[Ffmpeg-devel-irc] ffmpeg.log.20170206
burek
burek021 at gmail.com
Tue Feb 7 03:05:01 EET 2017
[00:56:21 CET] <xtina> wow.. building ffmpeg is really a slow process eh
[00:57:11 CET] <matthiaskrgr> do you use multiple jobs? (make -j2 / make -j 4 ) ?
[00:59:54 CET] <furq> it is on an rpi
[00:59:58 CET] <matthiaskrgr> :D
[01:00:01 CET] <furq> like i said, it's way faster if you cross compile
[01:00:20 CET] <matthiaskrgr> rpi is single core, right?
[01:00:26 CET] <furq> the pi zero is
[01:00:34 CET] <furq> pi 1 is dual core, pi 2/3 are quad core
[01:01:25 CET] <furq> it's probably also a good idea to build it on a network drive or in a tmpfs
[01:01:34 CET] <furq> i can't imagine how much slower it'll be if you're writing all the temp files to sd
[01:03:41 CET] <furq> one of these commentators just said "audibleizing"
[01:03:45 CET] <furq> that's the worst word i've ever heard
[01:04:10 CET] <furq> er. wrong channel
[01:10:10 CET] <xtina> i have a pi zero
[01:10:24 CET] <xtina> i did make -j2
[01:10:27 CET] <xtina> make install
[01:18:50 CET] <xtina> i really hope i built it correctly for arecord/raspivid
[01:19:01 CET] <xtina> since it would take so long to rebuild :O
[01:19:18 CET] <JEEB> that's why you use an ARMv6 cross-compiler with a sysroot
[01:19:25 CET] <JEEB> so you build and copy the stuff over
[01:22:51 CET] <dystopia> hello
[01:28:21 CET] <dystopia> can someone help me with my command
[01:28:21 CET] <dystopia> http://pastebin.com/0rSbuF5U
[01:28:37 CET] <dystopia> it works if i dont set -re and -ss timestamp
[01:28:47 CET] <dystopia> but i want to mark the time to begin the encode
[01:34:36 CET] <dystopia> got it
[01:34:50 CET] <dystopia> mm:ss:ns is not supported
[01:34:56 CET] <dystopia> it's mm:ss
[03:36:46 CET] <xtina> hey guys. i just compiled ffmpeg with alsa support. now i'm trying to understand how to turn my arecord command into a ffmpeg audio command
[03:37:05 CET] <xtina> my arecord command: arecord -Dmic_sv -c2 -r48000 -fS32_LE -twav -d10 -Vstereo test.wav
[03:37:18 CET] <xtina> how do i pass arecord params like -D (device name) into ffmpeg?
[03:38:26 CET] <furq> https://trac.ffmpeg.org/wiki/Capture/ALSA
[03:40:23 CET] <xtina> furq: from the manual it seems like -i is ffmpeg's 'input_device' equivalent for arecord -D. but the following command doesn't work
[03:40:25 CET] <xtina> ~/special/ffmpeg/ffmpeg -f alsa -ac 1 -ar 44100 -i mic_sv out.wav
[03:40:30 CET] <xtina> 'mic_sv: Input/output error'
[03:40:56 CET] <xtina> while this does work: arecord -Dmic_sv -c2 -r48000 -fS32_LE -twav -d10 -Vstereo test.wav [
[03:41:41 CET] <xtina> mic_sv is a device here:
[03:41:42 CET] <xtina> pi at raspberrypi:~/special/ffmpeg $ arecord -L null Discard all samples (playback) or generate zero samples (capture) mic_mm mic_rt mic_sv sysdefault:CARD=memsmic mems-mic, Default Audio Device dmix:CARD=memsmic,DEV=0 mems-mic, Direct sample mixing device dsnoop:CARD=memsmic,DEV=0 mems-mic, Direct sample snooping device hw:CARD=memsmic,DEV=0 mems-mic, Direct hardware device without any conversions p
[04:03:22 CET] <debianuser> xtina: But `ffmpeg -f alsa -ac 1 -ar 44100 -i mic_sv` sets 1 channel 44100 rate, while `arecord -Dmic_sv -c2 -r48000 -fS32_LE` uses 2 channels 48000 rate. Have you tried `ffmpeg -f alsa -ac 2 -ar 48000 -i mic_sv` ?
[04:04:49 CET] <xtina> you're right, i messed that up..
[04:04:55 CET] <xtina> i just got the following to record audio :) ~/special/ffmpeg/ffmpeg -f alsa -acodec pcm_s32le -i mic_sv -ar 48000 test.wav
[04:10:19 CET] <debianuser> Ah, it's good you have it sorted out. :) I wonder if `arecord -fcd -c1 -Dmic_sv test.wav` works for you... If it doesn't you may have a driver bug, worth reporting to your distribution.
[04:12:41 CET] <xtina> debianuser: that does work for me
[04:12:59 CET] <xtina> i'm now wondering the best way to stream both audio and video with ffmpeg
[04:13:21 CET] <xtina> which way is recommended: piping raspivid into my (ffmpeg w/ also audio), or (ffmpeg w/ audio and h264)?
[04:13:35 CET] <xtina> i'm concerned about getting good audio+video stream sync
[04:32:05 CET] <xtina> i decided to try piping raspivid into (ffmeg w/ audio) since i read this is more cpu efficient. my command throws no errors but i get lots of 'alsa buffer xrun' messages. report: http://pastebin.com/xeRA4VSY
[04:32:37 CET] <furq> it isn't more cpu efficient
[04:32:59 CET] <furq> -f v4l2 and -c:v h264_omx do more or less the same thing as raspivid
[04:35:21 CET] <xtina> furq: gotcha, i will try that instead then. do you know what i should do about alsa buffer xrun issues?
[04:38:48 CET] <debianuser> xtina: alsa buffer overrun/underrun is an indicator that application (ffmpeg) doesn't have enough CPU, or slowed down by something else, so it doesn't read soundcard buffer in time, it reads slower than soundcard puts the data into it, and the buffer overflows. It's not a fatal error, any alsa application should automatically recover from xrun. But you may hear slight audio click when xrun happens.
[04:44:52 CET] <furq> src/zimg/api/zimg.cpp:34:26: error: cannot compile this non-trivial TLS destruction yet
[04:44:55 CET] <furq> well this is a new one
[04:55:50 CET] <xtina> for some reason i'm seeing 'Unknown decoder 'h264_omx''
[04:55:58 CET] <xtina> when i attempt -c:v h264_omx
[04:56:03 CET] <furq> it's an output option, not an input option
[04:56:05 CET] <xtina> even though i included -enable-omx-rpi in my compile
[04:56:10 CET] <xtina> oh
[04:59:26 CET] <xtina> my latest attempt (errors everywhere): ~/special/ffmpeg/ffmpeg -f alsa -ar 48000 -acodec pcm_s32le -i mic_sv -f v4l2 -thread_queue_size 512 -framerate 20 -probesize 100 -i /dev/video0 -c:v h264_omx -acodec aac -f flv rtmp://209.85.230.23/live2/xbjs-a3gp-0yvk-ftdg
[05:00:26 CET] <furq> is there any reason you're using pcm_s32le
[05:00:38 CET] <furq> that's probably not helping if you're getting xruns
[05:02:17 CET] <xtina> i'm using a mems mic, my .asoundrc currently contains this:
[05:02:18 CET] <xtina> pcm.mic_hw{ type hw card memsmic channels 2 format S32_LE }
[05:02:36 CET] <xtina> (i'm not totally clear on why but this was suggested to me)
[05:03:12 CET] <furq> try changing that to S16_LE and use pcm_s16le
[05:03:17 CET] <furq> that'll be half the data rate
[05:12:09 CET] <xtina> furq: for some reason, changing the 32 to a 16 in asoundrc and the ffmpeg cmd throws an error
[05:12:17 CET] <xtina> [alsa @ 0x33a7760] cannot initialize hardware parameter structure (Invalid argument) mic_sv: Input/output error
[05:13:06 CET] <xtina> pcm.mic_hw{ type hw card memsmic channels 2 format S16_LE }
[05:13:23 CET] <xtina> ~/special/ffmpeg/ffmpeg -report -f alsa -ar 48000 -acodec pcm_s16le -i mic_sv -f v4l2 -thread_queue_size 512 -framerate 20 -probesize 100 -i /dev/video0 -c:v h264_omx -acodec aac -f flv rtmp://209.85.230.23/live2/xbjs-a3gp-0yvk-ftdg
[05:16:11 CET] <xtina> this doesn't happen if i use 32. this is the pastebin: http://pastebin.com/zXyzZFyT
[05:28:40 CET] <xtina> is it possible pcm_s16le is not supported by my mems mic
[05:59:21 CET] <xtina> i'm getting all kinds of errors with my ffmpeg command, and 0kbps/0fps. any guesses? http://pastebin.com/P6z0V7vy
[06:15:45 CET] <xtina> any ideas what's wrong here? http://pastebin.com/P6z0V7vy
[06:38:44 CET] <xtina> i'm running this cmd but getting 0kbps, anyone see what's wrong? ~/special/ffmpeg/ffmpeg -report -f alsa -ar 48000 -acodec pcm_s32le -i mic_sv -f v4l2 -thread_queue_size 512 -framerate 20 -probesize 100 -i /dev/video0 -c:v h264_omx -acodec aac -f flv rtmp://209.85.230.23/live2/key error log: http://pastebin.com/P6z0V7vy
[06:42:31 CET] <debianuser> xtina: "is it possible pcm_s16le is not supported by my mems mic" -- yes. Especially if you explicitly configured it to use s32le in asoundrc. :)
[06:43:46 CET] <xtina> debianuser: i did not, i explicitly switched it to s16le in asoundrc
[06:43:55 CET] <xtina> but it still didn't accept pcm_s16le
[06:44:31 CET] <debianuser> xtina: Does arecord still work after switching to s16le in asoundrc?
[06:53:11 CET] <xtina> debianuser: it looks like no.. pi at raspberrypi:~ $ arecord -Dmic_sv -c2 -r48000 -fS16_LE -twav temp.v Recording WAVE 'temp.v' : Signed 16 bit Little Endian, Rate 48000 Hz, Stereo arecord: set_params:1204: Broken configuration for this PCM: no configurations available
[06:53:28 CET] <xtina> worked OK as 32
[07:01:05 CET] <debianuser> xtina: What hardware devices you have? Can you show more details about your soundcard? You can use alsa-info script: https://wiki.ubuntu.com/Audio/AlsaInfo it should automatically suggest you to upload your data and give you a link to it (you can run it as a regular user, it doesn't need root).
[07:02:37 CET] <xtina> debianuser: i have a ics43432 i2s MEMS mic
[07:02:58 CET] <xtina> it is the one this guy has here: https://github.com/nejohnson2/rpi-i2s. i can run alsa-info too
[07:03:15 CET] <furq> apparently s24_le will work
[07:06:34 CET] <xtina> furq: no such luck with 24 (in asoundrc+ffmpeg) either! :(
[07:07:24 CET] <xtina> debian-user: thanks for the help! here's my alsa info: http://www.alsa-project.org/db/?f=ea4d1faf2eb15d8924e354b8eff4fd4f5abfdd78
[07:11:56 CET] <debianuser> Ah, so, two cards: playback-only card 0 "ALSA" and capture-only card 1 "memsmic". xtina, what hardware params its driver reports to support? Can you copy the output of `arecord -v --dump-hw-params -fdat -d1 -Dhw:memsmic /dev/null` to some pastebin?
[07:13:33 CET] <xtina> sure, here: http://pastebin.com/dcCS9qbn hmm it seems to say 24/32 are supported...
[07:15:47 CET] <meriipu> what is the difference between scale_npp and nvresize referenced in http://developer.download.nvidia.com/compute/redist/ffmpeg/1511-patch/FFMPEG-with-NVIDIA-Acceleration-on-Ubuntu_UG_v01.pdf ? Is nvresize part of ffmpeg_NVIDIA_gpu_acceleration.patch ? Does recent version of that patch that build against recent commits exist, or is using scale_npp better?
[07:23:16 CET] <debianuser> xtina: yes, S24_LE and S32_LE, but no S16_LE.
[07:24:26 CET] <xtina> i'm getting no joy with 24 :(
[07:24:27 CET] <xtina> pi at raspberrypi:~ $ arecord -Dmic_sv -c2 -r48000 -fS24_LE -twav temp.v Recording WAVE 'temp.v' : Signed 24 bit Little Endian, Rate 48000 Hz, Stereo arecord: set_params:1204: Broken configuration for this PCM: no configurations available
[07:24:27 CET] <furq> i take it you were using pcm_s24le in ffmpeg
[07:24:36 CET] <furq> i'll take that as a yes
[07:24:52 CET] <xtina> yes i was
[07:24:58 CET] <xtina> ~/special/ffmpeg/ffmpeg -report -f alsa -ar 48000 -acodec pcm_s24le -i mic_sv -f v4l2 -thread_queue_size 512 -framerate 20 -probesize 100 -i /dev/video0 -c:v h264_omx -acodec aac -f flv rtmp://209.85.230.23/live2/xbjs-a3gp-0yvk-ftdg
[07:25:03 CET] <xtina> [alsa @ 0x2837760] cannot initialize hardware parameter structure (No such file or directory) mic_sv: Input/output error
[07:25:07 CET] <furq> weird
[07:25:13 CET] <xtina> i tried rebooting too
[07:25:14 CET] <furq> ftr this might not be the issue
[07:25:24 CET] <debianuser> xtina: don't forget to change your ~/.asoundrc to use S24_LE too
[07:25:25 CET] <furq> or probably isn't the issue, rather
[07:25:38 CET] <furq> but if you're getting buffer xruns then sending less data is probably a good start
[07:25:42 CET] <xtina> i did, debian
[07:25:48 CET] <furq> i guess you could try setting the rate to 8000 to see if it works
[07:28:12 CET] <xtina> ok, thanks, i will try
[07:28:14 CET] <debianuser> xtina: does it work with arecord? Try using hardware device directly: `arecord -v -fS24_LE -c2 -r48000 -d10 -Dhw:memsmic test.wav` Does it work or print any errors?
[07:31:54 CET] <xtina> debianuser: that does seem to work. i should show you my full asoundrc file: http://pastebin.com/MHmPn0ud
[07:32:01 CET] <xtina> the device i am calling for arecord and ffmpeg is mic_sv
[07:33:18 CET] <furq> does it work with -i hw:1
[07:53:31 CET] <debianuser> xtina: does `arecord -v -fS24_LE -c2 -r48000 -d10 -Dhw:memsmic test.wav` print any errors?
[11:10:46 CET] <JEEB> meh
[11:11:00 CET] <JEEB> why do I always forget how to show help for the AVOptions of a specific demuxer
[11:12:34 CET] <JEEB> -h protocol=file
[11:12:37 CET] <JEEB> doesn't seem to work :/
[11:16:17 CET] <JEEB> yeah, just looked at cmdutils.c
[11:16:39 CET] <JEEB> I guess I'll have to patch it in myself
[11:26:02 CET] <abd5932> Hello, I have two mp4 files served by nginx: A ( https://goo.gl/lYGXjC ) and B ( https://goo.gl/ZsnX7M ). File A starts pretty fast ca. 2s and it buffers/caches "future" content, however to B it takes more than 6s and it doesn't seem to buffer/cache. However A gets stuck periodically which makes it impossible to view (probably when it fails to cache), while B plays nicely all the time. Here are my questions: 1. What makes A start up so fas
[11:26:05 CET] <abd5932> t? 2. Is it possible to have both - fast start AND stable playing? I have not encoded neither of the files but here is the ffmpeg -i output of both ( http://pastebin.com/SuwYZkWP ) and AtomicParsley -T of A ( http://pastebin.com/GRWJkqNy ) and B ( http://pastebin.com/YGq0HjCU ). If you need any additional information, please, let me know. Thank you very much!
[12:10:36 CET] <rodmacedo> hi guys, i dont know for you but for me it is good morning!
[12:12:18 CET] <rodmacedo> Last week I sent one doubut of mine and I am not sure if somebody has answered the question, i didnt see any answer
[12:13:54 CET] <rodmacedo> The issue is that when processing .MXF files that have closed-caption information (vbi_vanc_smpte_436M) on output file generatred there is not closed-caption
[12:15:12 CET] <rodmacedo> the information is lost after the processing even though I dont even I dont do anything whit this data
[12:15:37 CET] <rodmacedo> * the information is lost after the processing even though I dont do anything whit this data
[12:15:50 CET] <rodmacedo> can somebody help me?
[12:25:04 CET] <rodmacedo> I put all information here: http://pastebin.com/fXJcei9d
[13:23:12 CET] <kubanc> hello. What commend should I use to record my desktop for a period of time..." ffmpeg -timelimit 60 -f x11grab -r 25 -s 740x580 -i :0.0 test.mp4" does not work
[13:29:12 CET] <sfan5> ideally you would use something like OBS
[13:30:11 CET] <sfan5> ffmpeg -f x11grab -r 25 -s 740x580 -i $DISPLAY -t 60 test.mp4 should work though
[14:05:59 CET] <bubbely> can someone help me install ffmpeg on wamp
[14:08:22 CET] <bubbely> google can't help me.
[14:08:29 CET] <bubbely> there aren't any to-date posts
[14:50:53 CET] <bubbely> hi
[14:51:02 CET] <bubbely> can someone help me install ffmpeg on wamp
[14:54:11 CET] <BtbN> on what?
[14:55:46 CET] <bubbely> http://www.wampserver.com/en/
[14:56:36 CET] <BtbN> That' just some Windows Apache and PHP collection, how would you "install ffmpeg" for it?
[14:56:57 CET] <bubbely> hm. proper set of dll files with the exe and instructions ?
[14:57:06 CET] <BtbN> to achive what?
[14:57:15 CET] <bubbely> ffmpeg loaded on server
[14:57:29 CET] <BtbN> ffmpeg is not an apache module.
[14:57:45 CET] <bubbely> is there not smoe form of php impelmentation of it
[14:57:59 CET] <BtbN> no idea, if there is one, it's not part of ffmpeg.
[14:58:06 CET] <bubbely> ..damn
[14:58:39 CET] <bubbely> http://stackoverflow.com/questions/11154291/install-ffmpeg-on-wamp does this look like something u might know about
[15:59:51 CET] <Mooniac> how can I "cut out" a certain section, like from 12:55 to 18:27 of a .mkv file?
[16:02:12 CET] <klaxa> not with one command
[16:02:12 CET] <klaxa> afaik
[16:03:29 CET] <Mooniac> split into sections, and then "re-concatenate"?
[16:06:53 CET] <klaxa> pretty much
[16:18:09 CET] <Mavrik> huh?
[16:18:23 CET] <Mavrik> Just use codec:copy and set start time and length
[16:20:28 CET] <klaxa> i think he wants the first and the second part of the video
[16:20:30 CET] <klaxa> and not the middle
[16:29:03 CET] <ZexaronS> Hello
[16:30:15 CET] <ZexaronS> is there a utility that would scan my database of various video/audio files and report how much audio silence is in each one, so I can identify silent videos, which may also happen because of network errors and other connection glitches of these downloaded videos, it happens sometimes, but I have many thousands of videos I can't check them all one by one
[16:33:12 CET] <meriipu> ZexaronS: there is a silencedetect filter, though I have never used it https://ffmpeg.org/ffmpeg-filters.html#silencedetect
[16:42:26 CET] <ZexaronS> thanks
[16:54:21 CET] <bubbely> https://packagist.org/packages/php-ffmpeg/php-ffmpeg <- can i get help with this in this channel?
[17:05:12 CET] <petert_> Hi, am am trying to pipe output from a HDHomeRun cable tuner to ffmpeg for real-time transcoding. However the first part of the data coming from the tuner seems to be corrupted and ffmpeg is not able to transcode due to this. If I on the other hand save the file from HDHomerun and afterwards run the ffmpeg command separately then the conversion is successful. Is there a way to discard the fx first 1mb of data?
[17:07:29 CET] <DHE> petert_: pastebin the command and output you're using. show us what goes wrong
[17:09:07 CET] <petert_> hdhomerun FFFFFF /tuner0 save - | ffmpeg -f mpegts -i - -c:v h264 -an -sn -f mpegts out.ts
[17:09:14 CET] <petert_> that's the command line
[17:11:03 CET] <petert_> [mpeg @ 0xaf207a0] Format mpeg detected only with low score of 25, misdetection possible!
[17:11:04 CET] <DHE> and the output from ffmpeg when it fails?
[17:11:37 CET] <petert_> I checked on the ffmpeg forum and found the same issue;
[17:11:38 CET] <petert_> https://ffmpeg.org/pipermail/ffmpeg-user/2012-October/010297.html
[17:12:10 CET] <petert_> It turns out that the first 400kb of data from the tuner is corrupted
[17:12:40 CET] <petert_> So my idea was to insert an intermediate function that can discard the first part of the piped data...
[17:14:07 CET] <petert_> Maybe I should mention that I did this in Windows environment. I did not have the chance to test on my ubuntu machine
[17:18:05 CET] <petert_> I will paste my exact output to pastebin later today
[19:21:33 CET] <alexspeller> I'm getting some broken video files from webrtc recordings. Sometimes the audio and video streams are different lengths. Is there any way to pad the audio and video in a reliable manner without knowing in advance if this problem is going to exist in the current video file?
[20:01:14 CET] <thebombzen> ohey -f nut doesn't support opus
[20:01:22 CET] <thebombzen> wondering if that's a bug or if it's cause of a frozen spec
[21:11:24 CET] <xtina> hey guys. i've added some new configs to asoundrc to get a boost on my MEMS mic. when i try fS32_LE recording from my new config, it works, but when I try f24_LE, it fails.
[21:11:30 CET] <xtina> my asoundrc is here: http://pastebin.com/MHmPn0ud
[21:11:47 CET] <xtina> Note that I'm using the mic_sv device. my command is: arecord -Dmic_sv -c2 -r48000 -fS24_LE -twav temp.v
[21:12:31 CET] <xtina> If I use '-Dhw:memsmic', f24_LE recording works, but not if I use -Dmic_sv. why is that?
[21:12:46 CET] <Diag> oh well theres your problem
[21:13:34 CET] <xtina> it says:
[21:13:35 CET] <xtina> Recording WAVE 'temp.v' : Signed 24 bit Little Endian, Rate 48000 Hz, Stereo arecord: set_params:1233: Sample format non available Available formats: - S32_LE
[21:13:43 CET] <xtina> but i thought the MEMS mic was capable of 24LE recording
[21:14:12 CET] <xtina> (i'm getting xrun msgs in ffmpeg so trying to go lower than 32LE)
[21:15:23 CET] <xtina> (I want to use mic_sv instead of memsmic, because mic_sv is configured in asoundrc to have a volume boost)
[21:16:14 CET] <xtina> Diag: what do you mean? :)
[21:17:04 CET] <Tiddys> xtina: idunno, youll have to wait for someone smart
[21:17:21 CET] <xtina> haha, ok
[21:19:08 CET] <llogan> if there is an #alsa channel you should ask there. doesn't look like you're using ffmpeg.
[21:20:19 CET] <xtina> llogan: yea, fair enough. the reason i'm trying to change from 32->24LE is that my ffmpeg command is throwing alsa xrun messages. ffmpeg is what i'm using
[21:20:30 CET] <xtina> i'll check for an alsa room
[21:28:46 CET] <phillipk> sorry to repeat my question, but anyone have a clue why my .ts files are getting rendered fine it seems, but they have a "start" other than 0?
[21:29:06 CET] <phillipk> when I use ffprobe I see: Duration: 00:01:07.51, start: 1.422422
[21:32:29 CET] <xtina> hey guys. does anyone know what this error means?
[21:32:30 CET] <xtina> [video4linux2,v4l2 @ 0x3509250] Dequeued v4l2 buffer contains 3133440 bytes, but 3110400 were expected. Flags: 0x00000004.
[21:32:52 CET] <Diag> Well obviously there were more bytes than expected
[21:33:58 CET] <xtina> do you know which param i should change to fix that? here's my entire command
[21:34:07 CET] <xtina> ~/special/ffmpeg/ffmpeg -report -f alsa -ar 48000 -acodec pcm_s32le -i mic_sv -f v4l2 -thread_queue_size 512 -framerate 20 -probesize 100 -i /dev/video0 -c:v h264_omx -acodec aac -f flv rtmp://209.85.230.23/live2/KEY
[21:35:07 CET] <Diag> looks like you got too many bytes
[21:36:12 CET] <phillipk> It seems like I get that "start" in my output even with something simple: ffmpeg nice.mp4 -f mpegts funkyStartValue.ts
[21:36:20 CET] <sware> anyone here know what might cause a video to play back where it's constantly going ahead 1/2 a second, then back. then forward continuously?
[21:36:28 CET] <sware> with h264 videos
[21:36:57 CET] <xtina> diag: i don't know why i have too many bytes
[21:38:21 CET] <xtina> it looks like i have the same issue as this guy, but no answers here: http://superuser.com/questions/1048637/ffmpeg-video-recording-freezes-after-invalid-data-found-when-processing-input
[21:39:54 CET] <Diag> xtina: well there you go its invalid data
[21:40:14 CET] <xtina> Diag: and how would you suggest to fix? i am out of ideas
[21:40:26 CET] <Diag> xtina: well get valid data of course
[21:40:32 CET] <xtina> Diag: are you trolling?
[21:40:38 CET] <xtina> not appreciated, really
[21:40:57 CET] <Diag> well idunno
[21:41:02 CET] <Diag> if its invalid data its invalid
[21:43:23 CET] <xtina> if anyone sees what i've done wrong in constructing my command, please let me know! here it is again
[21:43:28 CET] <xtina> ~/special/ffmpeg/ffmpeg -report -f alsa -ar 48000 -acodec pcm_s32le -i mic_sv -f v4l2 -thread_queue_size 512 -framerate 20 -probesize 100 -i /dev/video0 -c:v h264_omx -acodec aac -f flv rtmp://209.85.230.23/live2/KEY
[21:45:21 CET] <xtina> error report: http://pastebin.com/SW6DgPFu
[21:48:20 CET] <furq> xtina: did you try the other input selection methods
[21:48:22 CET] <furq> https://trac.ffmpeg.org/wiki/Capture/ALSA#Selectingtheinputcard
[21:49:24 CET] <xtina> furq: if I try Dhw:memsmic, i can do s24_le. but right now, i'm keeping s32_le, and trying to debug the other errors...
[21:49:40 CET] <xtina> i don't think s32_le is causing the stream to completely fail, it must be something else
[21:50:04 CET] <xtina> the errors are in here http://pastebin.com/SW6DgPFu but mainly
[21:50:05 CET] <xtina> - [video4linux2,v4l2 @ 0x3509250] Dequeued v4l2 buffer contains 3133440 bytes, but 3110400 were expected. Flags: 0x00000004. - /dev/video0: Invalid data found when processing input
[21:53:13 CET] <xtina> i would also like to know why i can do 24 with memsmic and not mic_sv.. when mic_sv uses the memsmic car...
[21:53:16 CET] <xtina> card*
[21:55:30 CET] <sware> anyone know what would make video jump forward and backward constantly? like it keeps replaying the same 1/3 second of video until it gets to the next?
[22:39:09 CET] <xtina> furq: is there a difference between using -f alsa and -f s32le?
[22:40:11 CET] <xtina> ie i can go -acodec pcm_s32le -f s32le ... -acodec aac OR -f alsa -acodec pcm_s32le ...-acodec aac?
[22:43:47 CET] <mete> good evening
[22:45:49 CET] <mete> I've recorded a TS srtream with ffmpeg, now I would like to cut it. I found that -ss combined with -c copy resulted in a half-second audi chop at the start. is there a way to prevent this chop (without reencoding the stream)?
[22:48:50 CET] <Diag> wew sorry had some connection issues
[22:49:05 CET] <Diag> xtina: yo anyways
[22:49:29 CET] <xtina> Diag: ?
[22:49:50 CET] <Diag> was thinking of this video https://www.youtube.com/watch?v=3m5qxZm_JqM
[22:49:57 CET] <Diag> "Clarke and Dawe - The Front Fell Off - YouTube"
[22:59:15 CET] <thebombzen> xtina: -f s32le is raw pcm_s32le
[22:59:22 CET] <thebombzen> -f alsa is the also output device
[22:59:39 CET] <thebombzen> you might want -f alsa -c pcm_s32le if you're looking at alsa output
[22:59:57 CET] <thebombzen> honeslty though I think it's dumb that there's like seventeen different formats for various forms of raw pcm
[23:00:11 CET] <thebombzen> it really should be -f rawaudio -c pcm_s16le
[23:00:18 CET] <thebombzen> and -f rawaudio -c pcm_s32le
[23:00:23 CET] <thebombzen> not have two separate formats
[23:00:52 CET] <xtina> thebombzen: i don't know which output i need :? but i'm streaming to youtube live
[23:00:54 CET] <thebombzen> or if flt is incompatible with pcm, at least use -f pcmaudio or rawpcm
[23:01:08 CET] <thebombzen> xtina: -f alsa output will end up just playing the video on your computer
[23:01:18 CET] <xtina> oh
[23:01:18 CET] <thebombzen> -f alsa input grabs from an alsa device
[23:01:27 CET] <thebombzen> playing hte audio* on your compter
[23:01:34 CET] <xtina> -f alsa input vs. -f s32le input?
[23:01:39 CET] <thebombzen> streaming to youtube live is done with rtmp afaik
[23:01:42 CET] <thebombzen> or rtsp
[23:01:44 CET] <xtina> yes, rtmp
[23:01:45 CET] <thebombzen> one of the two
[23:01:55 CET] <thebombzen> yea rtmp is usually an flv container
[23:01:56 CET] <thebombzen> -f flv
[23:02:01 CET] <thebombzen> (I think)
[23:02:06 CET] <xtina> yep, at the end i'm doing -acodec aac -f flv
[23:02:20 CET] <xtina> for input not sure whether to do -acodec pcmbalhblah -f alsa or -acodec pcmblahblah -f s32_le
[23:02:29 CET] <furq> the first one
[23:02:44 CET] <thebombzen> well -f s32le is for reading from a file that contains raw pcm_s32le samples
[23:02:55 CET] <thebombzen> -f alsa input reads from an alsa grab device
[23:04:07 CET] <xtina> hmm, i'm reading from a grab device (mems mic)
[23:04:56 CET] <xtina> furq: for some reason i'm getting 20fps when using -f s32_le and 1fps when using -f alsa
[23:05:10 CET] <xtina> so i thought it should be -f s32_le
[23:05:43 CET] <xtina> but i'm not reading from a file. i'm reading from the mems mic
[23:05:57 CET] <thebombzen> first of all it's -f s32le
[23:06:00 CET] <thebombzen> not -f s32_le
[23:06:18 CET] <thebombzen> let's start with that
[23:06:42 CET] <thebombzen> you should get: Unknown input format: 's32_le'
[23:06:48 CET] <thebombzen> second of all, even without the typo
[23:06:57 CET] <xtina> thebombzen: sorry just typoed here. in my cmd line it is correct
[23:07:16 CET] <thebombzen> ffmpeg -f s32le -i default should give this: default: No such file or directory
[23:08:01 CET] <xtina> thebombzen: you're right itdoes. but i am doing -i mic_sv, for my mems mic
[23:08:28 CET] <mete> I tested it now with another command, but unfortunately I also get the 0.5 second chop at the start. here is the command I used and also the output of ffmpeg command line: http://pastebin.com/7jPn5cnG
[23:09:07 CET] <xtina> and i don't get any errors
[23:09:47 CET] <furq> is it actually broadcasting your mic
[23:13:45 CET] <xtina> furq: hard to tell because the stream crashes right away :( perhaps not. i'll switch to alsa based on your explanations..
[23:19:33 CET] <xtina> is it bad if i'm getting lots of this message?
[23:19:34 CET] <xtina> cur_dts is invalid (this is harmless if it occurs once at the start per stream)
[23:20:32 CET] <xtina> (it's not only happening once).
[23:22:28 CET] <xtina> it also looks like hardly any video packets getting muxed
[23:36:35 CET] <xtina> i am also seeing audio constantly cutting out in my stream, and xrun messages everywhere, but my CPU usage is only 10%
[23:36:37 CET] <xtina> [alsa @ 0x35c6870] ALSA buffer xrun. 17kB time=00:00:03.80 bitrate= 35.7kbits/s speed=0.537x [alsa @ 0x35c6870] ALSA buffer xrun. 39kB time=00:00:07.50 bitrate= 43.1kbits/s speed=0.704x [alsa @ 0x35c6870] ALSA buffer xrun. 55kB time=00:00:11.15 bitrate= 40.7kbits/s speed=0.785x [alsa @ 0x35c6870] ALSA buffer xrun. 74kB time=00:00:14.80 bitrate= 41.0kbits/s speed=0.811x [alsa @ 0x35c6870] ALSA buffer xrun. 95kB t
[23:54:23 CET] <xtina> furq: sorry for all the questions >_< i've noticed that i'm getting far worse performance now that i'm doing everything in ffmpeg, vs piping arecord and raspivid into ffmpeg. e.g. now i'm getting alsa xrun errors, previously i wasn't. now my fps is almost 0, previously it was around 10fps
[23:54:48 CET] <xtina> my commands look almost identical to me
[23:54:54 CET] <xtina> previously: sudo rm temp.v mkfifo temp.v arecord -Dmic_sv -c2 -r48000 -fS32_LE -twav temp.v | raspivid -o - -t 0 -vf -hf -fps 20 -b 6000000 | \ ffmpeg \ -re \ -i - \ -i temp.v \ -ab 32k \ -ac 2 \ -strict -2 \ -c:v copy \ -c:a aac \ -report \ -vsync 2 \ -async 1 \ -f flv rtmp://209.85.230.23/live2/KEY
[23:55:06 CET] <xtina> now: ~/special/ffmpeg/ffmpeg -report -re -f alsa -ar 48000 -acodec pcm_s32le -i mic_sv -f v4l2 -thread_queue_size 512 -framerate 20 -probesize 100 -i /dev/video0 -c:v h264_omx -acodec aac -f flv rtmp://209.85.230.23/live2/KEY
[23:55:24 CET] <xtina> one difference is that i'm now using later versions of encoders, with my newly compiled ffmpeg. but i can't see why my performance would be worse now
[23:56:24 CET] <xtina> you'll notice i'm on 32 both times, so that's not it
[00:00:00 CET] --- Tue Feb 7 2017
More information about the Ffmpeg-devel-irc
mailing list