[Ffmpeg-devel-irc] ffmpeg.log.20130608

[burek]

[00:36] <nellynel90> where does ffmpeg log stdout or stderr?
[00:36] <nellynel90> i have a small python scrypt im trying to use to filter the output of ffmpeg but it always passes everything
[00:46] <loopymofo> well, you know, I'm running it through python myself with subprocess and Popen and can get the output that way
[00:46] <Guest35974> STDOUT: ffmpeg -i infile outfile
[00:47] <Guest35974> STDOUT should be in "outfile"
[00:50] <Guest35974> You'll need probably to adjust --loglevel option to something other than quiet
[00:50] <Guest35974> from the man page
[00:51] <Guest35974>  "By default the program logs to stderr"
[00:52] <nellynel90> loopymofo: im no expert at python and ive had no luck
[00:52] <nellynel90> im using popen too but im not sure what im doing wrong
[00:53] <loopymofo> if you paste the code up I can take a look
[00:55] <nellynel90> http://pastebin.com/U10Lbr9B
[00:57] <nellynel90> in the loop at the bottom, it should actually read if 'Input' not in line: start_print = True
[00:57] <nellynel90> and dont mind the hard coded ffmpeg params
[00:59] <nellynel90> actually theres several mistakes heres the good one http://pastebin.com/kwEYe3wX
[01:03] <loopymofo> I see, so the idea is to only print lines once you've found an 'Input' in the output? and that's not working correctly?
[01:04] <loopymofo> I think you should wait until the process is finished, and then parse the output, that's what you want right?
[01:06] <nellynel90> yeah prints from the top
[01:07] <loopymofo> http://pastebin.com/NUHbPmma
[01:07] <loopymofo> I wonder if that will work
[01:08] <nellynel90> nah still prints from the top
[01:09] <nellynel90> on the shell, if i do ffmpeg foo bar foo bar 2>/dev/null i get no logging at all
[01:09] <loopymofo> hmm
[01:10] <nellynel90> but if i do Popen([foo, bar, stderr=subprocess.PIPE]) nothing ever comes through the pipe
[01:10] <nellynel90> that last bracket should come after bar. sorry
[01:11] <loopymofo> yeah I see what you're going after, trying to get it working myself :P
[01:12] <ubitux> nellynel90: http://git.videolan.org/?p=ffmpeg.git;a=blob;f=tools/normalize.py;hb=HEAD
[01:12] <loopymofo> hmm, nellynel90: did you try that out with stderr=subprocess.PIPE and my communicate code?
[01:12] <loopymofo> only print err instead
[01:13] <loopymofo> or check err vs out
[01:14] <loopymofo> nellynel90: http://pastebin.com/ru5ngu3j
[01:17] <nellynel90> im not feeding though stdin so i can remove that part right?
[01:17] <loopymofo> yeah
[01:18] <nellynel90> no output
[01:18] <loopymofo> are you sure the start_print = True is executing?
[01:19] <loopymofo> that code works for me for just printing the output after process ends
[01:19] <nellynel90> hmm
[01:19] <nellynel90> thats what communicate() does?
[01:20] <loopymofo> yeah, it waits for it to terminate and then reads it's out/stderr
[01:21] <nellynel90> if it works for you it should work for me too then. i need to go try this on a small clip then coz its taking forever
[01:21] <loopymofo> you could try printint the output regardless of if start_print is true or not, to see if you're getting it
[01:21] <loopymofo> printing*
[01:23] <nellynel90> but i still have to wait for communicate()
[01:25] <loopymofo> that's not what you want?
[01:27] <nellynel90> no im just saying, if i understand correctly, that if my ffmpeg encode is hella long, ill have to wait for communicate() to finish so the for loop can go through err. right?
[01:28] <loopymofo> yeah
[01:28] <loopymofo> you could go somewhat real time though, with what you had before
[01:28] <loopymofo> only call ffmpeg.wait() after your for loop
[01:29] <loopymofo> and you might also want/need to sys.stdout.flush(), to flush it so you can read it
[01:29] <loopymofo> and still set stderr to subprocess.PIPE
[01:30] <nellynel90> it didnt work for me
[01:30] <nellynel90> your method that is
[01:30] <nellynel90> neither out or err could match 'Input'
[01:31] <nellynel90> im interested in what you said though. how would i use flush()
[01:32] <loopymofo> well, did you try it just with wait() after your for loop? that might work but i'm unsure
[01:34] <loopymofo> i've always used subprocess in a blocking manner :P
[01:35] <nellynel90> nah nothing prints
[01:37] <nellynel90> im not sure how wait() would help though. going back to what i had before, if ffmpeg.stdout stores ffmpegs out, why cant it find the string
[01:37] <nellynel90> its definitely there. when i pipe stderr, nothing prints, when i pipe stdout, everything prints
[01:42] <loopymofo> parsing it as it come out in real time you mean? perhaps it's not contained in that buffer that's being read at that moment
[01:42] <loopymofo> It should've worked with the communicate(), beyond that I'm not really sure
[01:43] <nellynel90> i see. well ima play with it come monday. thanks a lot btw
[01:43] <loopymofo> no problem, sorry I couldn't be more help and get it actually working :P
[01:43] <nellynel90> if not for the internet and people such as yourself, i wouldnt get anywhere =D
[01:54] <loopymofo> I'm still looking for a good solution for constantly streaming a file through one instance of ffmpeg, thats being generated by another instance of ffmpeg
[01:54] <loopymofo> so that I can change what file the one instance is outputting, basically setting up a playlist structure for streaming stuff
[01:54] <loopymofo> dunno if it's even possible, no luck on my part
[02:00] <DeXa> hi guys, any official AVCHD support in FFmpeg?
[02:01] <DeXa> i'm converting like a gazilion gb and some files fails with re-encoding...
[02:04] <llogan> fails how?
[02:09] <DeXa> moov atom not found
[02:10] <llogan> so the issue is with decoding the original input, or decoding your re-encoded output?
[02:10] <DeXa> decoding my re-encoded output
[02:11] <DeXa> k, sec
[02:12] <DeXa> in the meantime, all corrupted files has the same file size 4.29GB
[02:12] <DeXa> if it rings any bells
[02:13] <ezekiel> hmm, weird - sounds like a filesystem size limit - NTFS? or FAT?
[02:14] <DeXa> It's a journaled os x partition
[02:14] <ezekiel> oh, the source files are bigger than that, and presumably on the same filesystem - so probably not relevant
[02:14] <DeXa> checking source size...
[02:15] <DeXa> hah. original has the same 4.29GB file size
[02:42] <DeXa> ok. I got this: "av_interleaved_write_frame(): File too large"
[02:42] <DeXa> (now pasting to pastebin...)
[02:44] <DeXa> http://pastebin.com/SLpdBDBF
[02:48] <DeXa> llogan: any idea?
[02:50] <llogan> no, sorry.
[02:51] <DeXa> thanks
[02:52] <llogan> why fifo?
[02:54] <DeXa> it's part of a script which analyse and add filters
[02:55] <DeXa> null will be better?
[02:55] <DeXa> (checking without...)
[03:08] <DeXa> same result without fifo
[03:19] <DeXa> omg, stupid me, the drive was formatted at fat32
[03:20] <DeXa> can't believe it...
[04:30] <skifreak> Having a problem with ffmpeg, seems it doesn't like remuxing some of my files which have PCM audio. Input file: http://dpaste.com/1218637/ Remux command: http://dpaste.com/1218638/ Resultant file: http://dpaste.com/1218639/
[04:30] <skifreak> Note: on master build of ffmpeg
[04:36] <skifreak> Just downgraded temporarily to v1.2.1, problem exists there as well
[05:48] <highgod> Hi, I want to ask a quesiton, I want to add a hardware decoder to my app, but I want to use ffmpeg to analyse the h264 bit stream to set some parameters, and as we konw, the context of h264 does not as a API to provide, is there any solution to implement that? thanks
[05:55] <llogan> an analog disc for video that used a needle and groove system like an audio record: http://en.wikipedia.org/wiki/Capacitance_Electronic_Disc
[06:24] <relaxed> skifreak: I don't see an error.
[06:24] <relaxed> try again with -ss and -t after the input
[06:31] <skifreak> relaxed: the output file's audio is "unknown"
[06:34] <skifreak> ffmpeg command ran: http://dpaste.com/1218889/ output file ("unkown" audio): http://dpaste.com/1218890/
[06:34] <skifreak> thanks for the help, relaxed
[06:35] <skifreak> (I didn't map all of the streams with -map 0 since there's a subtitle problem, outstanding bug #2622)
[06:43] <relaxed> It doesn't say unknown in the output.
[06:43] <relaxed> Did you try playing it with ffplay? mplayer?
[06:45] <skifreak> relaxed: tried both
[06:46] <skifreak> same thing happens on the sequel file as well, KB2
[06:47] <skifreak> line 22 in http://dpaste.com/1218890/:   Stream #0:1[0x101]: Unknown: none ([6][0][0][0] / 0x0006)
[06:48] <relaxed> Are you using a recent version from git?
[06:49] <relaxed> Can you put a small sample of your source up somewhere?
[06:49] <skifreak> of ffmpeg? tried v1.2.1 and master
[06:51] <skifreak> sure, where's a good place to post it?
[06:51] <skifreak> I can post on the ftp easy enough, but somewhere more public might be better
[06:53] <relaxed> Your input mpegts
[06:54] <relaxed> Dropbox or whatever, just so that we can reproduce.
[07:26] <james4k> hi, is it appropriate to ask libavcodec questions here?
[07:28] <james4k> ah yes, just read the #ffmpeg-devel topic :)
[07:30] <james4k> im trying to reuse a buffer to avoid copying when decoding h264 frames, but it seems to give some artifacts to the video until keyframes
[07:30] <james4k> so i tried passing my own buffers through get_buffer2: https://gist.github.com/james4k/05e83ba09a24ae76ee4a
[07:32] <james4k> ... i explained this in the wrong order. I tried passing my own buffers through get_buffer2, but i get artifacts. when i use avcodec_default_get_buffer2 and copy to those same buffers, there are no artifacts
[07:42] <skifreak> relaxed: currently uploading everything needed to reproduce on the ftp under incoming/PCM_audio_remux_fail.tgz Sorry if it's a little big, I had to make an input sample with dd starting at the beginning of the original input file until sound was exhibited, or else the results were not reproduceable
[07:44] <skifreak> ah, it just completed, thanks for the help! Let me know if you need anything more and if I should create a ticket
[07:46] <relaxed> skifreak: Ok. I'm at work so stick around and I'll look at it when I can.
[07:49] <skifreak> relaxed: ha, no worries, take your time. I'm wondering if there's something wrong with some of these files I'm running across possibly, they're 1st hand straight passthrough from blurays, but I'm wondering if #2622 I filed before is connected to this
[07:49] <skifreak> really appreciate the help relaxed!
[09:13] <relaxed> skifreak: where is it?
[09:33] <burek> <DeXa> http://pastebin.com/SLpdBDBF
[09:33] <burek> omg :)
[09:33] <burek> 20 minutes of mpegvideo
[09:33] <burek> and it got over 4 GB ? :)
[09:34] <braincracker> try h264
[09:34] <burek> i was just reading the logs in this channel and was surprised by that :) that's all :)
[09:35] <braincracker> mpeg is large
[09:36] <braincracker> they used that on svcd 10 years ago
[09:36] <braincracker> really crap quality fits on a cd
[09:38] <burek> yeah
[09:41] <burek> especially comparing to vp9 now http://www.youtube.com/watch?v=K6JshvblIcM#t=1m25s
[09:48] <skifreak> relaxed: on the ftp, incoming/PCM_audio_remux_fail.tgz
[09:58] <skifreak> going to coma for 7, thanks again relaxed!
[10:00] <relaxed> skifreak: spend two seconds and give me the direct url.
[10:01] <skifreak> relaxed: I assume it should be upload.ffmpeg.org/ffmpeg/incoming/PCM_audio_remux_fail.tgz
[10:02] <skifreak> but as anonymous on the ftp or through a simple web browser, I can't access it
[10:02] <skifreak> if that doesn't work, I'll just create a dropbox account
[10:02] <skifreak> whatever is the best for you
[10:03] <relaxed> skifreak: it's write only. Use my referral link to create an account please. http://db.tt/cDWPLzP
[10:03] <relaxed> For dropbox ^^
[10:06] <skifreak> relaxed: ok, working on it
[10:26] <skifreak> relaxed: https://www.dropbox.com/s/r6w8irxzo36zeei/PCM_audio_remux_fail.tgz
[10:26] <skifreak> Let me know if that link works for you
[10:27] <skifreak> again, sorry for the large file size, see note above ^ (almost 3 hours ago)
[10:45] <relaxed> skifreak: Ok. I have it now, thanks.
[12:36] <Aaronds> Hi all, I'm trying to convert video for using with HTML5 (mp4 and ogv files). My mp4 conversions seem to work OK, however when I convert to ogv, some files end up going from 3mb to almost 70mb. It doesn't happen all the time. The original file format is WMV. Conversion commands: zEYkPmQP
[12:37] <Aaronds> *http://pastebin.com/zEYkPmQP
[12:37] <Aaronds> Does anyone know what could cause this?
[12:41] <Mavrik> Aaronds, it much more helpful if you also pastebin full ffmpeg output
[12:41] <Mavrik> Aaronds, and also, did you consider webm which has noticably better quality and encoders than theora? Theora encoder is kinda bad
[12:42] <Aaronds> Oh, but does mp4 & webm have the same browser coverage?
[12:42] <Aaronds> I can get the full output but will have to repost later, takes a while to do.
[12:42] <Mavrik> well, can't help you more without a full probe output
[12:43] <Mavrik> IIRC theora and webm is supported in about the same browsers
[12:45] <Aaronds> Yeah looks like it... Nice I'll have to play around with webm later then.
[12:45] <Aaronds> I'll be back with ful output later if I continue having problems.
[12:45] <Aaronds> Thanks
[12:55] <diverdude> can anybody show me how i can use ffmpeg to stream a std::queue< std::vector<uint8_t> > to webclients using ffmpeg?
[12:56] <Mavrik> uh
[12:56] <diverdude> Mavrik, ?
[13:44] <khali> recent builds of ffmpeg flood the output with messages "Non-monotonous DTS in output stream 0:1"
[13:45] <khali> older build (1.0.6) does not print it and there doesn't seem to be any issue with the output file
[13:47] <ubitux> you tested with git/HEAD ?
[13:49] <khali> ubitux: yes, it did not help
[13:50] <khali> ubitux: might be that the DVB-T stream is indeed less than perfect, but that's no reason to flood stdout IMHO
[13:50] <ubitux> can you open a ticket with a sample, and mark it as a regression?
[13:50] <khali> ubitux: sure
[13:50] <ubitux> khali: mmh wait
[13:51] <ubitux> the output file is still valid?
[13:53] <khali> not sure
[13:53] <khali> hard to tell as this is a speechless program
[13:53] <khali> I'll redo it with ffmpeg 1.0.6 and see if the sound sync is the same
[14:08] <khali> ubitux: no, the output file is not valid, audio stream is shifted
[14:09] <khali> ubitux: but it happens that this recording has some corrupted data... old ffmpeg manages to recover from it, so that would still be a regression
[14:09] <khali> ubitux: but I'll try reproducing with a clean input file before I report
[14:12] <diverdude> can anybody show me how i can use ffmpeg to stream a std::queue< std::vector<uint8_t> > to webclients?  I have a camera device where i gather grayscale images(8bpp - or uint8_t) in a queue (a buffer of frames). i want to stream it over the network to standard desktop PC's. It should be accessible via a html browser like firefox/chrome using the html5 <video> tag.  It should be encoded using webm.
[14:56] <diverdude> any1?
[16:15] <skifreak> relaxed: let me know if you want me to open a ticket
[16:17] <jdolan> i have an .flv file recorded from a user's webcam / mic. the user mutes their camera in the middle of the recording, so video frames stop but audio keeps going. then the camera is turned back on and video frames resume.
[16:17] <jdolan> mplayer can play this .flv back just fine, ffplay breaks pretty badly.
[16:18] <jdolan> the video remains frozen in ffplay, and the file actually never stops playback :|
[18:11] <khali> ubitux: so far I am unable to reproduce when the input file has no error, so it might be an issue with how ffmpeg recovers from imperfect recordings
[21:18] <relaxed> skifreak: Can you play that audio stream? When muxed with mkvmerge it disappears.
[21:24] <skifreak> relaxed: I can't play the audio stream in the output file, out_bad_audio.m2ts with mplayer or ffplay
[21:24] <skifreak> haven't tried mkvmerge, seems it's not in portage
[21:25] <sacarasc> It's part of mkvtoolnix.
[21:25] <skifreak> ok, grabbing that package
[21:37] <relaxed> skifreak: I meant the input audio stream
[21:40] <skifreak> I can play the audio stream in the input file, input_video_with_PCM_audio.m2ts, with mplayer
[21:40] <relaxed> skifreak: Yeah, tsmuxer is able to remux it so you should file a bug report.
[21:41] <skifreak> ok, mkvtoolnix v5.0.1-r1 doesn't see the PCM audio when I run mkvmerge -I input_video_with_PCM_audio.m2ts
[21:42] <skifreak> I'll file I bug report then, thanks for the help relaxed!
[21:42] <relaxed> See if tsmuxer works on the who input and plays.
[21:42] <relaxed> whole*
[21:43] <skifreak> ok, will do
[21:56] <skifreak> relaxed, thanks, tsmuxer seems to be working nicely, I wonder if there's a nice way for me to work it in with ffmpeg with my x264 encoding (have tsmuxer take care of the subtitles and audio for me as well as the muxer)
[22:02] <skifreak> tsmuxer seems to  be nicer about muxing, in general, than ffmpeg. When I try to mux two files together (one 1080p x264 video the other input contain PCM audio, the new output file containing both streams) ffmpeg eats 18GB of my memory
[22:03] <sisco> hi all
[22:04] <sisco> i would like to know if there anyway to get stream of more then one rtmp from a server who have duplicate stream security
[22:04] <sisco> any advise will be greatfull
[22:04] <sisco> anyone can help ?
[22:12] <braincracker> "smack you" can mean "suck you" ?
[22:13] <relaxed> skifreak: It is much better at muxing mpegts than ffmpeg, but last I checked it's not easy to script with.
[22:13] <phr3ak> i tried to convert/compress a mpeg2video file to other format but i got bad picture quality. the output was mpeg4. i ran this: ffmpeg -i in.avi out.avi could you tell me please, how could I get better video quality?
[22:13] <relaxed> I use it to create avchd dics from my bluray player.
[22:13] <relaxed> for*
[22:14] <relaxed> phr3ak: do you need mpeg4/avi for a particular reason?
[22:15] <sisco> hiiiiiiiii
[22:15] <sisco> i would like to know if there anyway to get stream of more then one rtmp from a server who have duplicate stream security
[22:15] <skifreak> relaxed: yea...I've been trying to find a decent man page, but tsmuxer wants a meta file which seems much easier to generate in the GUI
[22:16] <phr3ak> relaxed: no
[22:16] <sisco> guys can u help me too ?
[22:16] <phr3ak> relaxed: i'd like to get a smaller file only
[22:16] <relaxed> phr3ak: https://ffmpeg.org/trac/ffmpeg/wiki/x264EncodingGuide
[22:17] <phr3ak> thanks
[22:17] <sisco> 0.o
[22:17] <sisco> need help :(
[22:18] <sisco> is there anyway to broke this security  http://www.wowza.com/forums/content.php?204-How-to-block-a-duplicate-publish-stream&page=2#comments
[22:18] <relaxed> phr3ak: for best comprression use "-preset veryslow" with libx264.
[22:18] <relaxed> skifreak: you can create the meta file on the fly with a "here doc"
[22:20] <relaxed> sisco: that's offtopic here
[22:21] <ezekiel> also, lrn 2 patience
[22:21] <ezekiel> :)
[22:21] <sisco> relaxed what u mean
[22:22] <sisco> i would like to broke this security
[22:22] <sisco> relaxed: a module that wowza set up to protect duplicate streaming
[22:23] <ezekiel> sisco: he means that ffmpeg is discussed here, not breaking some site security
[22:23] <skifreak> relaxed: ok, sounds great. Thanks again for all of the help, really appreciate it!
[22:23] <ezekiel> so your question is offtipic
[22:23] <sisco> well im using ffmpeg
[22:43] <braincracker> yo lmf4o
[23:22] <phr3ak> could I run batch convert to *.avi without any script or need I write one?
[23:31] <relaxed> phr3ak: for i in *.avi; do ffmpeg ...something_here... "${i%.*}".mkv; done
[23:33] <relaxed> phr3ak: sorry it's -->  for i in *.avi; do ffmpeg -i "$i" ...something_here... "${i%.*}".mkv; done
[23:44] <phr3ak> thank you
[00:00] --- Sun Jun  9 2013