the api with avr the same ? i think thats what wm4 is complaining about [13:41] <Compn> so people dont have to write two api... [13:41] <ubitux> Compn: no [13:41] <Compn> or a wrapper api [13:41] <Paranoialmaniac> viperfx_: you can't fast-seek a position until you read that position once since you don't know the position you want to seek at the first time [13:41] <ubitux> Compn: the api is almost the same, but some stuff are simpler in swr [13:42] <wm4> ubitux: I want my project to work both ffmpeg and libav (and other projects also want that), and the swr/avr makes this harder [13:42] <viperfx_> wm4: I cannot seek because I do not have the full duration [13:42] <wm4> viperfx_: you can just seek to a timestamp [13:42] <ubitux> wm4: and you want it to work with both ffmpeg and libav because ffmpeg is not available everywhere, that is, not in ffmpeg [13:42] <ubitux> not in debian* [13:43] <viperfx_> wm4: visually speaking on a UI, how would you limit the range without knowing the full duration? [13:43] <Compn> ubitux : well, because some users choose libav or ffmpeg and they want to support both [13:43] <wm4> viperfx_: well, I guess that's a UI problem [13:43] <Compn> and by choose , i mean debian chose for them :p [13:43] <ubitux> Compn: they don't, distro choose for them [13:44] <ubitux> so anyway, that's always the same issue [13:44] <wm4> viperfx_: but AFAIK dash has metadata to piece mp4 fragments together? [13:44] <ubitux> avr is prefered over swr only because of its distribution, and not its technical merits [13:44] <ubitux> that's not really a logic we should encourage [13:44] <Compn> ubitux : we can either work with downstream or we can ignore them [13:44] <viperfx_> Right, but FFmpeg does not do that yet? [13:44] <Compn> downstream being mpv [13:44] <wm4> viperfx_: no dash streaming support, no [13:44] <Compn> and some other projects that chose to support both [13:45] <ubitux> Compn: let's cut the discussion, there is a technical one currently we are interrupting [13:45] Action: ubitux & [13:45] <wm4> ubitux: I remember when I actually dropped support for an older version of ffmpeg, because swr was crashy, while avr wasn't [13:45] <wm4> but of course that's unfair because that's just a now-fixed bug [13:45] <Paranoialmaniac> wm4: viperfx_'s samle has no metadata to seek random access points [13:45] <ubitux> "one day there was an issue with swr, avr is better!" [13:46] <wm4> Paranoialmaniac: libavformat's generic seeking code will probably take care of it by skipping until the target PTS [13:46] <viperfx_> Paranoialmaniac: Does that void the suggest about seeking with timestamps then? [13:46] <wm4> ubitux: let's just say that my overall experience with swr was more negative [13:47] <Paranoialmaniac> wm4: yes. i know. but you can't fast seek at the first time without manifest file in this case [13:50] <Paranoialmaniac> viperfx_: sure [13:50] <wm4> so I wonder how youtube handles this... [13:50] <BBB> I dont see how changing avr/swr merging helps anyone; lavcodec is shared, and its still a nightmare to work with for downstreams [13:52] <wm4> the swr/avr mess is probably unfixable, so let's just be sad [13:54] <Paranoialmaniac> wm4: sidx atom? the libavformat mov demuxer has no support of sidx box. sorry i forgot the presence of sidx atom :P [13:54] <Compn> wm4 : i think i suggested to j-b once that he should fork ffmpeg/libav and pick features from both projects to make the 'one true ffmpeg' but i dont think he wanted to do that... [13:56] <wm4> Paranoialmaniac: aha, so if that were implemented, it would fix all issues? what exactly does the atom contain? [13:56] <wm4> Compn: that would in fact be the best and only way to solve the fork issue [13:57] <wm4> there'd be some significant danger of it just ending up as 3rd fork, of course [13:57] <Compn> its a possiblity [13:57] <wm4> but if it works, devs would slowly migrate to this project [13:57] <Compn> also maintainer questions would arise [13:57] <wm4> and the original projects would fade into obscurity [13:58] <Paranoialmaniac> wm4: sidx atom has collection of durations, size and random access type of subsegments in a segment file [13:59] <Paranoialmaniac> here. subsegment == moof + mdat [14:46] <BBB> has anyone actually compared avr vs. swr performance? [14:46] <BBB> the code I touched (I compared the avr version) was probably 1.5x as fast in swr [14:47] <BBB> and thats in the main resampling codepath, mind you [14:47] <BBB> not some obscure external outer loop nobody cares about [14:56] <BBB> kierank: Im highly doubtful avr vs swr trollwars in the context of debian are going to be beneficial for anyone, that should really be done outside the debian mailinglist... [14:56] <kierank> probably true [14:56] <kierank> but they are enabling avr [14:57] <BBB> that trollwar will get lost in the other debian trollwars, and that means I wont read half of it, why dont you do that in a new thread so theres a small chance that I actually care and read it [15:00] <kierank> I just continued the thread [16:09] <Daemon404> BBB, when in the history of time have any debian trollwars been beneficial to anyone? :) [16:12] <Daemon404> oh! i guess i can push that mplayer hack reversal [16:12] <Daemon404> i totally forgot [16:15] <wm4> I still don't understand how swr and avr came to be [16:15] <wm4> especially in such high similarity [16:15] <Daemon404> ffmetch fundign + youtueb funding + forkwars [16:15] <Daemon404> and lulz [16:15] <Daemon404> honestly, i really give 0 fucks abotu swr vs avr [16:16] <Daemon404> note how i have purposelt been uninvolved [16:16] <Daemon404> :D [16:16] <kierank> wm4: see ml [16:16] <kierank> wm4: swr came out of nowhere [16:17] <kierank> wm4: avr was because of libav/ffmtech drama [16:17] <Daemon404> kierank, translations: internet is srs bsns [16:46] <kurosu> ff_(vc1|wmv3)_decoder have ".flush = ff_mpeg_flush" [16:46] <kurosu> ff_mpeg_flush does "MpegEncContext *s = avctx->priv_data" [16:47] <kurosu> yet, avctx->priv_data should be afaik a VC1Context* [16:47] <kurosu> I don't get how that works, and I suppose I'm missing something [16:48] <kurosu> does that make more sense to someone else ? [16:50] <kierank> probably the beginning of a vc1context is an mpegenccontext [16:50] <kurosu> yes just confirmed that [16:51] <kurosu> vc1.h:198 [16:51] <wm4> D: [16:51] <kierank> that's pretty normal [16:51] <kierank> the kernel does things like that sometimes [16:51] <kierank> and also lets you go up with CONTAINER_OF [16:51] <kierank> afaik libavcodec only ever goes down [16:52] <ubitux> a lot of projects do that [16:54] <kurosu> yeah, it's just that I had taken MpegEncContext being there as a random event and not something intended [16:54] <kurosu> or even required, actually [16:55] <wm4> <ubitux> a lot of projects do that <- and they're all morally wrong [16:57] <ubitux> we might be wrong in the case of MpegEncContext, but in itself that's not wrong [16:59] <cone-492> ffmpeg.git 03Derek Buitenhuis 07master:43654a22c037: Revert "lavc/utils: Do not require dimensions for PNG." [17:00] <Daemon404> wm4, ^ [17:06] <wm4> sure took a while to undo this [17:30] <cone-492> ffmpeg.git 03Christophe Gisquet 07master:84bc45880ae1: hevc_mvs: set candidate availabilities [18:18] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:07262c719945: avcodec/options_table: leave xvidmmx AVOption in place [18:18] <cone-492> ffmpeg.git 03Christophe Gisquet 07master:0d0d24af0159: hevc_ps: verify P/T/L information [18:21] <BBB> Daemon404: so happy to be using mac; no need for all this distro trollcrap [18:21] <BBB> just some bofh that decides what my computer runs and it works [18:23] <wm4> yeah, and there are only 3 or so projects that try to play linux distro on the mac (like homebrew) [18:38] <cone-492> ffmpeg.git 03Luca Barbato 07master:fb1ddcdc8f51: avresample: Introduce AVFrame-based API [18:38] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:78eaaa857e95: Merge commit 'fb1ddcdc8f51b9d261ae8e9c26b91e81f7b6bf45' [18:47] <cone-492> ffmpeg.git 03Diego Elio Pettenò 07master:2597a842a0a2: tiff: Improve grayscale support [18:48] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:61b9ce00a30c: Merge commit '2597a842a0a2c7e8aa76f32733d27bf64817ae86' [18:54] <cone-492> ffmpeg.git 03Carl Eugen Hoyos 07master:273fc0139e0a: lavf/img2dec: Initialize a stack variable. [18:54] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:d6efed84b1da: Merge remote-tracking branch 'cehoyos/master' [18:55] <BBB> kurosu: and yes youre right end-of-buffer validation is a good idea regardless for sps/vps/pps [18:56] <kierank> the encoders i've seen have totally buggered vps [18:57] <nevcairiel> Luckily the decoder doesn't read vps then [18:58] <mraulet> it does but not too mcuh [18:58] <nevcairiel> (Or well use anything from it) [18:59] <mraulet> vps is mostly needed for hevc extensions [18:59] <kierank> it did try and validate it [18:59] <cone-492> ffmpeg.git 03Martin Storsjö 07master:f4312352fc52: configure: Add probe identification of MS armasm [19:00] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:f51c3e3a84e6: Merge commit 'f4312352fc52cc47c1ba398a33f629d32a737e91' [19:53] <cone-492> ffmpeg.git 03Diego Biurrun 07master:9f8cf50e3be0: configure: Enable gas-preprocessor on all OSes but only if available [19:53] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:7e4e45d227e1: Merge commit '9f8cf50e3be04b8e178086edaa4598826143303d' [20:31] <cone-492> ffmpeg.git 03Martin Storsjö 07master:ab0742d2f34d: configure: Suggest upgrading gas-preprocessor instead of just installing [20:31] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:3787495e1f2c: Merge commit 'ab0742d2f34d70415620441b7ed7da2f03a0c14f' [20:50] <cone-492> ffmpeg.git 03Reinhard Tartler 07master:59cc0812207e: Prepare for 11_alpha2 Release [20:50] <cone-492> ffmpeg.git 03Nidhi Makhijani 07master:8dca0877e3e1: mpegts: Return proper error code on invalid input data [20:50] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:c24be3c4a2e1: Merge commit '59cc0812207e3c2e2f2e559d60d0a8f9296a8847' [20:50] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:d123d9abef52: Merge commit '8dca0877e3e1457e9ec79ffa1ead1135aabb791c' [20:59] <cone-492> ffmpeg.git 03Anton Khirnov 07master:d63443b9684f: lavc: drop the av_fast_{re,m}alloc compatibility wrappers [20:59] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:bd879652476b: Merge commit 'd63443b9684fa7b3e086634f7b44b203b6d9221e' [21:08] <wm4> is trac down? [21:09] <kierank> it's up but slow [21:10] <wm4> hm yeah, now it seems to be loading... extremely slowly [21:23] <kurosu> same here [21:28] <viperfx_> Hi guys, I have come across another bug. Regarding openssl though this time. I can reproduce it in cmd line on desktop and on ios [21:28] <viperfx_> Wanted to run it by you guys before submitting a ticket [21:29] <viperfx_> First you will need to generate a URL with youtube-dl. $ youtube-dl -f 171 -g YOUTUBE_URL [21:29] <viperfx_> That should give a stream url for WebM format [21:30] <viperfx_> If you test that url with ffplay "stream_url" [21:30] <viperfx_> You will notice at the end of the audio. There is this error: [tls @ 0x7fb64af13f20] error:00000000:lib(0):func(0):reason(0)0/0 [21:35] <wm4> ah, lovely incomprehensible openssl errors (or is it gnutls) [21:35] <wm4> ah, openssl [21:36] <wm4> in general this code seems to work (there are people using it) [21:37] <cone-492> ffmpeg.git 03Mark Reid 07master:cab39afb1a9c: avformat/movenc: check that reel_name length doesn't exceed UINT16_MAX [21:43] <viperfx_> Yea, its strange. I did not have this issue til recently. [21:44] <viperfx_> It is not an issue really, but iOS panic's out [21:44] <viperfx_> and I dont know how to avoid it [21:52] <BBB> kurosu: I like that patch :) [21:52] <BBB> (the vps/sps/pps/slicehdr size validation) [21:54] <BBB> kierank: I dont see your new thread? [21:54] <kierank> BBB: i'm not going to start a new thread after all the angry pms and emails I got [21:54] <BBB> just the existing one, which is way too long and mythtv-central so Im not going to make that worse [21:54] <BBB> I wsnt angry [21:54] <kierank> not from you [21:54] <kierank> but from others [21:54] <BBB> oh ok [21:54] <BBB> fine then [21:54] <kierank> In fact I regret posting in that thread [21:55] <BBB> soryr to hear, anyway, I agree with the general comment that the api couldve benefited from a quick review [21:55] <BBB> I do fairly feel that swr is a lot faster than avr but some people dont seem to care about that [21:57] <kierank> BBB: are you going to vdd? [21:58] <BBB> no, my sisters wedding is around the same time [21:58] <BBB> maybe next year, although I have to admit flying hasnt gotten cheaper lately [21:58] <BBB> Im trying to buy a house so money is getitng more tight again [22:03] <kurosu> BBB: Michael already had fortified the decoder though but it was time to reduce the attack window [22:03] <kurosu> BBB: and the bigger city you live in, the more expensive they get [22:04] <BBB> yeah Ive noticed that, prices are like way out there [22:04] <kurosu> I kind of remember an image series showing NY flats and european mansions of equivalent prices [22:04] <BBB> its the problem if everyone has money; you only get stuff if you have MOAR MOAR MOAR [22:04] <kurosu> I don't know how disconnected are rates nowadays [22:05] <kurosu> Many French people have to take a mortgage for 20+ years [22:05] <kurosu> put another way, prices per square meters have doubled with inflation compensated [22:05] <kurosu> in 10 years [22:09] <kierank> many english people will just permanently rent [22:09] <kierank> generation of renters [22:12] <BBB> in manhattan, financial district is considered relatively cheap, and youll get a 2-bedroom appat (1000-1400sqft) for 1.2-1.4 million in a basic setting; the problem is that with all these rich people, theyre putting down double-price luxury appartments with things like gym, lounge, like a hotel more than a house, and that gets into 2-3million for the same appartment, and I just dont have that money but supply keeps going down in nor [22:12] <BBB> appts [22:12] <BBB> brooklyn is nice also& but Id like to buy in manhattan if I can [22:15] <ubitux> ffs i spent my w-e on something i hate with passion: threading [22:15] <ubitux> at least it seems to work now [22:16] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:9bcdb98839f3: avcodec/iff: Factorize av_pix_fmt_desc_get() out [22:16] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:3539d6c63a16: avcodec/iff: check pixfmt for rgb8 / rgbn [22:20] <BBB> threading is cool [22:23] <ubitux> no >:( [22:24] <ubitux> it's a pain, and it's cheating! [22:37] <ubitux> we lack a "developers" page on our web site [22:37] <ubitux> developers/contribute [22:37] <ubitux> something like, what's the main method to submit patches etc [22:37] <Compn> you mean donate or welcome to the porject ? [22:37] <Compn> ah [22:37] <ubitux> it's kind of mentioned in the mailing-list page but... [22:38] <ubitux> people often tend to look at github typically [22:38] <Compn> you mean like this page [22:38] <Compn> http://www.videolan.org/contribute.html [22:38] <Compn> kinda [22:38] <ubitux> yes [22:38] <ubitux> and merge the "MPlayer patch guidelines" into that page [22:43] <BBB> omg bitcoin [22:43] <Compnn> what i miss [22:43] <BBB> anyway hows that ffmpeg foundation thing going? [22:43] <BBB> like, can I request new hardware like a nice macbook pro laptop? my current one doesnt have avx2 [22:44] <Compn> yes, thats absolutely what donations are for [22:44] <Compn> so devels can work on specific hardwares [22:44] <Compn> assuming you are going to work on avx2 and not just .... get a new macbook :P [22:44] <Compn> not that i run the program or anything, just my opinion [22:46] <ubitux> BBB: https://ffmpeg.org/pipermail/ffmpeg-devel/2014-July/159929.html [22:47] <ubitux> we currently have about 3.5k it seems [22:48] <BBB> 3.5k doesnt get you a macbook, ok, Ill just buy my own then [22:48] <BBB> its funny, when I started ffmtech, it had 20-30k or so? where did all that money go [22:48] <ubitux> ah, ffmtech might have way more money [22:48] <BBB> I guess I had different methods of getting donations back then [22:48] <ubitux> http://ffmtech.org/projects.html [22:49] <BBB> omg diego inline asm conversion [22:49] <BBB> all whitespacers please alert [22:50] <BBB> ok anyway [22:50] <ubitux> it's funded it seems [22:50] <ubitux> ;) [22:50] <kierank> afaik it's not fonded [22:50] <ubitux> i'm just refering to the progress bar on the right [22:50] <kierank> oh yeah [22:50] <kierank> lol [22:51] Action: BBB gonna take kids out [22:51] <BBB> later [22:51] <ubitux> "hf" [23:32] Action: ubitux wonders why the diff planes of mpdecimate has increment of 4... and starts at x=8 [23:33] <cone-492> ffmpeg.git 03Chris \"Koying\" Browet 07release/2.3:6d2219e9f950: avcodec/h264_mp4toannexb_bsf: fix issue when sps/pps are already in the bistream (cherry picked from commit ad91bf854b559f1afc42fc06f8a7dfbb75def5c8) [23:33] <cone-492> ffmpeg.git 03Benoit Fouet 07release/2.3:7a1262fca37d: h264_mp4toannexb_bsf: account for consecutive IDR pictures. [23:33] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:594b1fa9615f: libavcodec/h264_parser: Increase parse_history, fix huge resolutions [23:33] <cone-492> ffmpeg.git 03Anshul Maheswhwari 07release/2.3:3ed27832e7b0: v4l2enc: adding AVClass [23:33] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:d6173ae341fa: avcodec/mpeg12dec: fix support for interlaced mpeg2 with missing last slice [23:33] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:aa0a8ef50e88: avcodec/dvdsub_parser: never return 0 when the input isnt 0 [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:76d59f1b34bb: avcodec/dvdsub_parser: Check buf_size before reading 32bit packet size [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:638c3aca645a: avcodec/dvdsub_parser: print message if packet is smaller than the packet size field [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:2eb72d5bdcd5: ffmpeg_opt: Use av_guess_codec() instead of AVOutputFormat->*codec [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:b63ec0cb0fb5: avformat/tee: flip assigment direction [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:fdfe94f4b16f: avcodec/wavpackenc: Fix log2sample() result value [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:7a6b5d7a867e: ffserver: initialize pbuffer in prepare_sdp_description() [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:901e275697de: avutil/cpu: add aarch64 entries to 2nd table [23:34] <cone-492> ffmpeg.git 03Anton Khirnov 07release/2.3:a72b7286e6db: cdgraphics: do not return 0 from the decode function [23:34] <cone-492> ffmpeg.git 03Felix Abecassis 07release/2.3:69281934935e: h264: fix interpretation of interleved stereo modes [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:ed15be7519af: avformat/matroskaenc: fix MAX_CUEPOINT_SIZE calculation [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:83956309cc16: avcodec/iff: check pixfmt for rgb8 / rgbn [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:fc57959fd57d: avcodec/iff: Factorize av_pix_fmt_desc_get() out [23:34] <cone-492> ffmpeg.git 03Michael Niedermayer 07release/2.3:c5129da726bb: update for FFmpeg 2.3.2 [23:41] <cone-492> ffmpeg.git 03Carl Eugen Hoyos 07master:788cf6f0c607: Remove panscan information in av_frame_copy_props() if resolution changes. [23:47] <cone-492> ffmpeg.git 03Diego Biurrun 07master:cc1d8c54c19d: avcodec: Postpone FF_IDCT_XVIDMMX removal until the next version bump [23:47] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:d177f275170c: Merge commit 'cc1d8c54c19dd14fb851e3e7a7793d6b3bd75e94' [23:51] <jamrial> michaelni: might be a good idea to undo commit 07262c71 to reinstate the FF_API_IDCT_XVIDMMX guards [23:53] <cone-492> ffmpeg.git 03Martin Storsjö 07master:08b94f160a2c: configure: Don't pass MSVC compiler options -M[TD] to armasm [23:53] <cone-492> ffmpeg.git 03Michael Niedermayer 07master:77273456137f: Merge commit '08b94f160a2c966bb83e32bde0e52246fafa2155' [23:55] <michaelni> jamrial, hmm, possible but imho we should keep the command line support longer than the next bump, its just one line [23:57] <jamrial> fine by me [00:00] --- Mon Aug 11 2014

1 0

ffmpeg.log.20140810
by burek 11 Aug '14

11 Aug '14

[00:37] <fajung> I've some avi (xvid/mp3) files and I want to convert them to mp4(x264/aac), how could I do it without losing vid quality ? [00:39] <c_14> If I have to be pedantic, use -qp 0. In all other cases, read https://trac.ffmpeg.org/wiki/Encode/H.264 [00:39] <diesel420> you cant transcode from xvid to x264 without losing quality [00:40] <c_14> x264 does have a lossless compression mode, so you can. [00:40] <diesel420> oh, sorry [00:40] <c_14> It just won't be any smaller (probably). [00:40] <Dark-knight> mp4 container supports mp3 [00:41] <fajung> yes, but my smart TV is not that smart and has a bug with mp3 sound in videos, I get a 3 sec without sound, but with aac, no problems [00:45] <fajung> thanks for the help [01:50] <llogan> fajung: try -crf 18. it'll be goodnuff. [01:50] <llogan> as mentioned in the link provided by c_14 [02:02] <ac_slater> hey all. I'm using avcodec_decode_video2() ... it seems the first time it's used, it returns the size but doesnt fill it's `got_picture_ptr` argument - indicating failure. I use it as the examples do and I cant seem to find out why it's failing [02:03] <ac_slater> the second time called, it works and does its business on the first frame ... meaning it didnt change internal state or anything ... using ffmpeg/libav* 1.2 [02:05] <ac_slater> 1.2.5 ... to be exact [02:08] <llogan> ac_slater: you may want to try the libav-user mailing list for FFmpeg library and API usage questions [02:30] <smo_> hi [02:35] <smo_> is it possible to do for exemple [02:35] <smo_> ffmpeg -i http://mirrorblender.top-ix.org/peach/bigbuckbunny_movies/big_buck_bunny_72… -sn -c:v libx264 -c:a libvorbis -f matroska bbunny.mp4 [02:35] <smo_> then start clc immediately to read the bbunny.mp4 AND get the original video length [02:36] <smo_> when i do things like that my players always report NAN:NAN or 00:00 as length [02:36] <smo_> so no seek nothing :( [19:28] <Mavrik> *grumbl* [19:28] <Mavrik> why is the filter chain eating my frames at the end of file [19:41] <jnvsor> Is it possible to change DAR with -c:v copy? [19:56] <chris_99> just a quick question, ffmpeg can decode quicktime right? [20:29] <hexhaxtron> How can I know if a video has a Creative Commons license? [20:31] <Chaz6> Ask the copyright holder [20:32] <hexhaxtron> Chaz6, isn't there metadata in the videos? [20:33] <Chaz6> In my experience very rarely [22:38] <Mavrik> chris_99, define "quicktime" :) [22:38] <Mavrik> quicktime isn't a format per se [22:38] <Mavrik> hmm... and why am I answering to a question from 2 hours ago [22:38] <chris_99> heh [22:38] <Mavrik> sorry :) [22:39] <chris_99> quicktime_ffmpeg_decode: unrecognized color model 12 -- doesn't mean anything to you, per chance [22:40] <chris_99> thats all i've got really, it's just output from a program called Cinelerra, and that's the only relavent error [22:40] <Mavrik> hmm, a quick google shows that it's a cinelerra error yes :) [22:41] <Mavrik> chris_99, can you do an ffprobe on your video and paste full output here? [22:41] <Mavrik> ah, found source which generates the error in cinelerra: http://cinelerra-cv.org/devcorner/doxy/svn_2.1_r1056/html/qtffmpeg_8c-sourc… :) [22:41] <chris_99> i'm not sure what that is, but in kdenlive and other video editors it works fine [22:42] <Mavrik> chris_99, what what is? [22:42] <chris_99> sorry, what ffprobe is [22:42] <chris_99> hmm so it looks like cinelerra is the issue, looking at that code? [22:42] <Mavrik> a video info tool, part of ffmpeg. [22:42] <Mavrik> run it on your video. [22:43] <chris_99> ok 1 tick [22:44] <chris_99> http://pastebin.com/PAKxavCe [22:45] <Mavrik> ah :) [22:45] <Mavrik> is that some kind of small camera output? [22:45] <Mavrik> Basically, cinelerra doesn't support YUVJ420P pixel format of video, just the standard ones [22:46] <Mavrik> and that's why you get the error message [22:46] <Mavrik> yuvj is a rather strange format tho, usually found in mjpegs and stuff [22:46] <chris_99> ah, thanks for your help Mavrik it's 1080p output from a canon 60d [00:00] --- Mon Aug 11 2014

1 0