[Ffmpeg-devel-irc] ffmpeg.log.20140604

Thu Jun 5 02:05:01 CEST 2014

[14:27] <bs4h> hello all. got a problem with HD capture via video4linux2. I'm using v4l2-ctl to set capture options to 1280x720 at 30fps and pixel format to MJPG, and when I query via "v4l2-ctl -VP" it reports these values. however as I try to start capture via ffmpeg (cmd: ffmpeg -f video4linux2 -i /dev/video0 -t 3 -y test.avi), the settings are reset to YUYV which supports only 10fps. any ideas?
[14:28] <bs4h> or to rephrase: how to stop ffmpeg+video4linux2 from messing with camera parameters passed via v4l2-ctl; or: how to pass these parameters via ffmpeg.
[14:58] <user3> cls
[14:59] <user3> question. on Windows, ffmpeg -i 1.mp4 gives the error message "at least one input file must be specified" ???
[14:59] <fflogger> user3, please use a pastebin site (like www.pastie.org or www.pastebin.com) to show your EXACT ffmpeg command and the COMPLETE console output. Please don't paste your scripts, just the exact ffmpeg command line from the console.
[15:00] <termos> is there a difference between the filter "idet,yadif=0:deint=interlaced" and yadif=0:deint=interlaced"? It seems to give the same result on both interlaced and non-interlaced video
[15:01] <termos> except for the runtime, running idet in addition is slower
[15:03] <kodomo> hi folks! Could someonle please point me into the direction of some documentation on what extradata is, or how to construct it?
[15:04] <kodomo> I'm trying to reencode an audiostream (into an .mkv)... using AV_CODEC_ID_MP3 works fine... but AV_CODEC_ID_VORBIS fails with the error message that extradata were corrupt.
[15:05] <kodomo> I can't seem to find a tutorial/structured documentation on how to properly create the required metadata.
[15:22] <bs4h> https://trac.ffmpeg.org/ticket/1387 this actually seems to be my issue
[16:06] <Fjorgynn> so
[16:06] <Fjorgynn> that's awesome
[16:33] <Fjorgynn> hm
[17:20] <fling> How to convert to something that will play fine on an old player?
[17:21] <fling> do I need something like dvd in mp4?
[17:23] <sacarasc> What kind of player?
[17:24] <fling> something connected to a tv set
[17:25] <fling> it is not playing mp4 files my gf tried to playback on it
[17:25] <fling> the device is reading flashdrives fine
[17:25] <fling> sacarasc: I think she dried to playback x264 or something
[17:25] <sacarasc> You could google for what it allows.
[17:25] <fling> is it good idea to encode in mpeg2?
[17:26] <fling> sacarasc: idk the name of it
[17:27] <sacarasc> Does it not have a name or something on it?
[17:27] <fling> & -target pal-svcd final.mpg ?
[17:27] <fling> sacarasc: she does not tell me
[17:28] <sacarasc> You don't want svcd. Try pal-dvd
[17:28] <fling> sacarasc: thanks ;>
[17:28] <fling> will know tomorrow if it is working
[17:28] <fling> sacarasc: which container? mpg?
[17:28] <sacarasc> Yes.
[17:28] <fling> thanks :>
[17:30] <fling> Stream #0:0 -> #0:0 (h264 -> mpeg2video) ; Stream #0:1 -> #0:1 (aac -> ac3)
[17:33] <fling> sacarasc: does this look good for you? -> http://dpaste.com/3RYZAHN/
[17:34] <sacarasc> Not really, you're encoding to MPEG2, but it's the most likely thing to work.
[17:35] <fling> vp8 is good for you? :P
[17:35] <sacarasc> No.
[17:37] <fling> then what? x265?
[17:38] <rsdrsdrsdrsd> When I do ffmpeg -i name.mp4 I get something like: 720x480 [SAR 40:33 DAR 20:11], 3767 kb/s, SAR 1:1 DAR 3:2
[17:38] <rsdrsdrsdrsd> I know what sar and dar mean, but what do the sar and dar mean just after 720x480
[17:39] <sacarasc> H264, fling, it is likely supported by your device, though which profile/level is anyone's guess.
[17:39] <fling> ok, thanks
[17:39] <JEEB> rsdrsdrsdrsd, I guess one of those is from the container and the other is from the stream. If it's the aspect ratio box in mp4 it should override the stream's one, so I wonder why it's showing that at all
[17:40] <JEEB> oh well
[17:43] <rsdrsdrsdrsd> I want to change the pixel aspect ratio with mp4box without reencoding, but that doesn't seem to do the trick right
[17:43] <JEEB> remux with ffmpeg and use the setsar filter and see if that does the job?
[17:44] <rsdrsdrsdrsd> but then I need to reencode eveything isn't it?
[17:47] <sacarasc> -c copy
[17:47] <Fjorgynn> ;D
[17:48] <JEEB> rsdrsdrsdrsd, no
[17:48] <JEEB> use -c copy
[17:48] <JEEB> (and -map 0 if you want to take in all tracks from input)
[17:50] <rsdrsdrsdrsd> when I do ffmpeg -i input.mp4 -c copy -vf setsar=1 output.mp4 I get the error: Filtering and streamcopy cannot be used together.
[17:51] <rsdrsdrsdrsd> I am using ffmpeg version 2.2.2
[17:58] <rsdrsdrsdrsd> from what i've found by serahcing copying and filtering can't be done together
[17:58] <rsdrsdrsdrsd> that is why i tried to do it with mp4box
[17:59] <rsdrsdrsdrsd> but this isn't also working good
[17:59] <JEEB> hmm, that most definitely should be possible, though :/ wonder if it was another option
[18:08] <SporeGodOfFungi> Hello everyone!
[18:10] <SporeGodOfFungi> I'm having trouble installing ffmpeg on Centos 6... I've been running into some trouble. I'm getting a c compler test failed. Here is the tail end of my config.log
[18:10] <SporeGodOfFungi> BEGIN /tmp/ffconf.TYv1qmXi.c
[18:10] <SporeGodOfFungi>     1   int main(void){ return 0; }
[18:10] <SporeGodOfFungi> END /tmp/ffconf.TYv1qmXi.c
[18:10] <SporeGodOfFungi> gcc -c -o /tmp/ffconf.AEpetSPM.o /tmp/ffconf.TYv1qmXi.c
[18:10] <SporeGodOfFungi> gcc -o /tmp/ffconf.dqa4vgBe /tmp/ffconf.AEpetSPM.o -ldl /usr/lib/libfdk-aac.so.0
[18:10] <SporeGodOfFungi> collect2: ld returned 1 exit status
[18:10] <SporeGodOfFungi> C compiler test failed.
[18:10] <SporeGodOfFungi> I have a paste-bin of my process so far at http://pastebin.com/Y4bDiWHU
[18:11] <SporeGodOfFungi> I'm doing a compile & install on a newly spun centos 6 VM that I'm using to do a hand compile so I can build a spec file to create an RPM of it
[18:20] <SporeGodOfFungi> actually I just got past the error... added LDFLAGS=lstdc++
[18:20] <SporeGodOfFungi> before ./configure
[18:22] <rsdrsdrsdrsd> It is so confusing
[18:23] <SporeGodOfFungi> how od I specify multiple paths with --extra-libs=?
[18:23] <SporeGodOfFungi> just specify it multiple times, multiple paths separated by space enclosed in ""?
[18:24] <SporeGodOfFungi> yeh, I'm not sure why it's such a pain in the ass when you start adding the other plugins in
[18:48] <HD|Lapto1> hi all
[18:55] <HD|Lapto1> I have from my video surveillance system a load of jpegs named d-m-Y-H-i-s-X.jpg, X being the camera ID
[18:56] <HD|Lapto1> now given a start and an end date/time, how do I make a video out of this, with 25 fps?
[19:01] <DHE> I have a video feed (multicast MPEG-TS) I'm trying to convert. I suspect the PMT is changing and reassigning PIDs on occasion and ffmpeg isn't tracking updates - output loses streams. Anyone know if this can be worked around or should I just restart ffmpeg?
[19:25] <sine_> hi folks. is it possible to screen capture lossless with ffmpeg on windows ?
[19:25] <sacarasc> Yes.
[19:25] <sine_> can you tell me the command and then i will scan the wiki
[19:25] <sine_> please
[19:27] <ParkerR> sine_, https://trac.ffmpeg.org/wiki/How%20to%20grab%20the%20desktop%20(screen)%20with%20FFmpeg
[19:27] <sacarasc> libx264 with -crf 0 will do lossless.
[19:29] <brontosaurusrex_> sacarasc, irc --qp 0 is more safe to use
[19:30] <JEEB> granted, most people will never see 10bit x264
[19:31] <JEEB> which is where crf zero no longer is lossless
[19:31] <ParkerR> JEEB, how so?
[19:32] <JEEB> most ffmpeg builds are built with 8bit libx264
[19:32] <JEEB> thus most people will never use a 10bit encoder
[19:33] <ParkerR> Well... it wouldn;t be completely lossless for all intents and purposes it would be tot he human eye
[19:33] <JEEB> that said, I always mention that crf 0 is only lossless with 8bit
[19:33] <JEEB> well, the definition of lossless is rather simple
[19:33] <JEEB> what goes in comes out as-is
[19:34] <brontosaurusrex_> offtopic: known linux browse that does NOT support svg?
[19:34] <brontosaurusrex_> browser*
[19:35] <ParkerR> Well webkit supports it so that means opera/otter/qupzilla all do and Firefox does and Chrome does
[19:35] <ParkerR> I dont think so
[19:36] <ParkerR> brontosaurusrex_, What's up?
[19:36] <brontosaurusrex_> well, iam just staring at this: <img src="your.svg" onerror="this.src=your.png">   ?
[19:36] <brontosaurusrex_> and not sure what browser to use to test the png scenario
[19:36] <brontosaurusrex_> hey ParkerR
[19:36] <ParkerR> Oh haha
[19:37] <ParkerR> Everything supports svg, cant test ong fallback #fossworldproblems
[19:37] <ParkerR> *pbg
[19:37] <ParkerR> *png
[19:38] <sine_> thanks
[19:38] <brontosaurusrex_> Seems to be working "fine" with Chrome, Opera, Midori and Firefox.
[19:38] <ParkerR> brontosaurusrex_, internat explorer in wine?
[19:39] <brontosaurusrex_> where fine=svg.
[19:39] <ParkerR> It comes bundled with a slimmed down ie
[19:39] <brontosaurusrex_> ParkerR, right!
[19:39] <ParkerR> Install wine
[19:39] <ParkerR> wine iexplore
[19:45] <brontosaurusrex_> yeah, fails with ie8
[19:45] <ParkerR> :D
[19:46] <ParkerR> Every browser supports svg, has to use ie in linux to test svg/png fallback #fossworldproblems
[19:46] <ParkerR> brontosaurusrex_, this could use some life http://www.reddit.com/r/fossworldproblems/
[19:46] <ParkerR> Submit this there :P
[19:48] <brontosaurusrex_> ParkerR, actually i'am not sure, perhaps its just blocking js
[19:49] <ParkerR> I think js works
[19:53] <brontosaurusrex_> ParkerR, you on windows?
[19:53] <ParkerR> I can be...
[19:53] <ParkerR> Want me to test in a new ie?
[19:53] <ParkerR> Ok brb
[19:53] <brontosaurusrex_> yeah, just wonder
[19:53] <brontosaurusrex_> ing
[19:57] <ParkerR> brontosaurusrex_, Need you to repaste the link
[19:57] <ParkerR> I thought you had posted to channel lol
[19:58] <brontosaurusrex_> in msg window now ParkerR
[19:59] <ParkerR> brontosaurusrex_, IE11 its PNG
[19:59] <ParkerR> If I right click save image
[19:59] <brontosaurusrex_> cool
[19:59] <ParkerR> brontosaurusrex_, also tip
[20:00] <ParkerR> Add in big text on the page that tells you what image it loads
[20:00] <ParkerR> And then use https://browsershots.org/
[20:00] <brontosaurusrex_> ah, good idea
[20:01] <ParkerR> "Browsershots was blocked by dl.dropboxusercontent.com/robots.txt. Please read the FAQhttp://browsershots.org/faq#Blockedbyrobots. "
[20:01] <ParkerR> Awww
[20:01] <ParkerR> :/
[20:01] <ParkerR> Host in on your own server maybe lol
[20:01] <ParkerR> I have a small http server ruuning if you want me to throw it up real quick
[20:02] <brontosaurusrex_> no, i have my own www as well
[20:03] <ParkerR> Ahh ok cool
[20:08] <brontosaurusrex_> yeah, it breaks with some ancient browser
[20:09] <brontosaurusrex_> browsers*
[20:09] <brontosaurusrex_> bbl
[20:13] <ParkerR> brontosaurusrex_, so every ie version ever? :D
[20:30] <spectralsun> ParkerR: in regards to you saying wine for ie...no just....no.
[20:30] <spectralsun> http://modern.ie
[20:35] <rsdrsdrsdrsd> If a movie has the following specs: 720x480 [SAR 40:33 DAR 20:11], 3767 kb/s, SAR 853:720 DAR 853:480
[20:35] <rsdrsdrsdrsd> How can i convert it to square pixels and a fixed height of 480
[20:36] <rsdrsdrsdrsd> with correct aspect ratio
[20:36] <synth_> could i get some help with my command? i'm using the command: http://pastebin.com/0pP6qJgr and i was under the impression that if i used -start_number when overlaying the frame number text='%{n}'; that it would start at whatever frame i told it to start on; however the overlay starts at 0
[20:36] <rsdrsdrsdrsd> I'm so confused by all those dar and sar info
[20:36] <fflogger> synth_, please use a pastebin site (like www.pastie.org or www.pastebin.com) to show your EXACT ffmpeg command and the COMPLETE console output. Please don't paste your scripts, just the exact ffmpeg command line from the console.
[20:37] <synth_> llogan i included a pastebin link
[20:37] <llogan> one fflogger is a fake. which one do i shoot?
[20:37] <llogan> synth_: it does not include the console output which the ffloggers mentioned
[20:38] <rsdrsdrsdrsd> I thought that -vf scale=480*dar:480,setdar=1 would always create good and square pixels. But this isn't true, because there is some other dar and sar info
[20:38] <synth_> ok i'll try to get the output, but it's incredibly long
[20:38] <llogan> you can add -t 10 to encode 10 seconds. should be enough
[20:42] <synth_> it's not even 10 seconds long
[20:42] <synth_> the output is too large for pastebin for 119 frames
[20:42] <synth_> i'll do 1 sec
[20:43] <llogan> the head will suffice
[20:45] <llogan> ill be back in 15 mins
[20:45] <synth_> llogan: http://pastebin.com/N1tVyQnh
[20:55] <brontosaurusrex_> ParkerR, hmm, i'am geting svg with ie9, weird
[21:02] <llogan> synth_: try %{expr\\:n+50}
[21:02] <llogan> the escaping might be a bitch depending on your command, etc
[21:02] <llogan> if in quesiton keep adding \
[21:13] <synth_> haha okay
[21:13] <synth_> so i should just add the start_number to it
[21:15] <llogan> synth_: yes
[21:16] <llogan> unfortunately the overlay will produce a double instead of just the frame rate, so it will look something like 42.0000, IIRC
[21:16] <synth_> yeah i just saw that
[21:16] <synth_> that won't work for me unfortunately
[21:16] <synth_> is there any way to turn that into an int?
[21:16] <llogan> a poor man's way to avoid that would be to maybe place the overlay far to the right to cut off the decimals, but i don't know if that would work and would probably look like shit
[21:17] <llogan> otherwise, not that i know of
[21:17] <synth_> lame
[21:17] <llogan> refer to the code i guess
[21:17] <synth_> if there isn't a way to then i'll have to abandon the idea of putting the frame number on the output
[21:18] <llogan> another superlame method would be to add 50 frames of junk and then cut it out
[21:18] <synth_> yeahhh that won't work lol
[21:19] <synth_> i guess i could potentially shift it over and have it cut off the .
[21:19] <llogan> that's all of the loltips i can think of for today
[21:20] <llogan> wait..another joke: place yet another overlay of just a section of the underlayment to overlay over your drawtext decimals. lolololololol
[21:22] <synth_> http://mellow.me/tmp/output.mov i might be able to get away with moving the number over since it's right aligned in the top right
[21:24] <DelphiWorld> sup guys
[21:24] <synth_> o herro prease
[21:24] <llogan> synth_: might your best bet for now
[21:26] <DelphiWorld> ffmpeg -re -i rtmp://radio.gmo.ps:1935/live/livestream -c:v libx264 -vb 380k -x264opts colorprim=bt709:transfer=bt709:colormatrix=bt709:fullrange=off -c:a libfdk_aac -b:a 32k -ar 22050 -f flv -af "volume=15dB" rtmp://192.168.0.13:1935/live/alray;
[21:26] <DelphiWorld> the issue guys is that volume filter is not taking any effect, any idea?
[21:28] <llogan> synth_: big block. now i remember where i saw your name.
[21:28] <synth_> :)
[21:29] <llogan> DelphiWorld: you've been here enough to know that...
[21:29] <fflogger> DelphiWorld, please use a pastebin site (like www.pastie.org or www.pastebin.com) to show your EXACT ffmpeg command and the COMPLETE console output. Please don't paste your scripts, just the exact ffmpeg command line from the console.
[21:29] <llogan> ignore the ffalse fflogger
[21:33] <synth_> llogan it looks like the expression eval is happening in libavutil/eval.h; perhaps i could just correct it in there and recompile
[21:36] <sine_> ok learn me a bit here. coming from xvid/mp3 and mpeg2 era and into the x264 what is the appropriate audio codec to use for today
[21:36] <sine_> i have powerful computers
[21:36] <sine_> is aac the way to go
[21:37] <sacarasc> Audio needs next to no computing power, really. But AAC with a good encoder is a pretty good choice.
[21:37] <sacarasc> For lossy, anyway.
[21:40] <sine_> the ffmpeg one vo_aacenc is good ye
[21:41] <llogan> no. that's the worst one.
[21:41] <llogan> https://trac.ffmpeg.org/wiki/Encode/AAC
[21:47] <sine_> llogan: I am using windows and and pre built binarys
[21:48] <llogan> sine_: the native AAC encoder is probably better than libvo_aacenc
[21:48] <sacarasc> Piping it out and using neroAacEnc is good. :D
[21:49] <llogan> or qaac since Windows
[21:49] <synth_> llogan: it looks like the double return from expr has been discussed by others as well http://superuser.com/questions/565258/overlay-frame-number-with-ffmpeg-starting-at-1
[21:50] <sine_> listing dshow devices only shows my webcam and my microphone. how can i screen capture the desktop
[21:50] <llogan> synth_: i don't recall a follow-up bug report submitted by that user, but there might be one
[21:55] <sine_> it seems as if the ffmpeg page is directing me to alternate software in order to captuure the desktop to video. the reason i wanted ffmpeg to do it was because the options were a bit thin on the ground for lossless. im using windows and the only devices that are dshow listed are webcam and microphone
[22:11] <matthiaskrgr> what does ./configure  --bit-depth=10  affect ?
[22:12] <matthiaskrgr> when building ffmpeg
[22:14] <llogan> matthiaskrgr: you mean when configuring x264?
[22:14] <llogan> there is no --bit-depth=10 configure option for ffmpeg, AFAIK
[22:19] <matthiaskrgr> ooh ok
[22:19] <matthiaskrgr> llogan: yes, when configuring x264
[22:19] <matthiaskrgr> why did I join this channel for asking this :/
[22:19] <matthiaskrgr> sorry
[22:19] <llogan> because #ffmpeg is teh brest.
[22:20] <matthiaskrgr> :p
[22:20] <matthiaskrgr> well, can #ffmpeg answer the question then? :)
[22:20] <llogan> that will enable 10-bit encoding in x264 (and therefore via libx264 if using ffmpeg with the 10-bit x264 build)
[22:21] <matthiaskrgr> but what does 10 bit encoding mean?
[22:21] <matthiaskrgr> arch linux provides a regular libx264 and a libx264-10bit package now, so I wondered
[22:22] <llogan> http://ateme.com/Why-does-10-bit-save-bandwidth
[22:22] <llogan> see PDF
[22:24] <matthiaskrgr> ah so it make videos smaller
[22:24] <matthiaskrgr> ok this was basically everything I wanted to know, thanks :)
[22:25] <Mavrik> of course
[22:25] <Mavrik> most players won't play that... so unless you KNOW you need 10-bit... don't.
[22:25] <matthiaskrgr> oh so it's not backwards compatible?
[22:25] <Mavrik> how would that work :)
[22:26] <llogan> if you like anime (I don't) you'll probably find a bunch of samples to play with
[22:27] <matthiaskrgr> well ok. I'll stay with the regular x264 pkg then
[22:27] <Mavrik> hmm, wonder why 10-bit compresses better.
[22:33] <llogan> synth_: would you guys be willing to sponsor the feature request?
[22:34] <llogan> or place a bounty i should probably say
[22:35] <sine_> i got a warning data not aligned can lead to frame loss (which it did) what does it mean data not aligned. i was screen capturing at 25fps with x264 crf 0
[22:36] <sine_> ignore me
[22:45] <c_14> Mavrik: http://www.ateme.com/Why-does-10-bit-save-bandwidth
[22:45] <Mavrik> I totally overlooked that download link.
[22:46] <Mavrik> thx.
[22:51] <Mavrik> man that's alot of crap for one sentence explanation
[22:52] <llogan> admittedly i didn't read it
[23:01] <synth_> llogan sorry just saw your message
[23:01] <synth_> sponsor the feature request? how so?
[23:03] <llogan> synth_: might be a good chance to try out bountysource
[23:05] <llogan> eventually your ticket will appear at https://www.bountysource.com/trackers/458692-ffmpeg
[23:07] <llogan> then you can click on it and offer an amount. BS will take 10% upon funds withdrawl by the developer once you are satisfied with the result  (IIRC)
[23:10] <llogan> synth_: looks like it has appeared now: https://www.bountysource.com/issues/2508038-ability-to-use-expressions-using-drawtext-to-offset-frame-number-without-returning-double
[23:18] <synth_> interesting
[23:18] <synth_> i don't know if our company will be willing to place a bounty for it
[23:21] <synth_> doesn't look like any ffmpeg issues have bounties set
[00:00] --- Thu Jun  5 2014