[Ffmpeg-devel-irc] ffmpeg.log.20130630

[burek]

[02:12] <flok420> hi. can I use ffmpeg in a pipe-construction (reading from stdin) and then resample a raw audio-stream from samplerate x to samplerate y? (emitting it to stdout so that I can feed it to an other program)
[02:18] <klaxa> flok420: does ffmpeg -i - not work?
[02:19] <klaxa> output to stdout could be achieved with ffmpeg <input options> -f <container> pipe:
[02:21] <flok420> klaxa: yes -i indeed works but I could not find how to both set the input and the resampled output samplerate
[02:22] <klaxa> ah raw audio stream... try: ffmpeg -f rawaudio -ar <in-sample-rate> -i - -ar <out-sample-rate> -f rawaudio pipe:
[02:23] <WolfpupL> relaxed can you pm me for a min
[02:25] <mark4o> -f s16le (or whatever format you are using)
[02:26] <klaxa> ah yes would probably be better tehe
[02:27] <flok420> yes, this seems to work: ffmpeg -f u16be -ar 44100 -i - -ar 22050 -f u16be -
[02:27] <flok420> thanks!
[02:27] <flok420> (to my surprise the input was nog 28kHz but 44.1kHz)
[02:27] <mark4o> your raw audio is u16be?  weird, never seen that
[02:28] <flok420> not entirely sure if it is signed or unsigned, am testing that now
[02:28] <mark4o> s16 is a lot more common
[08:21] <deadman36g> identify lastride1
[08:23] <deadman36g> can someone help me with screenshots? I need somescreets that are DAR but the ones I make end up SAR
[08:25] <sdl240> deadman36g, ffmpeg -ss 00:10:00 -i /foo/video.mkv -y -vframes 1 -vf "scale='max(sar,1)*iw':'max(1/sar,1)*ih'" /bar/screenshot.png
[08:34] <deadman36g> sdl240, I tried the command you gave at first and it returned /foo/Fall Of The Eagles.mkv: No such file or directory so then I did the command with /foo/ removed and I got [image2 @ 0x13cdf00] Could not open file : /bar/screenshot.pngbitrate=  -0.0kbits/s dup=0 drop=74
[08:34] <deadman36g> av_interleaved_write_frame(): Input/output error
[08:44] <sdl240> ffmpeg -ss 00:10:00 -i videofile.avi -y -vframes 1 -vf "scale='max(sar,1)*iw':'max(1/sar,1)*ih'" screenshot.png
[08:46] <deadman36g> thank you very much sdl240 that does the trick :O
[08:46] <sdl240> ffmpeg -ss 00:10:00 -i Fall Of The Eagle.avi -y -vframes 1 -vf "scale='max(sar,1)*iw':'max(1/sar,1)*ih'" /your.individual.path/screenie.png
[08:46] <sdl240> ok, fine
[12:32] <vulture-> using a custom io context I get this at the end of my streams: [mov,mp4,m4a,3gp,3g2,mj2 @ 000000002246bea0] stream 1, offset 0x5dc48c3: partial file
[12:32] <vulture-> is there some common pitfall I'm missing? ;d
[12:33] <vulture-> I do seem to be writing the whole stream and then sending EOF
[13:26] <luc4_mac> Hi! I'm using ffmpeg built using macports on Mac OS. I noticed that ffmpeg reports "using cpu capabilities: none!". Is this correct? Isn't there any CPU capability that can be used by ffmpeg to increase performance on a mac, like sse?
[13:31] <relaxed> luc4_mac: try the static build found on the download page.
[13:32] <Mavrik> yp
[13:32] <Mavrik> luc4_mac, afaik macports script disables all ASM optimizations on mac for some stupid reason
[13:33] <luc4_mac> Mavrik: oh thanks then! I'll try to download the built version.
[13:35] <cbsrobot>  luc4_mac or use homebrew
[13:44] <luc4_mac> Mavrik: in fact it is far better now :-) thanks!
[16:36] <gtlz> greetings
[16:37] <gtlz> i'm trying to do some vfs before sending my /dev/video device off to ffserver... however, no VFs are being obeyed.... if i create a sample output.mp4, it works perfectly. all the vfs work right.  but, if i view the ffserver stream, there is no vf (crop & text)
[16:38] <gtlz> what gives? is this a bug or expected behaviro?  if it's the latter, what's the best way to cope?  outputting to a temporary file/pipe and using a separate ffmpeg command to take that input and feed to cam?
[16:38] <gtlz> s/to cam/to ffserver/
[17:13] <gtlz> yeah the vfs are completely ignored if the output is an ffm rather than an mp4
[17:19] <shad0VV> hi
[17:21] <shad0VV> i am trying to play a network stream, its protocol is mms... but ffplay cannot find a protocol to open url
[18:42] <rabiescat99> Hello. I converted a 4K video clip and it turned from 6.95 GB to 79.6 MB with no loss of quality. However, it changed the colors slightly so they are a little darker. Comparison: http://i.imgur.com/gUxZHYF.jpg Command used: "ffmpeg.exe -i test.avi -q:a 0 -q:v 0 test_out.avi" What gives?
[18:55] <rabiescat99> Anyone alive in here?
[19:02] <durandal_1707> what colors are in 4k video?
[19:03] <rabiescat99> Huh?
[19:04] <durandal_1707> usually colors changes when doing rgb->yuv
[19:04] <durandal_1707> so what exactly you are doing is still unknown
[19:04] <rabiescat99> Who said anything about doing RGB => YUV?
[19:05] <rabiescat99> I gave you the command used.
[19:05] <durandal_1707> useless
[19:05] <rabiescat99> Useless?
[19:05] <sacarasc> But not the complete console output.
[19:09] <rabiescat99> OK, let's say it DID convert from RGB to YUV.
[19:09] <rabiescat99> Why did that fail?
[19:09] <rabiescat99> Bug in ffmpeg?
[19:10] <durandal_1707> it did not fail, it just may not be possible, or wrong coeff was used when doing colorspace conversion
[19:11] <durandal_1707> eg, rgb->yuv  is not same as rgb->yuvj
[19:13] <rabiescat99> May not be possible? Hrm...
[19:13] <rabiescat99> Are you saying YUV has many more colors or something?
[19:13] <rabiescat99> I just assumed it was yet another redundant standard.
[19:14] <rabiescat99> I have no idea what "yuvj" is.
[19:14] <durandal_1707> well if colors are darked, than inspect and tweak colorspace conversion which is usually done by swscale
[19:15] <durandal_1707> because you did not showed full, uncut console output, i cant say more
[19:34] <rabiescat99> Um...
[19:34] <rabiescat99> You saw the damn command.
[19:35] <rabiescat99> What about it is potentially wrong?
[19:35] <rabiescat99> It is not an actual problem, but just me wanting to know why it happened.
[20:27] <brontosaurusrex> rabiescat99, the full output will give some clues what color-conversions were done in the process
[20:27] <brontosaurusrex> if any
[20:39] <rabiescat99> I don't have any such thing.
[20:39] <rabiescat99> Why is it a problem?
[20:40] <rabiescat99> If it's possible to convert, then it's a bug.
[20:43] <brontosaurusrex> then report the bug
[20:49] <brontosaurusrex> unless you are using a version from 2001 that is...
[21:08] <rabiescat99> No old software herer.
[21:08] <rabiescat99> -r
[22:23] <LedM2> Hello? :p Anone can explain , why the MP3 headers on 64 bit are broken?.
[22:26] <LedM2> I dont have commands on 64Bit, ate least i have Computr 64Bit, these versions are from the web server, and all only i ever has used is the 32 Bit withou any trouble, these Damaged Mp3 comes with an illegal header , its impossible to recovery ,i need to reconvert on 32 Bit to normalize.
[22:27] <durandal11707> what ffmpeg version?
[22:27] <durandal11707> can you give link to one such file?
[22:28] <durandal11707> what os?
[22:28] <durandal11707> lot of relevant info is missing for diagnostic
[22:29] <LedM2> Well, on the id3 tagger that i use , with much difficulty, i discoverede the versio, Laf 8.5 OR Laf 8.6 is this...
[22:30] <durandal11707> hmm, i can't recognise that
[22:31] <LedM2> Ate least me, is a web server, TWO WEB SERVER, on 64Bit, .net
[22:32] <durandal11707> WHAT EXACT VERSION
[22:33] <LedM2> I am on mobile, but tomorrow, i will upload a MP3 , to opera server, and give a link, its oincredible, my phone and sd card corrupted by these Damaged headers, and the tagger dont fix...
[22:34] <LedM2> I dont know the versio exactly, say Laf 8.5...    and comes with supertagging property, :-|
[22:35] <durandal11707> 0.8.5 ?
[22:36] <LedM2> yes.
[22:36] <LedM2> On 64Bit web server .net
[22:36] <durandal11707> that is very old, and may not be FFmpeg at all but Libav fork
[22:37] <LedM2> No, tomorrow i will take ta versio exactly, and the link to Damaged Mp3...
[22:37] <LedM2> It says the version, Laf8.5.......
[22:39] <LedM2> i was created an MP3 aac+ based on 32BIT, and works very well, but these web servers, do the same formule on 64Bit, but missing Legal Headers! impossible to fix without reconverting the fiel on 32bit
[22:39] <BernzSed> I have a long video. There's about 10 seconds of video somewhere in the middle that I would like to replace with video from a different 10-second video file, but without replacing the audio. How can I do this?
[22:40] <BernzSed> wait, why is my text underlined?
[22:40] <BernzSed> &nvm?
[22:41] <BernzSed> ok, that was odd. Anyone know the answer to my question?
[22:41] <durandal11707> yes, i can answer your question.
[22:43] <LedM2> The id3tagger on java says : javaIOException: IndexOutOfBounds 0; the phone music player, its plays someday yes, somedays ?The file disappears! ha! the magic of missing header! :-|
[22:43] <BernzSed> Thanks. How?
[22:45] <LedM2> BernSzeg: You have to separate entre audio of the video, and convert your entire video mutted,
[22:46] <LedM2> Bernszed: sorry, when you have the mutted video, you cut untill the time tha you dont want tha appears, and replce with the new part, but the time have to be precisious!
[22:47] <BernzSed> I was hoping there would be a better way than that :(
[22:47] <durandal11707> BernzSed: if you need exact precision cut, you are probably forced to transcode video
[22:48] <BernzSed> Transcode?
[22:48] <LedM2> You will merge the three parts turning the video in one, and finally you re-merge the original audio... is this :-P
[22:49] <durandal11707> BernzSed: reencode video
[22:50] <BernzSed> I tried doing what LedM2 said I should do a while back, but ffmpeg couldn't consistently get the timings right. So either I was doing something wrong, or it was a bug that has hopefully since been fixed
[22:50] <BernzSed> Thanks
[22:50] <LedM2> I only see doing paciently, ffmpeg is hard to manipulate, if have transcode, it will be hardest time to encounter these commands(if exists)...
[22:51] <LedM2> i will be back later...
[22:52] <durandal11707> BernzSed: there are filters that can cut/trim video/audio and concat demuxer
[22:52] <BernzSed> How do I use them?
[22:53] <durandal11707> BernzSed: but i think same can be done just with filters, but that will force reencoding audio, wich may be needed anyway
[22:53] <durandal11707> BernzSed: this is channel and not consulting
[22:54] <durandal11707> there is filters documentation
[22:57] <durandal11707> anyway i think that doing something like that should be currently support by our libavfilter, even if there are no examples
[22:57] <durandal11707> basicaly you take one audio input and two video inputs, and outputs one audio and one vide
[22:58] <durandal11707> *video
[23:01] <durandal11707> and you are probably looking for interleave filters
[23:02] <durandal11707> so you need latest ffmpeg version from git
[23:04] <durandal11707> though it would be more logical that select filter does that: picking single frame from 2 inputs
[23:19] <BernzSed> Thanks, I might check that out
[00:00] --- Mon Jul  1 2013