[Ffmpeg-devel-irc] ffmpeg-devel.log.20120913

[burek]

[00:22] <ubitux> does anyone know if there is a way with freetype (and no fontconfig) to match a default common/monospaced font?
[00:29] <ubitux> mmh looks like this stuff will be a real pain
[00:30] <ubitux> fuck that shit, let's go with ff_vga16_font
[01:01] <saste> Daemon404, what about var.first.second = val; 
[01:01] <saste> is it supported by the converter?
[01:03] <saste> or is it a problem only with initializers?
[03:40] <CIA-57> ffmpeg: 03Michael Niedermayer 07master * rddabecbbf5 10ffmpeg/libavcodec/ (vc1.c vc1.h): 
[03:40] <CIA-57> ffmpeg: vc1dec: export adv profile color format indication
[03:40] <CIA-57> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[04:51] <CIA-57> ffmpeg: 03Ronald S. Bultje 07release/0.6 * r371e221d63 10ffmpeg/ (configure libavcodec/dxva2_internal.h): (log message trimmed)
[04:51] <CIA-57> ffmpeg: dxva2: include dxva.h if found
[04:51] <CIA-57> ffmpeg: Apparently, some build environments require dxva.h even for dxva2,
[04:51] <CIA-57> ffmpeg: while others lack this header entirely. Including it conditionally
[04:51] <CIA-57> ffmpeg: allows building in both cases.
[04:51] <CIA-57> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[04:51] <CIA-57> ffmpeg: (cherry picked from commit fa84506177f0246b30d4ea6a99ee5d419f3e4550)
[05:04] <CIA-57> ffmpeg: 03Ronald S. Bultje 07release/0.8 * r9a5e81235e 10ffmpeg/ (configure libavcodec/dxva2_internal.h): (log message trimmed)
[05:04] <CIA-57> ffmpeg: dxva2: include dxva.h if found
[05:04] <CIA-57> ffmpeg: Apparently, some build environments require dxva.h even for dxva2,
[05:04] <CIA-57> ffmpeg: while others lack this header entirely. Including it conditionally
[05:04] <CIA-57> ffmpeg: allows building in both cases.
[05:04] <CIA-57> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[05:04] <CIA-57> ffmpeg: (cherry picked from commit fa84506177f0246b30d4ea6a99ee5d419f3e4550)
[05:13] <CIA-57> ffmpeg: 03Ronald S. Bultje 07release/0.9 * r3c3ee65ebb 10ffmpeg/ (configure libavcodec/dxva2_internal.h): (log message trimmed)
[05:13] <CIA-57> ffmpeg: dxva2: include dxva.h if found
[05:13] <CIA-57> ffmpeg: Apparently, some build environments require dxva.h even for dxva2,
[05:13] <CIA-57> ffmpeg: while others lack this header entirely. Including it conditionally
[05:13] <CIA-57> ffmpeg: allows building in both cases.
[05:13] <CIA-57> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[05:13] <CIA-57> ffmpeg: (cherry picked from commit fa84506177f0246b30d4ea6a99ee5d419f3e4550)
[05:17] <CIA-57> ffmpeg: 03Ronald S. Bultje 07release/0.10 * r0f54c97f58 10ffmpeg/ (configure libavcodec/dxva2_internal.h): (log message trimmed)
[05:17] <CIA-57> ffmpeg: dxva2: include dxva.h if found
[05:17] <CIA-57> ffmpeg: Apparently, some build environments require dxva.h even for dxva2,
[05:17] <CIA-57> ffmpeg: while others lack this header entirely. Including it conditionally
[05:17] <CIA-57> ffmpeg: allows building in both cases.
[05:17] <CIA-57> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[05:17] <CIA-57> ffmpeg: (cherry picked from commit fa84506177f0246b30d4ea6a99ee5d419f3e4550)
[05:17] <CIA-57> ffmpeg: 03jamal 07release/0.10 * r50e6e494c9 10ffmpeg/library.mak: 
[05:17] <CIA-57> ffmpeg: build: Fix some paths in uninstall-libs
[05:17] <CIA-57> ffmpeg: Folder and file names weren't being separated with a slash.
[05:17] <CIA-57> ffmpeg: This resulted in .dll.a, .lib and .def files not being removed on uninstall.
[05:17] <CIA-57> ffmpeg: Signed-off-by: Alexander Strasser <eclipse7 at gmx.net>
[05:17] <CIA-57> ffmpeg: (cherry picked from commit 49440853d0c1e740daee0e2df1e65d5e67b1ad6b)
[05:18] <CIA-57> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[05:33] <ubitux> http://lucy.pkh.me/loudness.webm yepee
[05:33] <ubitux> now let's go for loudness range and integrated loudness 
[05:43] <llogan> i have no idea what i just watched
[06:17] <CIA-57> ffmpeg: 03Michael Niedermayer 07release/0.11 * rae8491ce33 10ffmpeg/configure: (log message trimmed)
[06:17] <CIA-57> ffmpeg: configure: fix sctp check.
[06:17] <CIA-57> ffmpeg: This fixes build failures on debian/kfreebsd
[06:17] <CIA-57> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[06:17] <CIA-57> ffmpeg: (cherry picked from commit bed63a5e98b1ba1f3bca6c81ea819666c2d81fde)
[06:17] <CIA-57> ffmpeg: Conflicts:
[06:17] <CIA-57> ffmpeg:  configure
[06:17] <CIA-57> ffmpeg: 03Ronald S. Bultje 07release/0.11 * rb45a8f6d24 10ffmpeg/ (configure libavcodec/dxva2_internal.h): (log message trimmed)
[06:17] <CIA-57> ffmpeg: dxva2: include dxva.h if found
[06:17] <CIA-57> ffmpeg: Apparently, some build environments require dxva.h even for dxva2,
[06:17] <CIA-57> ffmpeg: while others lack this header entirely. Including it conditionally
[06:17] <CIA-57> ffmpeg: allows building in both cases.
[06:17] <CIA-57> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[06:17] <CIA-57> ffmpeg: (cherry picked from commit fa84506177f0246b30d4ea6a99ee5d419f3e4550)
[06:17] <CIA-57> ffmpeg: 03jamal 07release/0.11 * rd14ce69869 10ffmpeg/doc/Makefile: 
[06:17] <CIA-57> ffmpeg: doc/Makefile: Add missing $(HOSTEXESUF) to print_options
[09:17] <cbsrobot-> ubitux: ebu128 ?
[09:40] <CIA-57> ffmpeg: 03Carl Eugen Hoyos 07release/0.10 * r2cf6afffe5 10ffmpeg/libavformat/swfenc.c: 
[09:40] <CIA-57> ffmpeg: Fix muxing mjpeg in swf.
[09:40] <CIA-57> ffmpeg: (cherry picked from commit 7680d99b4302e476076cc1b8f2567f47c2aaef4d)
[09:40] <CIA-57> ffmpeg: 03Carl Eugen Hoyos 07release/0.11 * r72791e41d4 10ffmpeg/libavformat/swfenc.c: 
[09:40] <CIA-57> ffmpeg: Fix muxing mjpeg in swf.
[09:40] <CIA-57> ffmpeg: (cherry picked from commit 7680d99b4302e476076cc1b8f2567f47c2aaef4d)
[09:40] <CIA-57> ffmpeg: 03Carl Eugen Hoyos 07release/0.5 * r09a278fdd1 10ffmpeg/libavformat/swfenc.c: 
[09:40] <CIA-57> ffmpeg: Fix muxing mjpeg in swf.
[09:40] <CIA-57> ffmpeg: (cherry picked from commit 79234907128390d775ceb1206dd0fed111a17896)
[09:40] <CIA-57> ffmpeg: 03Carl Eugen Hoyos 07release/0.9 * rf7d7b01e81 10ffmpeg/libavformat/swfenc.c: 
[09:40] <CIA-57> ffmpeg: Fix muxing mjpeg in swf.
[09:40] <CIA-57> ffmpeg: (cherry picked from commit 7680d99b4302e476076cc1b8f2567f47c2aaef4d)
[09:40] <CIA-57> ffmpeg: 03Carl Eugen Hoyos 07release/0.7 * r3ed505863b 10ffmpeg/libavformat/swfenc.c: 
[09:40] <CIA-57> ffmpeg: Fix muxing mjpeg in swf.
[12:03] <CIA-57> ffmpeg: 03Nicolas George 07release/0.11 * r3593f577b0 10ffmpeg/libavfilter/audio.c: lavfi/audio: fix an invalid boundary check.
[12:04] <CIA-57> ffmpeg: 03Nicolas George 07release/0.11 * r05e64b51bb 10ffmpeg/libavfilter/af_aresample.c: 
[12:04] <CIA-57> ffmpeg: lavfi/aresample: set buffer properties after copy.
[12:04] <CIA-57> ffmpeg: The aresample filter changes the format, layout and/or sample rate,
[12:04] <CIA-57> ffmpeg: it can not copy them blindly from its input.
[12:04] <CIA-57> ffmpeg: Fix trac ticket 1671 because af_volume uses the properties from the
[12:04] <CIA-57> ffmpeg: buffer and not from the link.
[12:13] <CIA-57> ffmpeg: 03Nicolas George 07master * rb5bea2f960 10ffmpeg/libavfilter/af_aresample.c: 
[12:13] <CIA-57> ffmpeg: lavfi/aresample: set buffer properties after copy.
[12:13] <CIA-57> ffmpeg: The aresample filter changes the format, layout and/or sample rate,
[12:13] <CIA-57> ffmpeg: it can not copy them blindly from its input.
[12:13] <CIA-57> ffmpeg: Related to trac ticket #1671.
[13:10] <ubitux> cbsrobot-: yes
[13:16] <cbsrobot-> ubitux: nice
[13:55] <burek> I have to ask a question here, because I've gotten different answers so far, whenever I asked this kind of question, so I'd like to be sure once and for all
[13:56] <burek> how can i take all the videos in a folder and concatenate them all into one output file (encoded normally, not using -vcodec copy)
[13:56] <burek> what is the "proper way" and is there any example command that can help me understand how to do this?
[14:02] <saste> burek: didn't your write a concat script?
[14:03] <burek> yes, but ffmpeg has got a big time bug
[14:03] <burek> which prevents it from using pipes
[14:03] <burek> so it's kinda useless
[14:03] <saste> uhm which one??
[14:03] <burek> mostly it stuck when people test it
[14:03] <saste> what bug?
[14:03] <burek> well, we are arguing if it is a bug or a feature
[14:04] <burek> but, let me give you a link to the ticket
[14:05] <burek> http://ffmpeg.org/trac/ffmpeg/ticket/1663
[14:07] <ubitux> saste: broken markup in #1731 :)
[14:07] <ubitux> (ohai)
[14:10] <ubitux> :)
[14:10] <cbsrobot-> burke, did you try to run two ffmpeg instances for decoding and 1 for encoding ?
[14:11] <cbsrobot-> burek: ^ (sorry)
[14:11] <burek> oh I see
[14:11] <cbsrobot-> meaning :
[14:11] <burek> yeah, I guess I could try that
[14:11] <cbsrobot-> shell 1:
[14:11] <cbsrobot-> ffmpeg -y -i 1.flv -map 0:a -f u16le -acodec pcm_s16le -ac 2 -ar 44100 /tmp/aaa &
[14:11] <cbsrobot-> shell 2:
[14:12] <cbsrobot-> ffmpeg -y -i 1.flv -map 0:v -f yuv4mpegpipe -vcodec rawvideo /tmp/vvv &
[14:12] <burek> yes, yes, I get it :) and shell 3 combines those outputs and encodes? :)
[14:12] <cbsrobot-> ok
[14:12] <burek> I'll test it right away, thanks for the hint :beer: :)
[14:14] <cbsrobot-> I think I did it that way a few months ago
[14:15] <burek> hmh
[14:15] <burek> it gets asynced that way (a/v)
[14:15] <cbsrobot-> if you decode files in multiple formats just make sure all the relevant parameters are corect
[14:15] <burek> wait, I need to switch shells
[14:15] <cbsrobot-> f.ex: for audio
[14:16] <cbsrobot-> -f u24le c:a pcm_s24le -ac 2 -ar 48000
[14:16] <cbsrobot-> f.ex: for video
[14:17] <cbsrobot-> -vf scale=1024:576 -pix_fmt yuv420p -f yuv4mpegpipe
[14:19] <cbsrobot-> and make sure, when you add them together to specify the same audio format again
[14:19] <burek> I see, well I used the same thing that faw item used: -vn -f u16le -acodec pcm_s16le -ac 2 -ar 44100 $TMP/mcs_a1 -an -f yuv4mpegpipe -vcodec rawvideo $TMP/mcs_v1
[14:19] <burek> oh :)
[14:20] <burek> u16le and pcm_s16le
[14:20] <burek> what a fail
[14:20] <burek> :D
[14:21] <burek> wait, is it correct to use U in format and S in codec?
[14:21] <cbsrobot-> i stumbled on that aswell
[14:21] <cbsrobot-> but it is correct
[14:21] <cbsrobot-> I think I tried the other way around, and it did not work
[14:22] <burek> shouldn't it be -f s24le c:a pcm_s24le
[14:22] <cbsrobot-> it won't work - if I remember well
[14:22] <saste> what happened of the ffserver rewrite in libav/gsoc?
[14:23] <merbanan> saste: ask luca
[14:23] <merbanan> there where some work but not complete
[14:23] <saste> merbanan,  do you have a link to a repo?
[14:24] <merbanan> no, lu_zero should have
[14:28] <saste> burek, why /tmp/aaa /tmp/vvv and then /tmp/mcs_aaa /tmp/mcs_vvv?
[14:28] <burek> it's a typo
[14:29] <burek> http://ffmpeg.org/trac/ffmpeg/wiki/How%20to%20concatenate%20(join%2C%20merge)%20media%20files
[14:29] <burek> the script was created from the faq item
[14:29] <burek> nothing new added except it was automated
[14:29] <burek> and while I was testing manually to see what's wrong
[14:29] <burek> I've used /tmp/aaa and /tmp/vvv
[14:29] <burek> so it left there while copy/pasting
[14:30] <burek> but, it's just a typo, the issue still exists if you correct it to mcs_aaa and mcs_vvv
[14:30] <cbsrobot-> burek, never copy paste after http://users.telenet.be/rgow/image/jupiler040.jpg !
[14:30] <burek> :D
[14:31] <burek> this is awesome! :D
[14:36] <burek> btw, what is the proper way to concat multiple input files using concat filter/format/protocol whatever exists in ffmpeg right now (and encode into some custom output, so no -vcodec copy)?
[14:37] <relaxed> burek: -i concat:1.mpg\|2.mpg -map 0 -c copy
[14:38] <relaxed> by the way, it's possible to use pipes and omit fifos from that script.
[14:38] <saste> burek, as far as i understand it, is not a threading deadlock problem
[14:39] <saste> it is a process deadlock, so it won't change much if the single process is using many threads
[14:39] <burek> I understood ffmpeg uses 1 thread to write both audio/video, so while it wants to write to audio pipe, the other side is ready to read the video pipe
[14:39] <burek> and that's where the problem starts
[14:40] <saste> thread deadlock are related to a single process using many threads
[14:40] <burek> if those were threads, it wouldn't happen, because both threads in the 1st process would make both the pipes always ready for read
[14:40] <saste> this is a process deadlock
[14:41] <burek> relaxed, does that work for any kind of input?
[14:42] <saste> burek, i'm not sure how the processes access the pipes, but it may be this problem
[14:43] <saste> process one is writing to pipe 1, since the queue is full it will wait the other process to read it
[14:44] <saste> process 2 is currently waiting on pipe2 queue, which will never be filled because process 1 hanged
[14:44] <saste> deadlock
[14:44] <burek> yes, I understand that
[14:44] <saste> you could increase the size of the queue pipe, but that will only decrease the possibility that the deadlock may occur
[14:45] <saste> changing the application logic would be more difficult, process1 may stop to wait for the queue1 to be emptied and switch to pipe2 after a given interval
[14:45] <saste> but this is not apparently how ffmpeg works
[14:46] <burek> well, I know, that's why I said I'm not sure if it's a bug or a feature
[14:47] <burek> but it seems more natural to me to put different outputs into different threads
[14:54] <burek> cbsrobot-, your logic works, now I just need to test if a/v is in sync
[14:55] <saste> burek, and turn ffmpeg into a multithreaded application?
[14:56] <saste> but even in that case it won't necessarily fix the issue, since you need to synchronize several threads
[14:57] <saste> right now i believe there is a blocking call which could (theoretically) replaced by a non-blocking one, through some timeout mechanism
[14:58] <burek> saste, yes, I realize that now, testing cbsrobot-'s idea, and I also believe async will happen.. it would really be beautiful if ffmpeg could support some kind of full concatenation or input playlist, so that all the inputs could be processed one by one and encoded into one output
[14:59] <saste> anyway yes i consider it a bug, but not a bug which can be easily fixed
[14:59] <burek> saste, it might help to write output in smaller chunks or something (because of pipe size)
[15:00] <saste> yes , one writer and two readers, or two readers and one writer, or two readers and two writers should overcome the problem
[15:01] <burek> although, it could be a good idea to make it optional like -small_chunks not to hurt general ffmpeg's performance
[15:01] <burek> anyway, I'll change the script for now to use real temp files, which will hurt, but still it'll work :)
[15:01] <burek> hm, or it wouldnt...
[15:02] <burek> i need a beer :)
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * r7ef4323405 10ffmpeg/libavformat/rtpdec_jpeg.c: 
[15:29] <CIA-53> ffmpeg: rtpdec_jpeg: Write the DHT section properly
[15:29] <CIA-53> ffmpeg: Currently the size header of the generated DHT section is
[15:29] <CIA-53> ffmpeg: incorrect, making the mjpeg decoder just skip it. Since the
[15:29] <CIA-53> ffmpeg: written huffman tables are the default ones, this failure had
[15:29] <CIA-53> ffmpeg: gone undetected.
[15:29] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * r932d8300d3 10ffmpeg/libavformat/rtpdec_jpeg.c: 
[15:29] <CIA-53> ffmpeg: rtpdec_jpeg: Merge two if statements
[15:29] <CIA-53> ffmpeg: This makes the code more readable and robust.
[15:29] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * r43957fcc71 10ffmpeg/libavformat/rtpdec_jpeg.c: 
[15:29] <CIA-53> ffmpeg: rtpdec_jpeg: Simplify writing of the jpeg header
[15:29] <CIA-53> ffmpeg: Generalize writing of any number of qtables. Don't manually write
[15:29] <CIA-53> ffmpeg: 16 bit values in two separate calls.
[15:29] <CIA-53> ffmpeg: Write out the numbers the way they are constructed, not just the
[15:29] <CIA-53> ffmpeg: final values.
[15:29] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * rcbaa9eeda3 10ffmpeg/libavformat/rtpdec_jpeg.c: 
[15:29] <CIA-53> ffmpeg: rtpdec_jpeg: Add more comments about the fields in the SOF0 section
[15:29] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * ra218deb856 10ffmpeg/libavformat/rtpdec_jpeg.c: 
[15:29] <CIA-53> ffmpeg: rtpdec_jpeg: Store and reuse old qtables for q values 128-254
[15:29] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * r1de9317bd0 10ffmpeg/libavformat/rtpdec_jpeg.c: 
[15:29] <CIA-53> ffmpeg: rtpdec_jpeg: Fold the default qtables case into an existing if statement
[15:29] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * rc3bcd22ed3 10ffmpeg/libavformat/rtpdec_jpeg.c: 
[15:29] <CIA-53> ffmpeg: rtpdec_jpeg: Error out on other unsupported type values as well
[15:29] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:29] <CIA-53> ffmpeg: 03Tomas Härdin 07master * r5e3572893d 10ffmpeg/libavformat/img2dec.c: 
[15:29] <CIA-53> ffmpeg: img2dec: always close AVIOContexts
[15:29] <CIA-53> ffmpeg: Zero sized files would cause the contexts to leak away.
[15:29] <CIA-53> ffmpeg: 03Christian Schmidt 07master * r07584eaf4a 10ffmpeg/libavformat/mpegts.c: 
[15:29] <CIA-53> ffmpeg: mpegts: check substreams before discarding
[15:29] <CIA-53> ffmpeg: Allow to extract the AC3 core from TrueHD with the "copy" codec.
[15:29] <CIA-53> ffmpeg: Signed-off-by: Luca Barbato <lu_zero at gentoo.org>
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * r62c9ae11a7 10ffmpeg/ (6 files in 2 dirs): (log message trimmed)
[15:29] <CIA-53> ffmpeg: Add a smooth streaming segmenter muxer
[15:29] <CIA-53> ffmpeg: This muxer splits the output from the ismv muxer into individual
[15:29] <CIA-53> ffmpeg: files, in realtime.
[15:29] <CIA-53> ffmpeg: The same can also be done by the standalone tool ismindex, but this
[15:29] <CIA-53> ffmpeg: muxer is needed for doing it in realtime (especially for live
[15:29] <CIA-53> ffmpeg: streams that need extra handling for updating the lookahead fields
[15:29] <CIA-53> ffmpeg: 03Martin Storsjö 07master * r1743938df1 10ffmpeg/libavformat/rtpdec_jpeg.c: 
[15:29] <CIA-53> ffmpeg: rtpdec_jpeg: Disallow using the reserved q values
[15:29] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:29] <CIA-53> ffmpeg: 03Michael Niedermayer 07master * rc942e8b1d7 10ffmpeg/: (log message trimmed)
[15:29] <CIA-53> ffmpeg: Merge commit '07584eaf4a95db3f11d3bc411f9786932829e82b'
[15:30] <CIA-53> ffmpeg:  file: Add an avoption for disabling truncating existing files on open
[15:30] <CIA-53> ffmpeg:  img2dec: always close AVIOContexts
[15:41] <CIA-53> ffmpeg: 03Tim Walker 07master * r8919fee7d2 10ffmpeg/ (4 files in 2 dirs): 
[15:41] <CIA-53> ffmpeg: lavu/audioconvert: add a second low frequency channel.
[15:41] <CIA-53> ffmpeg: Can be used by DTS-HD, TrueHD and E-AC-3, among others.
[15:41] <CIA-53> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[15:41] <CIA-53> ffmpeg: 03Tim Walker 07master * r3844572887 10ffmpeg/libavcodec/mlp_parser.c: 
[15:41] <CIA-53> ffmpeg: mlp_parser: avoid mapping multiple disctinct TrueHD channels to the same Libav channel.
[15:41] <CIA-53> ffmpeg: Signed-off-by: Justin Ruggles <justin.ruggles at gmail.com>
[15:41] <CIA-53> ffmpeg: 03Mans Rullgard 07master * r97cb9236cf 10ffmpeg/libavcodec/ (8 files in 2 dirs): 
[15:41] <CIA-53> ffmpeg: ac3: move ac3_downmix() from dsputil to ac3dsp
[15:41] <CIA-53> ffmpeg: Signed-off-by: Mans Rullgard <mans at mansr.com>
[15:41] <CIA-53> ffmpeg: 03Martin Storsjö 07master * r7b07fab567 10ffmpeg/libavformat/os_support.h: 
[15:41] <CIA-53> ffmpeg: os_support: Include io.h instead of direct.h on mingw32ce
[15:41] <CIA-53> ffmpeg: Windows CE doesn't have neither mkdir nor _mkdir officially (only
[15:41] <CIA-53> ffmpeg: CreateDirectoryW), but mingw32ce has compat wrappers with these names
[15:41] <CIA-53> ffmpeg: (declared in io.h since direct.h is unavailable).
[15:41] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:41] <CIA-53> ffmpeg: 03Martin Storsjö 07master * rca6b544ac9 10ffmpeg/ (configure libavformat/os_support.h): 
[15:41] <CIA-53> ffmpeg: os_support: Choose between direct.h and io.h using a configure check
[15:41] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:41] <CIA-53> ffmpeg: 03Michael Niedermayer 07master * re3e09f2bad 10ffmpeg/: (log message trimmed)
[15:41] <CIA-53> ffmpeg: Merge remote-tracking branch 'qatar/master'
[15:41] <CIA-53> ffmpeg: * qatar/master:
[15:41] <CIA-53> ffmpeg:  os_support: Choose between direct.h and io.h using a configure check
[15:41] <CIA-53> ffmpeg:  os_support: Include io.h instead of direct.h on mingw32ce
[15:41] <CIA-53> ffmpeg:  x86: ac3dsp: Only refer to the ac3_downmix_sse symbol if it has been declared
[15:41] <CIA-53> ffmpeg:  swscale: Remove two bogus asserts
[15:42] <CIA-53> ffmpeg: x86: ac3dsp: Only refer to the ac3_downmix_sse symbol if it has been declared
[15:42] <CIA-53> ffmpeg: This fixes building without inline assembly.
[15:42] <CIA-53> ffmpeg: Signed-off-by: Martin Storsjö <martin at martin.st>
[15:56] <burek> surprisingly, cbsrobot-'s idea works, to split the decoding of inputs into 2 ffmpeg processes, to avoid deadlock
[15:56] <burek> even the outputs are in sync :)
[15:56] <burek> I owe a big one to you cbsrobot- :))) :beer: :)
[15:57] <cbsrobot-> burek: np - just call me when you party !
[15:57] <burek> count on it :)
[16:44] <saste> uhm we got two donations in three months
[16:44] <saste> for a total of 66$
[16:46] <cbsrobot-> at least we can buy the candles for the xmas tree &
[16:46] <burek> :D
[16:46] <saste> or a pizza to share, and some beer for burek
[16:51] <Daemon404> saste, doesnt flv require the initial offset to be 0
[16:51] <Daemon404> or am i mistaken...
[16:52] <CIA-53> ffmpeg: 03Michael Niedermayer 07master * r0ed0f2e566 10ffmpeg/libswscale/swscale.c: 
[16:52] <CIA-53> ffmpeg: sws: revert: "swscale: Remove two bogus asserts"
[16:52] <CIA-53> ffmpeg: Also change them to av_assert
[16:52] <CIA-53> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:52] <CIA-53> ffmpeg: 03Michael Niedermayer 07master * rbc3d5173cc 10ffmpeg/libavformat/libavformat.v: 
[16:52] <CIA-53> ffmpeg: libavformat.v: export ffurl_protocol_next
[16:52] <CIA-53> ffmpeg: This is used by avidemux, and is likely usefull to others too.
[16:52] <CIA-53> ffmpeg: Patch by: gruntster (Avidemux Rev 7990  2012-05-30 13:02:27)
[16:52] <CIA-53> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:52] <CIA-53> ffmpeg: 03Michael Niedermayer 07master * r75f3e0ad95 10ffmpeg/libavcodec/mpegvideo.c: 
[16:52] <CIA-53> ffmpeg: mpegvideo: skip ff_print_debug_info() for VDPAU
[16:52] <CIA-53> ffmpeg: Based on code by: MEANX / gruntster (Avidemux Rev 7990  2012-05-30 13:02:27)
[16:52] <CIA-53> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[16:55] <saste> Daemon404, I have many FLV samples with the initial timestamp != 0 (generated by adobe FMS)
[16:56] <JEEB> yeah
[16:56] <JEEB> flv concateration is a messy job
[16:58] <saste> michaelni, you need to cd doc/examples to make examples
[16:58] <CIA-53> ffmpeg: 03Derek Buitenhuis 07master * r59db014b5b 10ffmpeg/libavcodec/libx264.c: (log message trimmed)
[16:58] <CIA-53> ffmpeg: Revert "lavc/libx264: remap X264_LOG_INFO loglevel from AV_LOG_INFO to VERBOSE"
[16:58] <CIA-53> ffmpeg: You should not need to enable verbose logging and all the output that comes
[16:58] <CIA-53> ffmpeg: with it in order to get things like the obtained bitrate from a first
[16:58] <CIA-53> ffmpeg: pass encode. Many things parse this output.
[16:58] <CIA-53> ffmpeg: Also, since it is marked as info in libx264, it should be marked as
[16:58] <CIA-53> ffmpeg: info in libavcodec as well.
[16:58] <saste> this is supposed to be run outside of the ffmpeg build environment
[17:03] <michaelni> saste, how could a user know this ?
[17:04] <michaelni> i mean the cd doc/examples
[17:04] <saste> michaelni, the directory is meant to be independent from the rest of ffmpeg
[17:04] <michaelni> again how could a user know this ?
[17:04] <saste> and examples are usually installed in share/doc/ffmpeg
[17:04] <saste> the user has eyes, and he can see that there is a Makefile
[17:05] <saste> i can add a README if you think so
[17:05] <michaelni> thats uhm useless
[17:05] <michaelni> the user doesnt read that :)
[17:05] <michaelni> make doc/examples/... should print it IMHO
[17:07] <michaelni> i mean something along the lines of "the examples are seperate from the ffmpeg build system please execute make ... in ,,,,"
[17:09] <michaelni> besides
[17:09] <michaelni> "make decoding_encoding" in the examples dir also doesnt work
[17:10] <michaelni> nor does a plain male
[17:10] <michaelni> maKe
[17:11] <michaelni> decoding_encoding.c:117:34: error: ‘AV_CODEC_ID_MP2’ undeclared (first use in this function)
[17:11] <michaelni> amongth others
[17:11] <mateo`> *
[17:12] <ubitux> michaelni: PKG_CONFIG_PATH=pc-uninstalled make works for me
[17:13] <saste> michaelni, you need to have a recent ffmpeg installed in the regular path
[17:13] <ubitux> PKG_CONFIG_PATH=pc-uninstalled should not be necessary if you have the current version of ffmpeg deployed
[17:13] <ubitux> maybe that could be part of a README
[17:13] <saste> so you need to set PKG_CONFIG_PATH
[17:13] <nevcairiel> requiring an installed ffmpeg is rather annoying just to build the examples that live in the source tree
[17:13] <ubitux> nevcairiel: PKG_CONFIG_PATH=pc-uninstalled make
[17:14] <ubitux> i guess we could have:
[17:14] <ubitux> dev: PKG_CONFIG_PATH=pc-uninstalled
[17:14] <ubitux> dev: all
[17:14] <ubitux> and we would just "make dev"
[17:14] <ubitux> dev, intree, whatever
[17:14] <saste> ubitux, you could even link it to the regular build, so that fate will notice build errors in examples
[17:15] <ubitux> i'm too busy ;)
[17:15] <saste> right now the requirements are: PKG_CONFIG_PATH pointing to a path with a recent ffmpeg install
[17:16] <michaelni> saste, IMHO if one tries to build the examples from within a git checkout (compared to share/doc) they should use the checkouts headers to be build
[17:16] <michaelni> otherwise you cannot checkout and test an old version at all
[17:16] <ubitux> michaelni: the problem is that these examples have to work with a deployed installed as well
[17:17] <ubitux> so the user just has to cp -r /usr/share/ffmpeg/examples and then just run "make" into this dir
[17:17] <ubitux> that works ATM
[17:17] <ubitux> as a developer, you have to type "PKG_CONFIG_PATH=pc-uninstalled make"
[17:17] <ubitux> a README explaining this could be added if necessary
[17:18] <ubitux> changing the example to make them work in-tree isn't a use case for most users developing their app
[17:18] <ubitux> (since they will be building their code against the installed library)
[17:18] <saste> indeed
[17:19] <ubitux> that's why the current examples are standalone codes working against an installed FFmpeg
[17:22] <ubitux> burek: you could add this script to the ffmpeg git in tools/
[17:22] <burek> only if I knew how :)
[17:22] <ubitux> that would avoid maintaining it on the wiki
[17:23] <ubitux> burek: cp mmcat.sh ffmpeg/tools
[17:23] <ubitux> git add tools/mmcat.sh
[17:23] <ubitux> git commit -m "tools: add mmcat.sh sript to concat medias."
[17:23] <ubitux> git format -1 --to ffmpeg-devel at ffmpeg.org
[17:23] <ubitux> format-patch*
[17:23] <ubitux> mmh
[17:23] <ubitux> git send-email sorry
[17:24] <ubitux> or format patch and then send it :p
[17:25] <ubitux> i wonder if Nicolas is going to continue the concat demuxer
[17:27] <cbsrobot-> what's the advance of the concat demuxer ?
[17:27] Action: cbsrobot- never tried it yet
[17:27] <ubitux> advance?
[17:27] <ubitux> benefit?
[17:27] <cbsrobot-> damn auto spelling 
[17:28] <cbsrobot-> advantage 
[17:28] <ubitux> i guess it would allow codec copy and would work with inputs ` mpeg
[17:28] <cbsrobot-> yeah or benefit, if you want
[17:28] <ubitux> (respectively in opposition with lavfi/concat and concat://)
[17:29] <cbsrobot-> ok - I need to try it then
[17:29] <ubitux> so you would be able concat .mp4 in a lossless op... i guess :)
[17:30] <cbsrobot-> btw ubitux - can you explain me something ?
[17:30] <burek> ubitux, with all regards, but it is far easier for me to edit the wiki article, than to remember all those git commands and switches :S
[17:30] <ubitux> cbsrobot-: i can try
[17:30] <ubitux> burek: as you wish, i don't care very much, that's for you :)
[17:30] <burek> if it will make some things better, I've got nothing against that idea
[17:31] <cbsrobot-> in drawutils.h there is a union in ffdrawcolor, whats the u32 and u16 for ?
[17:31] <burek> but it just seems too much to hassle with for me, that's all
[17:31] <ubitux> cbsrobot-: no idea, i didn't write nor used that code :p
[17:31] <cbsrobot-> and why only u8[4] is an array ?
[17:32] <ubitux> to access each component i guess
[17:32] <ubitux> like if you have rgbcolor = 0x00AABBCC
[17:32] <cbsrobot-> the union is defined as: union {&}  comp[MAX_PLANES]
[17:32] <ubitux> comp.u8[2] == 0xBB
[17:32] <ubitux> for example (i guess)
[17:33] <cbsrobot-> so I guess the comp[2].u8[0] == 0xBB
[17:33] <ubitux> IMO u16 should be u16[2]
[17:34] <ubitux> mmh?
[17:34] <ubitux> oh i see
[17:34] <ubitux> no i don't think so
[17:35] <cbsrobot-> see ff_draw_color
[17:35] <cbsrobot-> y -> comp[0].u8[0]
[17:35] <cbsrobot-> u -> comp[1].u8[0]
[17:36] <cbsrobot-> v -> comp[2].u8[0]
[17:36] <ubitux> maybe the u16 is for 10+ bits
[17:37] <ubitux> dunno, really
[17:37] <cbsrobot-> ok
[17:37] <cbsrobot-> thanks anyhow
[17:37] <ubitux> git blame and ask the guy
[17:37] <ubitux> (likely saste or nicolas)
[17:37] <cbsrobot-> nicolas' code is hard to follow
[17:37] <cbsrobot-> sometimes
[17:37] <ubitux> :)
[17:38] <cbsrobot-> i have vf_pad working for files with bpp > 8
[17:38] <cbsrobot-> but vf_ass if a bit harder
[19:36] <CIA-53> ffmpeg: 03Michael Niedermayer 07master * rbd3977267e 10ffmpeg/libavutil/x86_cpu.h: 
[19:36] <CIA-53> ffmpeg: x86_cpu.h: newline at the end to shutup some warnings in opencc
[19:36] <CIA-53> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[19:36] <CIA-53> ffmpeg: 03Michael Niedermayer 07master * r9d89d1d484 10ffmpeg/libswresample/resample.c: 
[19:36] <CIA-53> ffmpeg: swr/resample: mark mm_flags as potentially unused
[19:36] <CIA-53> ffmpeg: They are unused when asm is disabled
[19:36] <CIA-53> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
[20:18] <CIA-53> ffmpeg: 03Nicolas George 07master * rc262ddb8c5 10ffmpeg/libavfilter/audio.c: lavfi: set sample_rate in the default allocator.
[20:18] <CIA-53> ffmpeg: 03Nicolas George 07master * r8ad54a44eb 10ffmpeg/libavfilter/ (audio.c video.c): lavfi: check that buffer and link properties match.
[20:42] <Daemon404> why is ffmpeg's configure running perl
[20:43] <Daemon404> oh. docs.
[20:43] <gnafu> Daemon404: Because deep down, everything runs Perl.  It's camels all the way down!
[20:44] <Daemon404> i use perl a lot
[20:51] <CIA-53> ffmpeg: 03Stefano Sabatini 07master * r70b671a283 10ffmpeg/doc/examples/decoding_encoding.c: examples/decoding_encoding: apply various message log fixes
[20:51] <CIA-53> ffmpeg: 03Stefano Sabatini 07master * rbcf93f78a5 10ffmpeg/doc/examples/decoding_encoding.c: 
[20:51] <CIA-53> ffmpeg: examples/decoding_encoding: replace "picture" with "frame" in encode_example()
[20:51] <CIA-53> ffmpeg: The use of "picture" in place of "frame" is confusing.
[21:00] <Daemon404> i wish valgrind would show what lien of asm (yasm) teh read occurs on
[21:00] <Daemon404> -_-
[21:08] <Daemon404> michaelni, having trouble with an overread in some mmx2 code (utvideo enc valgrind fate instance). perhaps you could look, when you get a chance
[21:08] <Daemon404> it actually crashes on some openbsds
[21:09] <cbsrobot-> Daemon404: not on all ?
[21:09] <Daemon404> depends i guess
[21:09] <Daemon404> but valgrind sees it
[21:51] <llogan> that guide needs some work
[21:53] <Daemon404> a pitfall almost everyone (seemingly taht guide too) falls into is compilign ffmpeg with every dep under the sun
[21:53] <Daemon404> for absolutely no reason
[21:53] <Daemon404> massive waste of time
[21:58] <Daemon404> hmm
[21:58] <Daemon404> same thing on a diff box for af_pan, but no crash this time:
[21:59] <Daemon404> -map_channel is forwarded to lavfi similarly to -af pan=0x3:c0=c1:c1=c0.
[21:59] <Daemon404> [pan @ 00B1F780] Syntax error near "?·¼pT$"
[21:59] <Daemon404> Output pad "default" for the filter "Parsed_anull_0" of type "anull" not connected to any destination
[21:59] <Daemon404> Error opening filters!
[22:00] <Daemon404> O_O
[22:00] <llogan> Daemon404: i just trimmed some of the dumb libraries (faad!), but i am not a mac luser.
[22:01] <Daemon404> i didnt think ffmpeg even HAD faad support now
[22:01] <Daemon404> for like 1-2 years
[22:04] <Daemon404> def looks liek a pointer problem
[22:08] <Daemon404> i see
[22:08] <Daemon404> lots of stuff is fialing in fate due ot screwed option parsing
[22:08] <Daemon404> fun -_-
[22:21] <CIA-53> ffmpeg: 03Stefano Sabatini 07master * r51fa653829 10ffmpeg/doc/examples/decoding_encoding.c: examples/decoding_encoding: clarify operations in the usage message
[23:03] <Yexo> Would a patch to support new input/output devices (asi/sdi) have a chance of being included even if it'd require to link against a proprietary (but freely available) library to use them?
[23:03] <saste> Yexo: good question
[23:04] <saste> Yexo, no free libraries/drivers?
[23:04] <Yexo> driver is open source, library is not and will not be
[23:04] <Yexo> but just the driver is not enough to actually use the cards
[23:05] <saste> how's that that there are no free libraries?
[23:05] <Yexo> the library is free, just not open source
[23:05] <Yexo> and it's not open source because it contains a lot of IP
[23:06] <saste> Yexo, link?
[23:06] <Yexo> dektec.com
[23:07] <Yexo> http://www.dektec.com/Downloads/Drivers.asp <- Linux SDK available there
[23:09] <saste> Yexo, try to ask on list to check if there is the interest/will for integrating such thing
[23:09] <saste> since we have no similar cases I can't say
[23:09] <Yexo> I'd do the coding myself (will probably do that anyway, whether or not there is interest from other or not)
[23:09] <saste> as for me I have no strong opinion
[23:09] <Yexo> I was mostly curious how people here would look at it (but I'll send a mail to the dev list)
[23:10] <saste> configure --enable-proprietary?
[23:11] <Yexo> how would it be different from nonfree?
[23:11] <Daemon404> found out why af_pan is crashing
[23:11] <Daemon404> sscanf is being borky
[23:11] <Daemon404> perhaps a bad format
[23:11] <Yexo> hmm, not really true
[23:11] <Yexo> it's different than nonfree
[23:12] <Daemon404> http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavfilter/af_pan.c;h=753522b76e802eb9c20279da784da0e783210dc2;hb=HEAD#l81
[23:12] <Daemon404> this one
[23:12] <Daemon404> sometimes gives a length of <some huge #>
[23:14] <saste> honestly I always found "--nonfree" pretty confusing
[23:16] <Yexo> not sure whether --enable-proprietary would be needed. Its incompatible with GPL but LGPL shouldn't be any problem
[00:00] --- Fri Sep 14 2012