[Ffmpeg-devel-irc] ffmpeg-devel.log.20140303

[burek]

[00:04] <Snowrider_> BtbN: Im more referring to the frame of like a bmp for example. I may just have a fundamental misunderstanding of this
[00:09] <J_Darnley> You probably can get it (I'm not familiar with AVFrame) but it will be encoded
[00:09] <J_Darnley> You should make the libraries decode it for you into whatever format you ask
[00:11] Action: Compn sees nicolas' comment fly over don's head
[00:11] <Compn> ehe
[00:12] <Compn> Snowrider_ : probably you need an image or video editor that has those things
[00:12] <Compn> Snowrider_ : like if you want to compare the colorpalette or whatever
[00:12] <Compn> that would be more advanced , ffmpeg doesnt have those features yet
[00:13] <Snowrider_> yeah seems that would be the proper way to do it. This is more of experimentation on my part
[00:13] <Snowrider_> to get firmiliar with the code and how it works
[00:13] <Compn> there are some things available in the debug options, but i dont think theres an rgb values thing
[00:13] <Compn> it would be interesting task to add an analyzer if you wanted to try
[00:14] <Compn> you could do it with the filter api 
[00:14] <Snowrider_> I will look into that
[00:14] <Compn> Snowrider_ : did you see this http://ffmpeg.zeranoe.com/forum/viewtopic.php?f=15&t=805
[00:14] <Snowrider_> I figured it would be useful in some circumstances.
[00:15] <Snowrider_> yeah i was looking through that
[00:15] <Snowrider_> what about this?
[00:15] <Snowrider_> https://stackoverflow.com/questions/1047200/extract-rgb-values-from-a-avframe-ffmpeg-in-c?rq=1
[00:17] <Snowrider_> the first array there i believe is the frame
[00:18] <Snowrider_> and for a picture it should always be zero
[00:18] <Snowrider_> since it is not a video 
[00:19] <Compn> no clue. i'm not a programmer
[00:20] <Compn> also this channel is for ffmpeg development, bugreports go to #ffmpeg and 3rd party libavcodec apps is in libav-user mailing list or the ffmpeg forum on gusari
[00:21] <Snowrider_> this isnt really a bug in any way just trying to add code to ffmpeg and experiment
[00:21] <Snowrider_> thanks tho
[00:21] <Compn> no problem :)
[00:21] <Compn> sometimes its hard to tell what people want when they ask questions here
[00:22] <Compn> but yeah outputting rgb values in an easy format, maybe tying it into json thing we have now, that would be useful to someone maybe
[00:23] <Snowrider_> yeah understandable can see why it would be hard to tell
[00:23] <Snowrider_> it is hard to ask the right questions
[00:24] <Compn> it takes a while to learn all of the terms and definitions for each part of the video too 
[00:27] <Snowrider_> yeah Im finding that out
[00:28] <Compn> hows the weather in utah? anything above freezing ?
[00:28] <Compn> got over 100 days under freezing here in michigan, cant even thaw the snow
[00:31] <J_Darnley> That's pretty impressive
[00:31] <Snowrider_> its pretty warm
[00:31] <Snowrider_> Im from michigan haha
[00:31] <Snowrider_> but livin in utah
[00:32] <Snowrider_> 56 degrees here now
[00:32] <Compn> 16F .... gonna be ... 16F tomorrow too
[00:33] <Snowrider_> My parents feel your pain
[00:49] <llogan> New GSoC FAQ. "Why was my project not accepted? Because there was no room for you." "How did other projects get accepted? There was room for them."
[00:59] <Compn> llogan : cant tell if joke or not :)
[00:59] <llogan> that's why it is funny
[00:59] <llogan> now laugh
[01:00] <Compn> i mean , thats in the official faq ?
[01:00] Action: Compn wonders if reynaldo or saste emailed google for the secret reason :P
[01:02] <llogan> no, it's a FAQ I'm making for them. the "secret" reason was just a repeat of "there was no room".
[01:03] <Compn> it was ?
[01:03] <Compn> why did she want it emailed then ?
[01:03] <llogan> Why was Lincoln buried on a hill?
[01:03] <llogan> Because he was dead.
[01:03] <llogan> that's how they answer
[01:03] <Compn> ah
[01:03] <Compn> forget it jake, its chinatown
[01:04] <llogan> i don't know why she wanted it emailed, and when she was emailed she didn't give more info
[01:04] Action: llogan has to go get a bus now since jalopy car is broken again
[01:04] <Compn> did whomever emailed her say he was authorized by the project to recieve the secret answer ?
[01:04] <Compn> hehe
[01:04] <Compn> maybe he didnt present the required credentials to see the confidential information
[01:05] <Compn> we're going to have to buy saste some presents, for doing all that gsoc work
[01:05] <Compn> j-b knew, he knew what was happening
[01:55] <rcombs> what's the starting (i.e. time=0) PTS for MPEG-2?
[02:39] <cone-265> ffmpeg.git 03Michael Niedermayer 07master:2dcaa1b9d142: avformat/utils: detect MPEG streams with faulty DTS and discard affected DTS
[05:23] <Zeranoe> If I save a frame with FFmpeg, will it save it with the same display dimensions as ffplay shows?
[05:29] <Compn> Zeranoe : if the container for the frame you save saves the aspect ratio and dimensions
[05:30] <Compn> avi, mp4, mpeg, asf etc yes. raw? no unless yuv4mpeg or other 'raw format'
[05:32] <Zeranoe> it's vob, and I see it isn't 
[05:35] <Zeranoe> Possible bug in ffplay, when not using -ss, the aspect/display seems skewed, but with ss it seems fine. Is this normal?
[05:36] <Zeranoe> I want to pause ffplay on a frame, right at 2second, I know space does it, but can I pass it through the command?
[05:38] <Zeranoe> Even doing ffplay -i input.vob -ss 00:00:00.01 I get a completely different display than without
[09:07] <plepere> hello all
[09:11] <plepere> is there a way to do a fate on only 10bit bitstreams ?
[09:13] <nevcairiel> not automatically no
[09:14] <nevcairiel> you can figure out which tests are for 10-bit and run them manually
[09:14] <plepere> ok ok. :)
[09:14] <plepere> thanks
[10:31] <plepere> OK, I've sent a new patch on ffmpeg-devel. :)
[10:33] <plepere> BTW, how do I know if my code has been validated ?
[10:36] <ubitux> plepere: ok so now every function is almost identical, you can just create a macro generating them 
[10:36] <ubitux> your code will be pushed when "validated"
[10:36] <plepere> ok
[10:37] <ubitux> basically you can have the parameter of LOOP_INIT/END and the number of loop as parameter
[10:38] <ubitux> also, just squash those changes in your patch instead of iterations
[10:44] <ubitux> plepere: so untested, but something like that: http://pastie.org/pastes/8832107/text
[10:45] <plepere> ubitux : I did a 3rd patch since I've already given the 2 previous patches at the start of the thread
[10:45] <ubitux> (i randomly picked the last 3 entries)
[10:45] <ubitux> plepere: they're not applied, so you can update them
[10:45] <plepere> yes, we use that more or less for prototypes.
[10:46] <ubitux> you could add 16 and 24 as well
[10:46] <ubitux> also, probably something doable with the _8 with another parameter
[10:46] <ubitux> anyway, you get the point
[10:47] <ubitux> you can probably halve the size of the asm by doing so
[10:47] <plepere> so you say that I should do macros only for the functions with repetitions or should I do it even for tricky ones ? (width = 12 when in 10 bit, width = 24 when in 8bit
[10:48] <ubitux> wherever you can avoid the duplication
[10:48] <ubitux> i just took the last one as an example
[10:50] <ubitux> http://pastie.org/pastes/8832213/text is simpler than duplicating 5x the code
[10:54] <plepere> ok thanks
[11:03] <plepere> I've done %rep %1/16  to avoid the second parameter. )
[11:03] <plepere> it' %1/8 when in 10bit though. :)
[11:04] <ubitux> you can probably add parameter for the other variants
[11:05] <plepere> I'm doing a simple pass macro and a repeat macro for the functions
[11:05] <plepere> I'll do a pastie
[11:11] <plepere> ubitux : http://pastie.org/8832615
[11:11] <ubitux> sounds better :)
[11:41] <Case> are the vorbis maintainers still alive and active with FFmpeg?
[11:43] <JEEB> well, there are quite a bit of people hacking on it, IIRC
[11:44] <Case> I see two names in maintainers. Couldn't find any such info for libav though
[11:45] <JEEB> you can just see who last poked the file http://git.videolan.org/?p=ffmpeg.git;a=history;f=libavcodec/vorbisdec.c;h=a2f7dd27f8e7594927c2dac91817607c8543b93c;hb=HEAD
[11:46] <JEEB> mini, ruggles, lu_zero, elenril etc etc
[11:47] <Case> ok, seems there's still work going on. Good to know
[11:49] <JEEB> yeah, I've had bugbears fixed in it, too
[11:50] <Case> I'm just annoyed about foobar2000's situation. This bug would be circumventable by reverting to libvorbis if file isn't recognized at all by ffmpeg: https://trac.ffmpeg.org/ticket/2893
[11:51] <Case> but this issue isn't as simple: https://trac.ffmpeg.org/ticket/3432
[12:00] <wm4> Case: you could try to narrow down the problem by finding out whether it's the demuxer's or the decoder's fault
[12:01] <Case> pretty sure decoder. I believe foobar2000 uses its own demuxers and the problem is present in it
[12:03] <wm4> I'd guess there's no way to revert to another decoder if the decoder fails (unless the application does it explicitly)
[12:03] <wm4> and it doesn't sound like a proper solution anyway
[12:04] <Case> ideal solution would be if ffmpeg's decoder worked reliably for all situations
[12:04] <wm4> yep
[12:36] <JEEB> heh, cehoyos being useful'ish for a change :P
[12:38] <ubitux> JEEB: come on, he's often tracking the regression commit :)
[12:38] <JEEB> if he is, great
[12:39] <JEEB> I've just herped a derp ever since my exchange of e-mails with him
[13:11] <anshul> Hi I was looking at bug #1778, when i try to convert that bugy video subtitle, it gives me error that encoder not found, but what I expected was decoder for that subtitle not found.
[13:38] <cone-430> ffmpeg.git 03Jan Ekström 07master:40feed524854: libutvideoenc: Add support for the new BT.709 FourCCs for YCbCr
[15:04] <ubitux> j-b: does this bonus point system efficient in practice to get bugs solved?
[15:04] <j-b> yep
[15:04] <ubitux> also, can the points be used for trade or something?
[15:04] <ubitux> or that's merely symbolic
[15:07] <j-b> Please read the email
[15:08] <ubitux> my bad, thx
[15:18] <plepere> ubitux, I'm done doing the macros
[15:19] <plepere> should I put out a patch 4 or should I simply squash previous commits and have a new patch 2
[15:19] <ubitux> squash
[15:19] <plepere> ok thanks
[15:19] <ubitux> as long as it's not applied upstream, you update your patchset
[15:19] <ubitux> does it reduce significantly the size of the patch?
[15:21] <plepere> well I'll give back the 2 patches so you can just apply them on the master
[15:23] <plepere> there
[15:26] <plepere> feedback is welcome
[15:27] <ubitux> yeah maybe later for me, that's a large patch
[15:28] <plepere> well it's only cosmetic changes.
[15:30] <nevcairiel> patch size at least changed from like 130kb in the first version to under 60kb now
[15:34] <plepere> yeah. :)
[17:03] <cone-430> ffmpeg.git 03Willy Aubry 07master:65c95de213d9: avformat/HLS: modified EXT-X-MEDIA-SEQUENCE header in order to follow the HLS standard
[17:11] <cone-430> ffmpeg.git 03Michael Niedermayer 07master:ab5d46df1656: avformat/httpauth: add comment about "" and algorithm
[17:13] <ubitux> who wants a hack for icy support in ffplay?
[17:14] <nevcairiel> ffplay doesnt really have much of a GUI, where would it even show it?
[17:14] <ubitux> av_log + win title
[17:14] <ubitux> http://pastie.org/8839837
[17:14] <ubitux> ^ this allows to reproduce the issue wm4 is fixing
[17:14] <ubitux> but now i'm too lazy to review the code, i can just confirm it fixes the problem
[17:14] <nevcairiel> lol
[17:15] <wm4> ubitux: probably better to just dump the packet contents to stdout?
[17:15] <ubitux> how so? :)
[17:15] <wm4> if it's for debugging
[17:15] <ubitux> it's not
[17:15] <ubitux> it's a real feature ;)
[17:15] <wm4> especially considering that the icy packets are pretty wacky
[17:16] <wm4> e.g. if the title contains a "'", most servers don't escape it
[17:16] <nevcairiel> i was wondering about that
[17:16] <ubitux> yeah; am i supposed to look for the ';' or EOS and go back from here to the previous ' ?
[17:16] <wm4> no idea, really
[17:16] <nevcairiel> i postponed my parser work for a bit
[17:16] <ubitux> (assuming there is no ';' in the stream title... heh)
[17:17] <ubitux> but well, it's good enough for now
[17:19] <ubitux> "Stream update: Makoto Fujiwara - Runner" how awesome is that.
[17:20] <wm4> what's the point of adding features to ffplay anyway?
[17:22] <ubitux> testing your patches
[17:24] <ubitux> wm4: are you afraid ffplay is going to be a competitive threat to mpv ? :D
[17:24] <nevcairiel> hahahaha
[17:25] <ubitux> watch out for sdl2 support!
[17:25] <ubitux> and when lavfi will have subtitles support, you'll be doomed
[17:25] <wm4> mpv is too small to be in competition with anyone
[17:27] <nevcairiel> personally i never got the whole idea of competition on open source projects, well certainly not with small ones (ie. 1-2 developers)
[17:27] <nevcairiel> What do i care how many people use my stuff, its not like i get money from them
[17:27] <wm4> I don't even care how many use mine
[17:27] <ubitux> it seems nevcairiel is already admitting his defeat
[17:27] <wm4> if I did, I'd probably start contributing to vlc or so
[17:27] <Plorkyeran> you don't get mad donation moneys?
[17:28] <Plorkyeran> I get almost a tenth of a cent per user per year in donations
[17:28] <Plorkyeran> adds up so fast!
[17:29] <nevcairiel> ubitux: nah, my marketshare on windows isn't all that small, a large share of all non-vlc windows players use some of my stuff now
[17:29] <nevcairiel> and my work has landed me a  new job
[17:29] <nevcairiel> so it was worth something!
[17:30] <ubitux> :)
[17:30] <wm4> ubitux: in general, it would be nice if ffplay would focus on debugging, because trying to make it a serious player will never go anywhere (and also, it's a >3000 lines single file POS)
[17:30] <nevcairiel> occasionally i get a couple hundred in donations from some company that use my stuff, so its not too bad
[17:31] <ubitux> wm4: what are you thinking about?
[17:31] <wm4> vlc can buy their own test hardware with donation money, that's pretty nice
[17:31] <ubitux> wm4: you probably want ffprobe
[17:31] <wm4> ubitux: ffprobe (or ffmpeg.c) isn't always sufficient
[17:33] <ubitux> so far i'm pretty satisfied with ffplay
[17:35] <wm4> I'm just saying that in this case, attempting to parse the icy title and setting it as window caption is not terrible useful for testing icy
[17:35] <ubitux> btw, i wonder if sdl2 support in ffplay will somehow bring opengl support
[17:35] <wm4> you could just hook up libavdevice
[17:35] <wm4> doesn't that have opengl support
[17:36] <wm4> but sdl2 still has yuv support, if you mean that
[17:36] <ubitux> wm4: the rational is that it's good enough to trigger the bug and it's also a small useful feature (we already update the win_title with the "title" metadata typically)
[17:37] <ubitux> adding more debug is actually easy if needed
[20:46] <kurosu_> BBB / plepere: maybe pmulhrw is not needed here, as surprisingly there's no roun-before-final-shift here
[21:13] <wm4> ubitux: ping about my microdvd and icy patches... I think you said ok, but didn't write anything on the ML
[21:14] <ubitux> yes sorry, will apply tonight the microdvd if i find the time
[21:14] <ubitux> need to check just a few things
[21:14] <cone-430> ffmpeg.git 03Michael Niedermayer 07master:b79dfd16cbf0: ffmpeg: dont try to run the subtitle encoder on non subtitle streams
[22:30] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:f87ce262f68a: build: The MPEG-4 video parser depends on h263dsp
[22:30] <cone-430> ffmpeg.git 03Reinhard Tartler 07release/2.2:4be1b68d52c7: Prepare for 10_beta2 Release
[22:30] <cone-430> ffmpeg.git 03Luca Barbato 07release/2.2:9b6ccf0f243c: hevc: Always consider VLC NALU type mismatch fatal
[22:30] <cone-430> ffmpeg.git 03Vittorio Giovara 07release/2.2:63169474b392: h264: Lower bound check for slice offsets
[22:30] <cone-430> ffmpeg.git 03Luca Barbato 07release/2.2:de187e3e9ec4: h264: Fix a typo from the previous commit
[22:30] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:8ad2f45964ac: Merge commit 'de187e3e9ec4803575deb1c293ccad84d2a88da8' into release/2.2
[22:37] <cone-430> ffmpeg.git 03Anton Khirnov 07release/2.2:1d1df82093fd: pthread_frame: flush all threads on flush, not just the first one
[22:37] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:c4149c4d5406: Merge commit '1d1df82093fdacb2cbc443c70c80f8f801002d28' into release/2.2
[22:50] <cone-430> ffmpeg.git 03Martin Storsjö 07release/2.2:db6b2ca0b3af: qt-faststart: Increase the copy buffer size to 64 KB
[22:50] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:c9f015f1c688: qt-faststart: Simplify code by using a MIN() macro
[22:50] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:92edc13d6918: qt-faststart: Check fseeko() return codes
[22:50] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:298d66c8de1d: qt-faststart: Fix the signedness of variables keeping the ftello return values
[22:50] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:b3f106cb1f00: qt-faststart: Check the ftello() return codes
[22:50] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:7754d4838178: qt-faststart: Check offset_count before reading from the moov_atom buffer
[22:50] <cone-430> ffmpeg.git 03Martin Storsjö 07release/2.2:9841617b7f86: qt-faststart: Avoid unintentionally sign extending BE_32
[22:50] <cone-430> ffmpeg.git 03Lou Logan 07release/2.2:a6a2d8eb8f12: qt-faststart: Add a note about the -movflags +faststart feature
[22:50] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:7ff4cd2acc8c: Merge commit 'a6a2d8eb8f125a2edb512a7a47df33dbd70d6b35' into release/2.2
[22:59] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:35694706938e: doxygen: Add a number of missing function parameter descriptions
[22:59] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:1779cd7695ab: doxygen: Replace @parblock syntax with manual linebreaks
[22:59] <cone-430> ffmpeg.git 03Martin Storsjö 07release/2.2:cd6281abef6e: arm: Mark the stack as non-executable
[22:59] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:33e1bca6517d: gitignore: Add all examples below doc/examples
[22:59] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:39dc4a6bb34b: x86: dca: Add missing multiple inclusion guards
[22:59] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:b967c10029bd: Merge commit '39dc4a6bb34baf833ce1e5eabad7d0dbf933237d' into release/2.2
[23:14] <cone-430> ffmpeg.git 03Luca Barbato 07release/2.2:affc7687d33a: doc: Sort the muxer documentation
[23:14] <cone-430> ffmpeg.git 03Luca Barbato 07release/2.2:1853d8bb7a50: doc: Name the MOV muxer as it should be called
[23:14] <cone-430> ffmpeg.git 03Luca Barbato 07release/2.2:646c564de545: hevc: Mention the missing SPS in the error message
[23:14] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:9cc22be03298: svq3: Adjust #endif comment
[23:14] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:4015829accc2: bit_depth_template: Use file name as multiple inclusion guard
[23:14] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:eb2244ece9cb: Merge commit '4015829accc2382393d42d62654eb96d896d1326' into release/2.2
[23:27] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:bb4820727f4b: x86: dsputil: Use correct file name as multiple inclusion guard
[23:27] <cone-430> ffmpeg.git 03Janne Grunau 07release/2.2:697be8173bc7: h264: avoid undefined behavior in chroma motion compensation
[23:27] <cone-430> ffmpeg.git 03Christophe Gisquet 07release/2.2:2897481f64c8: dca: replace some memcpy by AV_COPY128
[23:27] <cone-430> ffmpeg.git 03Luca Barbato 07release/2.2:bc2c9a479aee: avconv: Do not divide by zero
[23:27] <cone-430> ffmpeg.git 03Anton Khirnov 07release/2.2:7933039ade01: af_resample: preserve frame properties
[23:27] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:222e7549a74d: Merge commit '7933039ade01b39638ec3d9e638b6ae06ee84984' into release/2.2
[23:35] <cone-430> ffmpeg.git 03Anton Khirnov 07release/2.2:58556826a80e: af_volume: preserve frame properties
[23:35] <cone-430> ffmpeg.git 03Vittorio Giovara 07release/2.2:3c72204ae0eb: doc: name correct header
[23:35] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:b76871d87004: parser: Remove commented-out cruft
[23:35] <cone-430> ffmpeg.git 03Luca Barbato 07release/2.2:437179e9c89e: parser: K&R formatting cosmetics
[23:35] <cone-430> ffmpeg.git 03Diego Biurrun 07release/2.2:596d3e20ae69: parser: cosmetics: Drop some unnecessary parentheses
[23:36] <cone-430> ffmpeg.git 03Michael Niedermayer 07release/2.2:82031e41f820: Merge commit '596d3e20ae69a278d562eea08f5e8c0ae5a5bfc4' into release/2.2
[23:37] <cone-430> ffmpeg.git 03wm4 07master:0786598f51ce: avformat/microdvd: skip BOM properly
[23:37] <cone-430> ffmpeg.git 03wm4 07master:b416517cce6b: avformat/microdvd: export the declared frame rate
[23:37] <wm4> ubitux: <3
[23:37] <ubitux> e
[23:40] <ubitux> wm4: i microbumped lavf so you can check for that version in your code
[23:40] <wm4> thanks, though I'll just check whether the subfps option exists
[23:40] <ubitux> j
[23:40] <ubitux> k
[23:41] <wm4> what about the icy one? or does that need further "review"
[23:42] <ubitux> i don't know how the icy actually works and my brain is melting tonight, that will have to wait i guess
[23:42] <ubitux> who reviewed that last time?
[23:43] <wm4> > +    if(!strcmp("translate.google.com", hoststr) && !s->user_agent)
[23:43] <wm4> > +        s->user_agent = av_strdup("Mozilla/5.0 (X11; Ubuntu; Linux i686; rv:15.0) Gecko/20100101 Firefox/15.0.1");
[23:43] <wm4> wow
[23:44] <wm4> (that patch was rejected, I guess)
[23:45] <ubitux> yes
[23:45] <ubitux> somehow
[23:46] <ubitux> i suggested to use a "browser" mode all the time, but it wasn't a good idea in practice
[23:46] <wm4> hm saste reviewed the initial icy patch
[23:46] <ubitux> (since some web radio act as a web page summary if detected as a browser, and a ogg/mp3 stream if not)
[23:47] <wm4> yeah that's annoying
[23:47] <wm4> no user agent will work for everything
[23:49] <ubitux> oh the vorbis this has been pushed?
[23:49] <ubitux> i missed that
[00:00] --- Tue Mar  4 2014