[Ffmpeg-devel-irc] ffmpeg.log.20130426

[burek]

[00:07] <yns_> Hi
[00:08] <yns_> I would like to convert pjpeg
[00:08] <yns_> cant find ut how tough
[00:10] <yns_> any suggestions?
[00:13] <yns_> cant find anything about pjpeg in the documanetation, is there another name for pjpeg maybe?
[00:19] <Trashlord> hey. I'm trying to stitch together audio and a still image to create a video. I'm doing -pix_fmt yuv420p, but it says unknown format, why?
[00:20] <Trashlord> actually, it says it can't find a suitable output format for that
[00:55] <tonsofpcs> anyone have experience using ffmpeg to render XDCAM HD 50?
[01:13] <yns_> annyobdy knows how to convert into pjepg with ffmpeg
[02:15] <Anaphaxeton> hi! i have a TV set that supports playback with a usb stick. it limits you a lot about containers and what streams they should include
[02:16] <Anaphaxeton> so basically i am  trying to convert a dvd to a .mpg file that is a PS container with mpeg2 video and pcm audio
[02:16] <Anaphaxeton> the problem is with the audio
[02:17] <Anaphaxeton> ffmpeg -i Downton.Abbey.Season.2.Episode.5.mkv -c:a pcm_u16be -f vob -c:v copy Downton.Abbey.Season.2.Episode.5.mpg
[02:17] <Anaphaxeton> one of my attempts to make it work...
[02:17] <Anaphaxeton> packet too large, ignoring buffer limits to mux it
[02:17] <dozm9> Does does anyone have any good resources/tutorials for taking arbitrary input videos and outputting to mp4 container for multiple devices? 1080,720,480,360
[02:18] <Anaphaxeton> ^ that one always appears with whatever pcm and with or without -f vob
[05:16] <defaultro> hey folks, can ffmpeg transition from one video to another? I mean, the first video is about to end in 10 seconds, our next video starts to fade-in in that time frame
[08:43] <Aziroshin> I am trying to encode a "preview" video. The perk would be that the encoding process finishes really fast.
[08:43] <Aziroshin> However, I belive I am misunderstanding how this works, because, despite horrible quality settings and 2 fps (testwise), the encoding process doesn't really speed up much.
[08:44] <Aziroshin> The CPU is also not really used too much by ffmpeg during that preview encode.
[08:44] <Aziroshin> The line I am using goes like this: ffmpeg -y -i ../take1.video.mkv -r 2 -c:v libx264 -s 320x240 -crf 40 -preset superfast 1.mkv (testwise)
[08:45] <Aziroshin> Here's the output: http://dpaste.com/1074101/
[08:47] <Mavrik> hmm
[08:48] <Mavrik> Aziroshin, ok, whats the framerate of your input?
[08:50] <Aziroshin> It's 30
[08:50] <Aziroshin> It's a yuv4mpeg encoded file in an mkv container (the input)
[08:51] <Mavrik> yeah well
[08:51] <Mavrik> as you can see theres alot of dropped frames
[08:51] <Mavrik> since youre forcing it to 2 fps on output
[08:51] <Aziroshin> Yeah
[08:51] <Mavrik> and youre probably hitting your I/O limit
[08:52] <Mavrik> since raw 720p video DOES need a huge amount of I/O bandwidth to read from disk
[08:52] <Aziroshin> Hmm
[08:52] <Mavrik> that also explans why your CPU is idling ;)
[08:52] <Aziroshin> Interesting point, I should do a test sometime with small snippets on a ramdisk and see how it flies.
[08:53] <Aziroshin> It's interesting that the fps spikes quite high in the beginning, and then flattens down quickly.
[08:53] <Mavrik> Aziroshin, for 720p 30fps raw video to be read from HDD you need a 110MB/s sustained read
[08:54] <Mavrik> that's without the writes needed to store results :)
[08:54] <Aziroshin> To be read at realtime, that is?
[08:55] <Mavrik> yeah
[08:55] <Aziroshin> I see. o.o
[08:55] <Aziroshin> That's like some more than 1/3 of a quite fast harddisk.
[08:55] <Mavrik> um
[08:55] <Mavrik> actually I havent seen 7200rpm harddisk that could do that ;)
[08:55] <Mavrik> SSDs only
[08:56] <Aziroshin> I think I've seen a magnetic one in a review somewhere, but it might have been a 15k one.
[08:56] <wolf85kunst> hello
[08:56] <Aziroshin> Hello.
[08:56] <Mavrik> Aziroshin, most HDDs top out at 100MB/s on inner platters
[08:57] <Mavrik> usually the curve is like 70-100MB/s
[08:57] <Mavrik> quite annoying for video work :)
[08:57] <Mavrik> Aziroshin, try storing output to another disk
[08:58] <Aziroshin> Yeah,well, even if I had the money for more harddisks, that alone would probably not give the speed increase I had dreamed of. xD
[08:58] <wolf85kunst> i wonder if ffmpeg is the good solution to use in a script... in fact i developp a script who get some videos info.. and i wonder if ffmpeg is not deprecated. i sound libav.. what is the best solution ? thanks a lot
[08:58] <Aziroshin> It's not deprecated at all.
[08:58] <Mavrik> Aziroshin, mhm& why do you have video in raw format? :)
[08:59] <Mavrik> wolf85kunst, no, ffmpeg is not deprecated
[08:59] <wolf85kunst> *** THIS PROGRAM IS DEPRECATED ***
[08:59] <wolf85kunst> This program is only provided for compatibility and will be removed in a future release. Please use avconv instead.
[08:59] <Aziroshin> It's just that the debian/ubuntu developers put in avconf instead of ffmpeg and renamed it ffmpeg instead, putting in their own avconf ffmpeg fork a fmpeg-deprecation message.
[08:59] <Mavrik> wolf85kunst, yes, ffprobe JSON output is perfect for use in script
[08:59] <Mavrik> wolf85kunst, thats misleading and obsolete
[09:00] <Mavrik> Aziroshin, because the better option is to just use a little CPU when saving video and store it into one of the lightlt losless compressed formats like huffyuv
[09:00] <Aziroshin> huffyuv, interesting. Ive heard of it.
[09:01] <Aziroshin> I should probably do that.
[09:01] <Mavrik> yeah, it uses simple lossless huffman compression
[09:01] <Mavrik> which makes a huge difference when I/O is your bottleneck :)
[09:02] <Aziroshin> I record gameplay footage with glc. It stores the video in yuv4mpeg in an internal .glc format. I need to pipe it out before I can do anything. Before, I used libx264 for that. I figured I could get some more quality during thet re-encoding processes further down the pipeline if I would not do that.
[09:02] <Mavrik> mhm
[09:03] <Mavrik> Aziroshin, I found out that using light compressing when grabbing video from gameplay is worth it
[09:03] <Aziroshin> And good quality settings for libx264 takes ages
[09:03] <Mavrik> since otherwise I always saturated disk
[09:03] <Aziroshin> Well, glc does yuv4mpeg or bust.
[09:03] <Mavrik> which is way worse than give part of a core over for compression
[09:03] <wolf85kunst> Mavrik, thanks you mean it's better to use ffprob that's right ?
[09:04] <Aziroshin> I need to pipe it out after the file is completely recorded.
[09:04] <Aziroshin> I have stuffed around with direct piping, though. These things don't work out sound-wise so far.
[09:04] <Mavrik> mhm
[09:04] <Mavrik> I think I used some other game recording software
[09:04] <Aziroshin> Which one?
[09:06] <Mavrik> um
[09:06] <Mavrik> playclaw I think
[09:06] <Mavrik> only one that could do compression so I could grab 1920
[09:06] <Aziroshin> Oh, sometimes I forget that not everyone on these freenode channels is automatically living in the UNIX world
[09:06] <Aziroshin> I use linux. XD
[09:07] <Aziroshin> And, well, through glc I am used to files that get insanely large.
[09:07] <Mavrik> mhm
[09:07] <Mavrik> Aziroshin, I figured gaming == windows
[09:07] <Aziroshin> I stopped doing Windows years ago.
[09:08] <Mavrik> havent tried grabbing video for games form linux, usually performance was bad enough as it is
[09:08] <Mavrik> anyway, sleep time
[09:08] <Aziroshin> That yuv4mpeg extract I am working with is like 173 GB in size
[09:08] <Aziroshin> Uh... see ya. XD
[09:15] <wolf85kunst> is ffprobe is based on ffmpeg ?
[09:20] <Iv0ry> Hey guys if mmpeg says: [mov,mp4,m4a,3gp,3g2,mj2 @ 03781960] moov atom not found Invalid data found when processing input that means the file is damaged right? :(
[10:36] <ubitux> wolf85kunst: ffprobe is one of the tool of the FFmpeg project, just like ffmpeg, ffserver or ffplay
[10:48] <ghost145> hello? the question about FFV1 and SNOW codecs - are there any changelog info or any info about new versions of this codecs or they have 1 version at all?
[11:01] <wolf85kunst> i have a question
[11:01] <wolf85kunst>  why ffmpeg return result on exit error stream ?
[11:02] <wolf85kunst> i have to do that to get ffmpeg return in a variable :
[11:02] <wolf85kunst> ffmpeg_result=`ffmpeg -i "$line" 2>&1
[11:03] <wolf85kunst> the question is why i have to do that ?
[11:03] <wolf85kunst> thanks for answer
[11:08] <Peace-> hi everyone i am trying to do this  ffmpeg -f  alsa  -ac 1 -ar 48000  -i hw:1,0   -acodec pcm_s16le -vn audio_recording.wav  but  i get
[11:09] <Peace-> this error Unrecognized option 'f '.
[11:09] <Peace-> why?
[11:15] <ubitux> ghost145: i don't think snow had recent changes lately, but ffv1 got improved
[11:15] <ubitux> (’ version 1.3 iirc)
[11:16] <Peace-> ubitux: http://paste.ubuntu.com/5603943
[11:17] <TheSchaf> hello! i'm generating preview audio mp3s from mp3s using ffmpeg. i want to add a fadeout. however it seems to ignore the ss param, no matter what i do...
[11:17] <TheSchaf> -ss 20 -t 10 -i foo.mp3  -ab 192k -ar 48000 -acodec mp3 -af "afade=t=out:ss=9:d=1" -y foo_demo.mp3
[11:17] <TheSchaf> this is the command line i use
[11:17] <TheSchaf> i've tried various parameters for ss but it always fades out after 1s directly
[11:17] <TheSchaf> can someone help me? :)
[11:17] <Peace-> ubitux: i tried even to use the command line from ffmpeg doc http://paste.ubuntu.com/5603951 this is for screen grab
[11:17] <ubitux> Peace-: works here...
[11:18] <Peace-> ubitux: that infact the strange thing
[11:18] <Peace-> ubitux: i remembered it worked
[11:18] <ubitux> Peace-: maybe a broken revision, try to upgrade
[11:18] <Peace-> ubitux: mm i tried to use another ffmpeg , avstuff
[11:18] <Peace-> ubitux: it says the same thing
[11:19] <ghost145> hello! the question about FFV1 and SNOW codecs - are there any changelog info or any info about new versions of this codecs or they have 1 version at all?
[11:19] <Peace-> maybe some library that is fucked up ?
[11:19] <ubitux> Peace-: --enable-random
[11:19] <ubitux> remove this...
[11:19] <Peace-> ok
[11:20] <ubitux> ghost145: read the backlog.
[11:23] <ubitux> TheSchaf: afade=t=out:ss=9:d=1  ss = start sample; it's expressed in sample, not second
[11:23] <ubitux> try "ns"
[11:23] <ubitux> Peace-: "st" sorry
[11:23] <ubitux> aaarg
[11:23] <ubitux> TheSchaf: "st" sorry ^
[11:24] <ubitux> afade=t=out:st=9:d=1
[11:24] <TheSchaf> aaah, okay :)
[11:25] <TheSchaf> http://www.ffmpeg.org/ffmpeg-filters.html#Examples-29 is wrong then?
[11:25] <ubitux> what is wrong?
[11:25] <TheSchaf> it says ss there
[11:26] <ubitux> erm..
[11:26] <ubitux> you're right
[11:26] Action: TheSchaf is confused :p
[11:27] <ubitux> it will be fixed
[11:27] <TheSchaf> thanks :)
[12:06] <xlinkz0> i want to make a hash for the data in a frame, what buffers should i look at?
[12:06] <xlinkz0> i still don't understand how this should be read
[12:52] <spaam> w
[12:52] <spaam> ops
[12:57] <Peace-> ubitux: :S http://paste.ubuntu.com/5604198
[12:57] <Peace-> ubitux: just compiled
[12:58] <Cobra_Fast> Hi! I have a problem applying a watermark to a scaled video. The watermark is positioned relatively to the original size rather than the scaled size. A full description is at http://stackoverflow.com/q/16148068/522479 Any ideas?
[12:59] <Peace-> ubitux: ok worked with another resolution
[13:00] <ubitux> Peace-: --enable-pic --enable-sram --enable-hardcoded-tables --enable-memalign-hack
[13:00] <ubitux> why do you do that?
[13:00] <ubitux> where did you pick all those flags?
[13:01] <ubitux> btw, --enable-random is to randomize completely the configuration...
[13:01] <ubitux> it's only meaningful for testing compilation
[13:01] <relaxed> Cobra_Fast: use the most recent version, look at -filter_complex in the man page, and have scale first in the filter chain.
[13:02] <ubitux> Cobra_Fast: multiple -vf, huh?
[13:03] <relaxed> I didn't notice that :/
[13:03] Action: relaxed needs sleep
[13:04] <ubitux> Cobra_Fast: also, wtf is this pad?
[13:04] <Peace-> ubitux: wel the most of times i compile ffmpeg with every possibile flags
[13:04] <Peace-> ubitux: and it worked fine untill a week ago
[13:04] <ubitux> Peace-: those flags are insane, don't do that.
[13:04] <Peace-> :D hahaha
[13:04] <ubitux> you should only --enable external libraries
[13:05] <Peace-> ok ok
[13:05] <Peace-> i will try again without those flags
[13:05] <ubitux> the rest is automatic and will pick what's best for you
[13:05] <Cobra_Fast> ubitux: whats wrong with multiple -vf?
[13:05] <ubitux> Cobra_Fast: one will be ignored i'd say.
[13:05] <ubitux> Cobra_Fast: just overlay the pic and get done with it
[13:06] <ubitux> you don't need to pad anything
[13:06] <ubitux> nor separate the filtergraph
[13:06] <Cobra_Fast> ubitux: yes, I've already changed to using the X and Y arguments of the overlay filter
[13:06] <Cobra_Fast> so all in one vf?
[13:06] <ubitux> you need to upgrade anyway, your ffmpeg is 2 years old
[13:07] <Cobra_Fast> N-52381-g2288c77 is only a couple days old
[13:07] <ubitux> ah missed that, i thought it was 0.8
[13:07] <ubitux> anyway, yeah then just overlay
[13:08] <ubitux> -i in.mp4 -i watermark.png '[0:0][1:0] overlay'
[13:08] <ubitux> -i in.mp4 -i watermark.png -filter_complex '[0:0][1:0] overlay'
[13:08] <wolf85kunst>  i have a question
[13:08] <wolf85kunst>  why ffmpeg return result on exit error stream ?
[13:08] <wolf85kunst> i have to do that to get ffmpeg return in a variable :
[13:08] <wolf85kunst> ffmpeg_result=`ffmpeg -i "$line" 2>&1
[13:09] <wolf85kunst> the question is why i have to do that ?
[13:09] <wolf85kunst> thanks for answer
[13:09] <ubitux> you mean why it's on stderr?
[13:10] <Cobra_Fast> ubitux: -i in.mp4 -i watermark.png -filter_complex "[0:0] scale=...,[1:0] overlay=..." ?
[13:10] <ubitux> Cobra_Fast: what do you want to scale?
[13:10] <Cobra_Fast> in.mp4
[13:10] <ubitux> yeah then it should work
[13:11] <wolf85kunst> ubitux, yes i do (xcuse me for my english)
[13:11] <wolf85kunst> i mean the return is in fact stderr...
[13:12] <ubitux> wolf85kunst: you're likely not supposed to parse what's on stderr; i don't remember exactly why it's done that way though
[13:12] <wolf85kunst> that's why i have to tap that ffmpeg_result=`ffmpeg -i video.avi 2>&1`
[13:12] <ubitux> what information are you trying to grab?
[13:13] <wolf85kunst> video resolution, codec, duration
[13:13] <ubitux> use ffprobe
[13:13] <ubitux> and one of the various output
[13:13] <wolf85kunst> i ear that
[13:13] <ubitux> you have json, xml, csv, etc
[13:13] <ubitux> wait simpler and appropriate to parse
[13:13] <ubitux> way*
[13:13] <wolf85kunst> for the moment i work with regex and ffmpeg
[13:14] <ubitux> that's not a good idea
[13:14] <wolf85kunst> i have other question...but is not necessary about ffmpeg
[13:15] <wolf85kunst> well , if i tap :
[13:15] <wolf85kunst>  ffmpeg_result=`ffmpeg -i video.avi 2>&1`
[13:15] <wolf85kunst> echo "$var" and echo $var is not the same result, i just want to understand why
[13:15] <wolf85kunst> any idea ?
[13:16] <ubitux> shell escaping
[13:16] <wolf85kunst> with echo $var --> all on one line
[13:16] <ubitux> try: ffprobe -v 0 -show_entries stream=duration,width,height -of flat video.avi
[13:16] <wolf85kunst> with echo "$var" --> severall line
[13:17] <ubitux> now try to replace "flat" with "json" or whatever
[13:17] <ubitux> if you want to use shell, you can use flat=s=_
[13:17] <ubitux> which allows you to eval directly the result
[13:17] <ubitux> and have the results into shell variables directly.
[13:18] <ubitux> replace "-show_entries stream=duration,width,height" with "-show_streams" to see what other keys are available
[13:19] <wolf85kunst> ffprobe: unrecognized option '-show_stream'
[13:19] <wolf85kunst> ffprobe: unrecognized option '-show_entries'
[13:19] <ubitux> upgrade.
[13:20] <ubitux> your ffmpeg is at least 6 months outdate
[13:20] <Cobra_Fast> ubitux: how do i select an input for -b:v?
[13:20] <ubitux> -b:v:0 ?
[13:21] <ubitux> (you have multiple video streams?)
[13:21] <ubitux> ah you have the png; try -filter_complex "[0:0] scale=...,[1:0] overlay=... [out]" -map "[out]"
[13:23] <Cobra_Fast> wait my command is all spaghetti now... let me sort it out
[13:24] <wolf85kunst> ubitux, i will upgrade thanks
[13:25] <wolf85kunst> but somedy can explain to me the problem --> "$var" and $var - the difference please
[13:28] <ubitux> that's a shell question
[13:29] <ubitux> and the question is unclear given the context
[13:29] <Cobra_Fast> -y -ss 0 -t 0:0:30.0 -i 'in.mov' -i 'watermark.png' -filter_complex "[0:0] scale=400:225,[1:0] overlay=305:0 [out]" -map "[out]" -b:v 896k -r 20 -an  'target.flv' - for some reason target.flv is 95x39 and .05 seconds long!?
[13:31] <ubitux> Cobra_Fast:  try "[0:0] scale=400:225 [p]; [p][1:0] overlay=305:0 [out]"
[13:32] <Peace-> ubitux: http://paste.ubuntu.com/5604280
[13:33] <ubitux> Peace-: mmmh.
[13:33] <Cobra_Fast> ubitux: YES! IT WORKS! Thank you very much, you are my hero today! :D
[13:33] <ubitux> Peace-: -ar is an output option afaik
[13:33] <ubitux> Cobra_Fast: no prob, being a hero is my daily life
[13:34] <ubitux> Peace-: but i admit the error is unclear..
[13:35] <Peace-> ubitux: mmm i guess it's the kernel ffmepg has never done this shit http://paste.ubuntu.com/5604298
[13:37] <ubitux> your original command line works fine here
[13:38] <ubitux> Peace-: http://pastie.org/private/zqjahjtsbna6ubu51yujca
[13:40] <Peace-> ubitux: mm untill few days ago ffmpeg worked fine ...
[13:40] <Peace-> ubitux: i will try to reboot with another kernel
[13:42] <ubitux> have fun
[13:52] <Peace-> mah i dunno undestand even with this kernel i got problems
[13:53] <ubitux> given the level of insanity of your previous configure flags, you may want to make distclean and run configure and make again
[13:54] <Peace-> ubitux: i did before
[13:54] <Peace-> the point it was that ffmpeg untill a week ago worked fine with those insane flags
[13:55] <Peace-> i suspect that i did a upgrade to my distro
[13:55] <Peace-> and it's that the problem
[13:55] <ubitux> mmh
[13:56] <ubitux> Peace-: can you pastebin again with -v verbose?
[13:56] <ubitux> -v debug even
[13:59] <Peace-> ubitux: http://paste.ubuntu.com/5604358  ; http://paste.ubuntu.com/5604360
[14:01] <ubitux> Reading option '-f ' ...Unrecognized option 'f '.
[14:01] <ubitux> fix your script.
[14:01] <Peace-> mm
[14:02] <Peace-> ubitux: i always used that script
[14:02] <Peace-> anyway let me try see doc
[14:03] <ubitux> i doubt ffmpeg ever recognize "-f "
[14:03] <ubitux> +d
[14:03] <yns> Hello, I'm trying to see what frames of my converted video are keyframes. Tried using -vstats_file but I have no clue how to interpret the file. How can I see what frames are keyframes?
[14:04] <ubitux> yns: you can use -vf "drawtext=%{pict_type}"
[14:04] <ubitux> also, you can use -vf select to pick only the i-frames
[14:04] <ubitux> for vstats_file i don't know, maybe there is some doc on ffmpeg.org
[14:05] <yns> thanks for help? sorry for the newbie-nes but are i-frames the same as keyframes?
[14:05] <Peace-> ubitux: mmm http://ffmpeg.org/ffmpeg.html  i found here   ffmpeg -f alsa -ac 1 -i hw:1 -f video4linux2 -i /dev/video0 /tmp/out.mpg
[14:06] <ubitux> Peace-: i'm talking about the space after the -f
[14:06] <ubitux> in your script
[14:06] <Peace-> ah damn shit the space
[14:06] <ubitux> copy paste the cmd line you showed me, it works.
[14:06] <ubitux> but your script is broken
[14:06] <ubitux> it sends the space
[14:06] <yns> meant thanks for help!* of course :)
[14:07] <ubitux> yns: not exactly the same, but you can likely consider it to be :p
[14:07] <yns> hehe ok :)
[14:07] <yns> so: ffmpeg -i [input] -vf "drawtext=%{pict_type}" [output]  ?
[14:08] <ubitux> yes
[14:08] <ubitux> you can use ffplay to see what it looks like
[14:08] <Peace-> ubitux: :S worked fixing the space
[14:08] <ubitux> yns: note that all i-frames might not be considered "key-frames"; basically key-frame is format level, while i-frame is codec level
[14:09] <yns> hm okay
[14:09] <ubitux> yns: anyway, ffplay input -vf "drawtext=%{pict_type}:fontsize=30:x=20:y=20:fontcolor=yellow"
[14:09] <ubitux> i'm assuming you have a recent enough ffmpeg though
[14:09] <yns> hm
[14:10] <ubitux> otherwise, you can just -vf showinfo and look at your terminal :)
[14:10] <yns> but that would just playback only the keyframes of the movie right?
[14:10] <yns> I would like to get a list f all frames that are keyframes, that possible?
[14:10] <ubitux> no, it will write a text on each frame
[14:10] <ubitux> yes
[14:10] <yns> oh ok I'm going to try thanks :)
[14:11] <ubitux> see the documentation of vf select
[14:12] <yns> I'll have a look at that too yes :)
[14:14] <yns> ffplay not found :(
[14:15] <yns> it doesn't automatically install when installing ffmpeg?
[14:15] <ubitux> it should be available
[14:15] <ubitux> did you build ffmpeg yourself?
[14:15] <ubitux> make sure it was built on a system where the sdl headers were available
[14:15] <yns> no
[14:15] <yns> i didnt build myself
[14:16] <ubitux> what distro?
[14:16] <yns> http://www.evermeet.cx/ffmpeg/
[14:16] <yns> downloaded ffmpeg-1.2.7z
[14:16] <ubitux> ah, mac os
[14:17] <ubitux> no idea then
[14:17] <ubitux> "--disable-ffplay"
[14:17] <yns> unpacked it, copied it to /usr/local/bin and ran it
[14:17] <yns> -bash: --disable-ffplay: command not found
[14:17] <ubitux> http://www.evermeet.cx/ffplay/
[14:17] <ubitux> no i meant it was built with "--disable-ffplay"
[14:18] <ubitux> that's what i read in the page
[14:18] <yns> ah ok. sorry I'm a super newbie :)
[14:18] <yns> thanks
[14:18] <yns> that procedure I mentioned is the way to go right?
[14:18] <yns> unpacked it, copied it to /usr/local/bin and ran it
[14:23] <yns> [drawtext @ 0x10207bc00] Key '%{pict_type}:fontsize' not found.
[14:25] <yns> using: ffplay input -vf "drawtext=%{pict_type}:fontsize=30:x=20:y=20:fontcolor=yellow"
[14:25] <ubitux> yes, 1.2 is too old :)
[14:25] <ubitux> you need git ;)
[14:26] <yns> ok :)
[14:26] <ubitux> anyway, check -vf showinfo, it will print all kind of stuff
[14:26] <ubitux> notably the "type" you are interested in
[14:26] <yns> ok :)
[14:26] <ubitux> you can also extract all the frame with -vf select, but i let you read the doc to see how to achieve that
[14:26] <ubitux> (hint: ffmpeg.org ’ documentation ’ filters)
[14:27] <yns> thank you
[14:27] <yns> -vf showinfo worked
[14:27] <yns> going to have a look at the doc now :)
[14:28] <yns> type:I means it is a I-frame which could be considered keyframe?
[14:31] <ubitux> yns: yes
[14:32] <ubitux> yns: I-frame means it's a "complete" frame standalone; key-frame is more like a format wording which points to a point where you can seek (it's often a keyframe, but not all i-frame could be marked as key-frame)
[14:32] <yns> hm
[14:32] <ubitux> meh random wording, but i hope you get the point
[14:32] <yns> so no way to see actual keyframes then?
[14:33] <ubitux> you're really interested in the key frames?
[14:33] <ubitux> (what's the purpose?)
[14:33] <ubitux> get the seek points?
[14:34] <yns> no, I'm really of track I guess :) Working with openFrameworks, want to encode my videos so every frame has a keyframe and want to make sure it worked
[14:36] <ubitux> if all frames are intra, you can technically seek anywhere; even if the format hasn't marked all the frames as key frames... i guess
[14:36] <ubitux> oh you could use ffprobe to get the key frames
[14:36] <ubitux> that should be faster
[14:37] <yns> okay
[14:37] <yns> what does it mean hen all frames are intra?
[14:37] <ubitux> ffprobe -v 0 -show_entries frame=key_frame -of flat input
[14:37] <ubitux> try this :)
[14:38] <yns> I will, Guess I have to install ffprobe first :)
[14:38] <ubitux> you can download it from the same website
[14:38] <ubitux> it's one of the FFmpeg tools
[14:38] <ubitux> just like ffmpeg, ffplay and ffserver
[14:39] <yns> okay :)
[14:39] <ubitux> btw, the above command will just work at format level, so it should be faster than looking into the codec level like with any -vf
[14:41] <yns> okay :) and what did you by if all frames are intra? or is that out of scope?
[14:42] <yns> ffprobe -v 0 -show_entries frame=key_frame -of flat input
[14:42] <yns> nothing really happens?
[14:44] <ubitux> remove the -v 0 and see if there is any error
[14:44] <ubitux> (replace "input" with your file of course)
[14:44] <ubitux> yns: all frames intra = all frames with type:I
[14:44] <yns> ah okay :)
[14:45] <yns> Invalid data found when processing input
[14:45] <yns> thats the error I get when remove -v 0
[14:45] <ubitux> is your input file even valid?
[14:45] <ubitux> (can you ffplay it?)
[14:45] <ubitux> (or even transcode it with ffmpeg)
[14:45] <yns> sorry :s
[14:46] <yns> it is invalid as hell :)
[14:46] <yns> and it works!
[14:47] <yns> Thanks a million!
[14:47] <yns> helped me big time!
[14:47] <ubitux> your welcome
[14:47] <yns> by the way: you know by any chance how to convert into photo jpeg?
[14:47] <ubitux> ?
[14:48] <yns> aka P-JPEG
[14:48] <yns> am I asking crazy stuff?
[14:48] <ubitux> i don't know
[14:49] <yns> okay thanks anyway :)
[14:49] <yns> anyone?
[14:49] <ubitux> never heard of p-jpeg
[14:49] <ubitux> any reference link?
[14:49] <yns> http://www.siggraph.org/education/materials/HyperGraph/video/codecs/JPEG.html
[14:54] <ubitux> i wonder how that differs from standard jpeg
[14:54] <yns> no clue
[14:55] <yns> I've read it should be even faster in scrubbing than MJPEG
[14:56] <blez> I have a slight problem with dshow
[14:57] <blez> ffmpeg -f dshow -audio_decice_number 0 -f flac -
[14:57] <blez> Output file #0 does not contain any stream
[14:58] <yns> can I convert to jpeg maybe? would that give me a playable movie?
[14:59] Action: blez pokes ubitux
[14:59] <blez> that's "device"
[15:00] <xlinkz0> why does the demuxing example call avformat_find_stream_info with NULL for the second argument?
[15:02] <ubitux> blez: i never used dshow
[15:03] <blez> most people haven't
[15:03] <ubitux> i didn't use windows since years as well
[15:03] <ubitux> ...so i can't help you
[17:00] <yns> anybody a clue how to encode to Photo JPEG? http://www.siggraph.org/education/materials/HyperGraph/video/codecs/JPEG.html
[17:39] <simonsimcity> Hi, I know that it's possible to add an additional audio-track to an existing video ... but is it possible to change the name of the track?
[17:40] <simonsimcity> In VLC for example I can then choose the track and it has a name ... On some VOB files I got it to be named after the language, but I can't see the option in ffmpeg's command ...
[17:40] <simonsimcity> An example I found for just adding audio: http://blog.noizeramp.com/2011/04/21/adding-audio-track-to-video-with-ffmpeg/
[17:46] <simonsimcity> *yess* I found it. "ffmpeg -i INPUT -metadata:s:a:1 language=eng OUTPUT", right?
[17:59] <simonsimcity> I found an option -alang ... what about that? Does it do the same?
[18:15] <simonsimcity> so ... the parameters alang, vlang and slang are removed? http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=35c5d7d1ef7b623779c860887f9ddbeb3265c71c
[18:17] <simonsimcity> It would be nice to update the documentation for metadata then. Currently you have a hint in there, that it's possible, but no hint, that the language should be a 3 letters ISO 639 code.
[19:01] <simonsimcity> It's nice to know that some container-formats do not support every metadata setting :) http://wiki.multimedia.cx/index.php?title=FFmpeg_Metadata
[19:01] <simonsimcity> Could this be linked here? http://ffmpeg.org/ffmpeg.html
[19:40] <xlinkz0> i remember putting a banana metadata key in mp4 so are you sure it's restricted?
[20:08] <ghost231> anyone here
[20:08] <ghost231> &
[20:08] <ghost231> ?
[20:10] <ghost231> are there snow and ffv1 encoders in ffmpeg last build?
[20:11] <ghost231> and how to use them?
[20:13] <ubitux> just like the others
[20:14] <ghost231> stop, snow still in pack?
[20:15] <ubitux> snow is still in ffmpeg if that's the question
[20:16] <ghost231> ok?
[20:16] <ghost231> ok, but tell me were can I get the versions of these codecs (in pack)
[20:18] <ubitux> i don't know what pack you are talking about
[20:19] <ghost231> the newest, lastest pack by zeranoe for win (32, 64 bit)
[20:20] <ubitux> these codecs are part of FFmpeg, so there version depends are basically the ffmpeg version
[20:20] <ubitux> but i don't even get your question anyway
[20:36] <ghost231> i want to know about snow and ffv1 -  what are the last versions, and what are in ffmpeg packs for windows?
[20:37] <ghost231> and how to use them (they are lossless codecs)
[20:42] <ghost231> I know about ffv1 - last version is 3, as I know, but not shure.. and snow is one version at all (no changelog) as I know...   And I want to get info about them here (where else?).
[20:43] <ghost231> i want to know about snow and ffv1 -  what are the last versions, and what are in ffmpeg packs for windows?
[20:43] <ghost231> and how to use them (they are lossless codecs)
[20:46] <ubitux> snow is not lossless afaik
[20:47] <ubitux> and both are ffmpeg codecs
[20:47] <ubitux> so look at the git log history of ffmpeg
[20:48] <ghost231> where are the info about versions of codecs?
[20:49] <ghost231> please give me the link (so look at the git log history of ffmpeg)
[20:51] <ubitux> ffmpeg.org
[20:51] <ubitux> go to download, clone the repository
[20:52] <ubitux> git log libavcodec/ffv1dec.c libavcodec/ffv1.c etc
[20:53] <ghost231> but i have windows (??)
[20:58] <ubitux> i can't help you. check on the web interface
[20:58] <ubitux> http://source.ffmpeg.org
[20:59] <ghost231> I;
[20:59] <ghost231> ok
[21:01] <beastd> ghost231: besides the source in ffmpeg, ffv1 has also an external spec written by its author
[21:11] <beastd> ghost231: here is the source document for the FFV1 spec: https://github.com/FFmpeg/FFV1
[21:32] <durandal_1707> ubitux: snow have lossless mode....
[21:33] <ubitux> durandal_1707: ah, i see, ok
[21:35] <durandal_1707> instead of dwt_97 for lossless and qscale = 0 you use dwt_53
[22:00] <Zarx> x264 has an option to set the SAR to stretch the video on playback. What is the equivalent option in ffmpeg?
[22:01] <durandal_1707> -vf setsar ?
[22:01] <durandal_1707> anyway, you can pass all x264 options to libx264 encoder
[22:02] <Zarx> oh, i forgot i can do that
[22:02] <Zarx> ill do it that way
[22:32] <Zarx> hmmm, i cant quite figure out the syntax for this
[22:33] <Zarx> x264's sar option takes a colon in it
[22:33] <Zarx> but the colon is also used to separate options
[22:47] <shadowing> does anyone have experience merging ts files?
[22:48] <shadowing> I've tried running "ffmpeg -f concat -i ts/dvr -async 1 -c copy concat.mpeg"
[22:48] <shadowing> where ts/dvr is a text file for the concat protocol
[22:48] <shadowing> I keep on getting this error spouted out
[22:48] <shadowing> [h264 @ 0x3080200] non-existing SPS 32 referenced in buffering period
[22:49] <shadowing> [h264 @ 0x3080200] non-existing PPS referenced
[22:49] <shadowing> [h264 @ 0x3080200] non-existing SPS 32 referenced in buffering period
[22:49] <shadowing> [h264 @ 0x3080200] non-existing PPS 0 referenced
[22:49] <shadowing> [h264 @ 0x3080200] decode_slice_header error
[22:49] <shadowing> [h264 @ 0x3080200] no frame!
[23:43] <Anaphaxeton> i was expecting some mpeg-ps format but it doesnt exist
[23:44] <Anaphaxeton> i used mpeg and the result was 80 instead of 3 streams...
[23:47] <Anaphaxeton> what i really want is an SD-video container
[23:54] <Anaphaxeton> the mpeg-ps wont fit pcm btw
[23:54] <Anaphaxeton> why is that?
[23:58] <Mavrik> probably because it doesnt carry stream information
[23:58] <Mavrik> and expects streams to do that
[23:59] <Mavrik> also, I doubt theres a use case for raw audio streaming
[00:00] --- Sat Apr 27 2013