[Ffmpeg-devel-irc] ffmpeg.log.20140930

[burek]

[00:00] <c_14> You'd have to programatically find the resolution
[00:01] <spectralsun> Yeah, I'm able to do that
[00:02] <spectralsun> Here is what I would be doing:
[00:03] <spectralsun> splash.mp4 (1280x720) + 6 second clip of video (480x360 or 568x320) + gauge_preview.mp4 (1280x620)
[00:04] <spectralsun> I'll know the resolution of the 6 second video clip (sometimes the videos are less than 6 seconds, in that case I use the whole video)
[00:07] <spectralsun> I figured it out
[00:07] <spectralsun>  '[0:0]scale=568:320,setsar=1[v1];[1:0]scale=iw*sar:ih,setsar=1[v2]; [v1] [0:1] [v2] [1:1] concat=n=2:v=1:a=1 [v] [a]'
[00:08] <spectralsun> w00t!
[00:09] <spectralsun> c_14++ c_14++ c_14++
[01:17] <voip1__> guys, any help regarding my problem ?
[01:19] <c_14> I can't see anything that wrong with the command, can you play the input stream with ffplay without problems? Can you try playing the output with ffplay instead of vlc? Maybe try saving the stream to a file and try playing that?
[01:22] <voip1__> c_14, fflpayer is windows player
[01:22] <voip1__> ?
[01:22] <c_14> ffplay is the sdl player included with ffmpeg
[01:23] <voip1__> so i shld play from linux
[01:23] <voip1__> ?
[01:24] <c_14> If you can. I just want to see if it might just be vlc having the issues and not the stream itself.
[01:28] <voip1__> c_14, i just tryed from linux proppt ffplay stream URL, it plays , but i dont konw abaut quality
[01:29] <voip1__> can you please chek stream quality from source: http://176.56.178.36:8042/9275617894-ru-sts ?
[01:30] <c_14> stream plays fine here
[01:31] <voip1__> intersting, in which country you are ?
[01:31] <c_14> Germany.
[01:32] <voip1__> oo, i am from USA, my ping is ~140, but i dont tinks its problem, because it TCP
[01:33] <voip1__> also y have 3600 fbufer
[01:36] <JodaZ> playing fine in vlc
[03:18] <voip1__> how can chek inbound videostream rate ? ffmpeg - streamSRC gaves me :  Duration: N/A, start: 82119.524033, bitrate: 141 kb/s
[03:18] <voip1__>   Program 1
[03:18] <voip1__>     Stream #0:0[0x44](rus): Audio: aac ([15][0][0][0] / 0x000F), 48000 Hz, stereo, fltp, 141 kb/s
[03:18] <voip1__>     Stream #0:1[0x45]: Video: h264 (Constrained Baseline) ([27][0][0][0] / 0x001B), yuv420p, 720x576 [SAR 16:15 DAR 4:3], 25 fps, 25 tbr, 90k tbn, 50 tbc
[03:18] <voip1__> where is video bitrate ?
[04:29] <spectralsun> Can anyone help me extract the audio from this file: http://spectralsun.com/extract_audio.mov
[04:29] <spectralsun> or help me figure out how it has been improperly made
[04:29] <spectralsun> I think it came off an iphone 5s
[04:31] <JodaZ> are you geoff gilbert ? :)
[04:34] <spectralsun> negative
[04:34] <spectralsun> my friend sent me that file
[04:34] <spectralsun> asking if i could figure it out
[04:34] <JodaZ> file looks broken/truncated
[04:34] <spectralsun> yeah
[04:34] <JodaZ> low entropy data too
[04:35] <spectralsun> JodaZ: stream 1, error opening alias: path='/Users/', dir='Roll 1', filename='IMG_0181.MOV', volume='Macintosh HD', nlvl_from=3, nlvl_to=3
[04:35] <spectralsun> is the file some kind of link
[04:35] <spectralsun> ?
[04:36] <JodaZ> possibly
[04:36] <JodaZ> lol
[04:36] <spectralsun> sigh
[04:36] <JodaZ> thats an interesting format
[04:40] <Peter_Occ> I am capturing videos from a camera but when I play them back, they play too fast and end too soon. I can adjust the speed and  length of a video as I capture them, but they only come out right if I set the input rate to 10 and the output rate at 5. I figured this out by trial and error. How would you figure this out logically? Why does the input rate have to be different from the output...
[04:41] <Peter_Occ> ...rate in order for it play at the correct speed.
[04:47] <JodaZ> spectralsun, definitively a link
[04:57] <kryo_> hey, this might be the wrong place to ask...
[04:58] <kryo_> but i was wondering if anyone knows how to configure ffmpeg to use a socks5 proxy
[04:59] <kryo_> the idea is to capture the desktop and stream via proxy to an rtmp url
[05:01] <JodaZ> why
[05:02] <kryo_> why not?
[05:31] <MikeJoel> odd - If I make an mp4 directly from images it will not play in windows media player. If I make an avi and then convert that to an mp4 it works fine.
[05:47] <ZZRMike> I'm trying to convert .flv to mp4 to be used as html5 video, but using "ffmpeg -i video.flv video.mp4" the video won't play in a browser, I can hear audio but there's no video. Is there a flag I'm missing?
[06:04] <kryo_> ZZRMike: you might need -codec copy
[06:04] <ZZRMike> kryo_ thanks, I just used that. But I'm pretty sure it was a file permissions issue.
[06:04] <kryo_> o_O
[06:05] <kryo_> that sounds impossible... audio and video are not seperate files :P
[09:20] <brontosaurusrex> how would i go from 24 to 25 speed up correctly? (including audio speedup)
[09:22] <brontosaurusrex> same as this basically; http://paste.debian.net/plain/123735
[09:22] <brontosaurusrex> (thats avisynth)
[09:40] <chuckleplant> Howdy
[09:43] <chuckleplant> I'm using ffmpeg source to encode MPEG4 video. There aren't plenty of examples that use source rather than the built ffmpeg application. I want to reduce the bitrate but changing the encoding settings is not working very well
[09:44] <chuckleplant> Does anyone know about MPEG4 resources or guides that I can take a look at¿
[09:54] <brontosaurusrex> chuckleplant: x264?
[09:54] <brontosaurusrex> chuckleplant: https://trac.ffmpeg.org/wiki/Encode/H.264
[10:20] <chuckleplant> @brontosaurusrex Can't use x264 as it's GPL, I need LGPL
[10:25] <JodaZ> chuckleplant, vpx?
[10:42] <chuckleplant> JodaZ, Isn't that only for webM encoding?
[10:43] <JodaZ> well, yes (you can put it in other containers, but thats not the point)
[10:46] <chuckleplant> JodaZ, I need either H264, MPEG4 or MJPEG.
[10:46] <chuckleplant> JodaZ, thanks btw
[10:47] <chuckleplant> My implementation works well with MPEG4 but I can't control the settings. My next bet would be to switch to MJPEG but I wanted to prevent it if possible
[10:48] <JodaZ> well, what exactly did you try to reduce the bitrate?
[10:53] <Wessie> Is it documented anywhere what is still legal when your AVIOContext has no seek callback. The AVFormat probing code for example assumes seek is available and ignores the returned error if any, and I'm wondering if I should try and always implement seek instead
[10:59] <chuckleplant> Manually reducing the bit_rate parameter from the AVCodecContext struct. But this does not seem to work, do I have to change the quantization parameters?
[11:00] <JodaZ> what exactly is not working with it?
[11:01] <chuckleplant> Bitrate isn't reduced, result is the same whatever bitrate is set
[11:03] <JodaZ> chuckleplant, also set rc_min_rate and rc_max_rate
[11:06] <chuckleplant> chuckleplant, trying it out !
[11:07] <JodaZ> :)
[11:13] <chuckleplant> Works!! I knew it had to be a small detail
[11:13] <chuckleplant> thanks JodaZ
[11:14] <chuckleplant> Any recommendation for quality tweaks in mpeg4?
[11:15] <timothy> chuckleplant: for youtube?
[11:15] <chuckleplant> no, it's for streaming over rtsp, I need a balanced quality and bitrate, with 4Mbps I'm getting really poor quality
[11:16] <timothy> 1080p?
[11:16] <chuckleplant> Yes, that would be the maximum res, but could be less
[11:21] <brontosaurusrex> avc could look decent with 4 Mbps, mpeg4 will not for most video material
[11:22] <brontosaurusrex> and if you have a balanced bitrate, that means the quality will be variable
[11:22] <brontosaurusrex> simplified
[11:26] <chuckleplant> brontosaurusrex, by avc you mean h264?
[11:26] <brontosaurusrex> yes
[11:27] <chuckleplant> I need a non-GPL implementation though
[11:33] <brontosaurusrex> there is none afaik, unless you get a commercial license
[11:33] <JodaZ> heh, cisco is giving one away for free
[11:34] <brontosaurusrex> JodaZ: isnt that decoder only?
[11:34] <JodaZ> hmm
[11:34] <JodaZ> i thought voip was like their reason for it; and asumed that'd have to include both
[11:34] <JodaZ> but let me check
[11:36] <chuckleplant> :O
[11:37] <chuckleplant> that's awesome news
[11:39] <chuckleplant> Will it be included into ffmpeg? Or is BSD non compatible?
[11:40] <JodaZ> brontosaurusrex, the source they compile it from has both de and encoder, i can't find any definitive assertions if their free license blob has both too tho
[11:40] <brontosaurusrex>  JodaZ should have imho
[11:40] <JodaZ> brontosaurusrex, well, you questioned it
[11:41] <brontosaurusrex> yes, if they are both in existance i would imagine they would license both the same
[11:41] <JodaZ> chuckleplant, for licensing reasons (cisco pays the licensing costs for it for you is what basically happens) it has to be separate
[11:41] <brontosaurusrex> now the only question is the quality of the encoder
[11:41] <JodaZ> brontosaurusrex, they open sourced it now too
[11:43] <bencoh> afaik it as supports for low levels/profiles only
[11:44] <bencoh> has*
[11:44] <bencoh> Q. Which profiles of H.264 will be supported?
[11:44] <bencoh> A: The initial code has the baseline profile. We look forward to working with the open source community to add high profile and others.
[11:44] <JodaZ> bencoh, still maybe better than mpeg4 (also, why IS mpeg4 license free as chuckleplant seems to think?)
[11:46] <bencoh> I guess he means free as in "I can link it with closed-source/non-gpl software"
[11:46] <bencoh> not as in "royalty/patent-free"
[11:46] <JodaZ> oh, right, he is linking directly
[11:47] <bencoh> and ... yeah I guess openh264 cant be worst than mpeg4part2
[11:48] <chuckleplant> JodaZ, afaik mpeg4-2 is LGPL and I can link to it right?
[11:48] <chuckleplant> bencoh, yep that's what I though.. will give openh264 a try definitely
[11:48] <JodaZ> i think you'd still have to pay the mpeg-la for every unit sold?
[11:48] <bencoh> yup, you're still supposed to pay mpeg-la
[11:48] <brontosaurusrex> if its baseline only, its probably worse than mpeg4 anyway
[11:49] <bencoh> someone should bench it :p
[11:49] <JodaZ> indeed
[11:49] <bencoh> but part2 has no deblock loop
[11:49] <JodaZ> someone should do stuff
[11:49] <brontosaurusrex> yeah someone
[11:50] <bencoh> definitely not me :D
[11:50] <JodaZ> bencoh, if that info is from the initial announcement maybe its outdated by now, they open souced it and its on github
[11:51] <bencoh> JodaZ: I see nothing in commitlog
[11:51] <JodaZ> myeah
[11:51] <bencoh> I suspect we'd see a big "\o/" or something ;)
[11:53] <chuckleplant> so let me get it straight, mpeg4-2 is not license free under ffmpeg LGPL 2.1 builds? Where can I find more info about this?
[11:55] <JodaZ> chuckleplant, http://www.mpegla.com/main/programs/M4V/Pages/Intro.aspx
[11:59] <JodaZ> chuckleplant, bloodsuckers aint they? ;)
[11:59] <chuckleplant> JodaZ, you are the man
[12:06] <JodaZ> chuckleplant?
[12:06] <JodaZ> hmm, actually, how do things like the obs project get away with it
[12:06] <JodaZ> any idea, bencoh?
[12:07] <chuckleplant> back to the mpeg4 encoding via source, what should be the value of rc_buffer_size ?
[12:07] <chuckleplant> Thanks for the license info, I'll read it through
[12:14] <JodaZ> chuckleplant, i think you can leave most of that stuff alone? otherwise you can just google and look how other people set it
[12:19] <brontosaurusrex> any good explanation on how atempo filter is actually working?
[12:21] <brontosaurusrex> iam trying to get 23,976>25fps conversion, so far: ffmpeg -r "25" -i "in.mov" -filter:a "atempo=1.0427" "out.mov"
[12:35] <chuckleplant> hi again, sorry but I can't find guidelines on the rc_ parameters. And my stream fails if I don't set at least the rc_buffer_size.. but it crashes after a while. I get this error "ffmpeg error evaluating rc_eq "
[15:18] <chuckleplant> Best to all!
[16:00] <Peter_Occ> I am capturing videos from a camera but when I play them back, they play too fast and end too soon. I can adjust the speed and length of a video as I capture them, but they only come out right if I set the input rate to 10 and the output rate at 5. I figured this out by trial and error. How would you figure this out logically? Why does the input rate have to be different from the output rate...
[16:00] <Peter_Occ> ...in order for it play at the correct speed.
[16:01] <c_14> Have you tried not setting the output rate and only the input rate?
[16:09] <Peter_Occ> Yes, it makes them both the same and the video plays too fast
[16:09] <c_14> v4l2?
[16:09] <Peter_Occ> just a minute
[16:19] <Peter_Occ> The version is ffmpeg version N-60997-g983c7f4
[16:21] <c_14> What I was asking was whether the input device is a video4linux2 device.
[16:22] <Peter_Occ> It a DLink camera, so no I don't think so.
[16:40] <Peter_Occ> I'm trying to get that together ffloger.
[16:40] <Peter_Occ> What does 	!pb mean?
[16:40] <c_14> fflogger is a bot, !pb is the command I used to trigger him
[16:49] <Peter_Occ> Here is the paste bin. I'm using a bash script, the script is at the top
[16:49] <Peter_Occ> http://hastebin.com/hazufijabu.vhdl
[16:51] <c_14> Can you try running ffprobe on the http url?
[17:05] <Peter_Occ> I'm having trouble using ffprobe. Here is what I have done. What am I doing wrong?
[17:05] <Peter_Occ> http://hastebin.com/ecobihenaw.rb
[17:06] <c_14> Try with -analyzeduration 1G -probesize 1G (1G might be a bit excessive, at least for a stream, but I want to see if it changes anything)
[17:07] <sacarasc> Cameras like that just create a web page which puts a JPG there every x seconds...
[17:07] <sacarasc> It's not a stream as much as a CGI script.
[17:10] <Peter_Occ> I still get the same "could not find" http://hastebin.com/ajaqabomer.rb
[17:14] <c_14> Hmm, ok. Back to your ffmpeg command, if you leave off the -r 10, what does the input framerate get detected as?
[17:16] <Peter_Occ> How would I tell? If I leave off both the input and the output frame rate, the frame rate of the video is 25. Does that tell you?
[17:17] <c_14> On your first pastebin, line 28 says 10fps, if you leave off the -r what does it say?
[17:17] <c_14> The Stream #0:0 line
[17:19] <Akagi201> If I don't enable --enable-librtmp during I build ffmpeg from source, what I will lose about RTMP
[17:22] <c_14> rtmp muxing and demuxing support afaik
[17:24] <c_14> ie, reading from rtmp{,s,e} streams and outputting rtmp{,s,e} streams
[17:25] <BtbN> so basicaly, you loose rtmp support..
[17:25] <c_14> pmuch
[17:26] <Peter_Occ> If I leave off the -r 10 the output to the log file is a bunch of machine language. I don't understand why that would make any difference.
[17:27] <c_14> Machine language?
[17:43] <Akagi201> c_14: but ffmpeg itself already some rtmp feature in the libavformat right?
[17:43] <Peter_Occ> Well I don't know what it really is, but something is causing the log file to have different format. I can't copy and paste it and when I uploaded the file the serve won't let you view it, but you can download it and look at it in a programmers editor. http://geotonics.com/test/ffmpeg.log
[17:44] <Alina-malina> frame=    0 fps=0.0 q=-1.0 Lsize=      10kB time=00:00:00.00 bitrate=N/Avideo:0kB audio:0kB subtitle:0kB other streams:0kB global headers:0kB muxing overhead: unknown
[17:44] <Alina-malina> C:\ffmpeg\bin>ffmpeg -i 1.avi -vcodec copy -acodec copy -ss 12:15:00 -t 12:27:00
[17:44] <Alina-malina>  output3.avi
[17:44] <Alina-malina> :(
[17:44] <Alina-malina> it is been an hour i cant cut a piece from .avi file
[17:44] <Alina-malina> please help
[17:45] <c_14> Akagi201: I think those are mainly wrappers.
[17:45] <c_14> Peter_Occ: I'm getting a 403 Forbidden
[17:46] <Alina-malina> so what is this error means?
[17:46] <Peter_Occ> Yes you can't look at it, but you can click on the link and save the file.
[17:47] <c_14> Peter_Occ: the HTTP server is giving me an access denied. ie I can't download it
[17:47] <c_14> Alina-malina: I do not see an error in the output you pasted.
[17:47] <Alina-malina> headers:0kB muxing overhead: unknown  and exits
[17:48] <Alina-malina> http://pastebin.com/YCGqV3n8
[17:49] <Peter_Occ> Yeah I see what you mean. I have no idea what this log file is or how to display it.
[17:53] <c_14> Is the file smaller than 50MiB?
[17:54] <Alina-malina> c_14, http://pastebin.com/YCGqV3n8  see the error here
[17:55] <Alina-malina> erm
[17:55] <Alina-malina> hold on
[17:55] <c_14> Peter_Occ: if it is, can you try uploading it to pomf.se ? or you can dcc it to me or something
[17:56] <Alina-malina> c_14, http://pastebin.com/p0Ugd2uc  here
[17:58] <c_14> Alina-malina: can you play the input with ffplay, vlc or something?
[18:00] <Alina-malina> yes i can watch it with vlc without problems
[18:00] <c_14> Try using -ss and -t as input options?
[18:03] <Alina-malina> C:\ffmpeg\bin>ffmpeg -i 1.avi -vcodec copy -acodec copy -ss 12:15:00 -t 12:27:00 output.avi   this is my command
[18:04] <c_14> ye, try C:\ffmpeg\bin>ffmpeg -ss 12:15:00 -t 12:27:00 -i 1.avi -c copy output.avi
[18:05] <Alina-malina> c_14, http://pastebin.com/VEw83JCk
[18:06] <c_14> That looks like it worked.
[18:07] <c_14> well
[18:07] <c_14> If the video is 267KiB
[18:07] <Alina-malina> erm
[18:07] <Alina-malina> yes
[18:07] <Alina-malina> it worked
[18:07] <Alina-malina> heh
[18:07] <Alina-malina> but the timing
[18:07] <Alina-malina> it cut something different
[18:07] <Alina-malina> then it should
[18:07] <Alina-malina> hmmmm
[18:08] <relaxed> -t goes after the input
[18:08] <Alina-malina> [NULL @ 02e9c120] Unable to find a suitable output format for '12:27:00'
[18:08] <Alina-malina> 12:27:00: Invalid argument
[18:08] <c_14> relaxed: it works before as well
[18:08] <Alina-malina> hmmm
[18:09] <c_14> Alina-malina: what are you trying to cut?
[18:09] <Alina-malina> ffmpeg -ss 12:15:00 -t 12:27:00 -i 1.avi -c copy output.avi  this worked, but it did not cut what i expect
[18:09] <Alina-malina> i want to cut a part of movie
[18:09] <c_14> From where to where/how long?
[18:09] <Alina-malina>  12:15:00 12:27:00
[18:09] <Alina-malina> from here to here
[18:09] <relaxed> -t is time to encode
[18:09] <Alina-malina> erm
[18:10] <Alina-malina> i dont get it
[18:10] <relaxed> -t 12
[18:10] <Alina-malina> :(
[18:11] <c_14> -t is duration
[18:11] <Alina-malina> yes but it still cut something different
[18:11] <c_14> entirely different or close?
[18:11] <Alina-malina> the end
[18:11] <Alina-malina> the total videio duration is 41:40
[18:12] <Alina-malina> o want to cut from here 12:15:00 to here 12:27:00
[18:12] <relaxed> 00:12:15
[18:12] <c_14> -ss 12:15:00 -t 12:00
[18:12] <Alina-malina> ah
[18:12] <relaxed> HH:MM:SS
[18:12] <c_14> eh, oh
[18:12] <c_14> right
[18:12] <Alina-malina> 12 seconds
[18:12] <Alina-malina> not 12 minutes
[18:12] <relaxed> your seek was wrong
[18:13] <relaxed> you want to seek to the 12:15 point and cut after 12 seconds, correct?
[18:13] <c_14> Did not even notice that, oh man.
[18:13] <Alina-malina> now it cuts everything before 00:12:15 and removed the sound:(((
[18:14] <Alina-malina>  want to cut only this   [12:15:00 to here 12:27:00]
[18:14] <relaxed> what are you even trying to do?
[18:14] <Alina-malina> eh
[18:14] <Alina-malina>  want to cut only this   [00:12:15:00 to here 00:12:27:00]
[18:14] <relaxed> as we dicussed, 12:15:00 means 12 hours and 15 minutes
[18:15] <Alina-malina> no just 12 seconds
[18:15] <Alina-malina>  want to cut only this   [00:12:15 to here 00:12:27]  like this
[18:15] <relaxed> ffmpeg -i input -ss 00:12:15 -t 12 -c copy output
[18:16] <Alina-malina> yes this is it, bu now the sound gone :-0/
[18:16] <Alina-malina> :-/
[18:17] <relaxed> use output.mkv
[18:18] <Alina-malina> heh yes it works:)
[18:18] <Alina-malina> but with .mp4 not mkv
[18:18] <Alina-malina> thank you very much
[18:35] <XHFHX> Hi there. I need an advice on a codec. It needs to be very fast and without much quality loss? Lossless x264 isn't an option and it's too slow with x264 -qp 1.
[18:35] <XHFHX> Filesize isn't that important as long as it doesn't hit 5gb/min or something like this
[18:36] <c_14> you can try ffv1, or huffyuv. Tried adjusting the preset with x264 to speed it up?
[18:38] <XHFHX> you mean with -preset ultrafast?
[18:39] <c_14> yep
[18:40] <XHFHX> only tried it with the lossless x264, but will try this out. Will make some now tests, but huffyuv seems nice
[18:50] <iive> I though x264 lossless is with -qp 0
[18:51] <iive> ffv1 is imho, slower than x264 -qp 0.
[18:52] <relaxed> ffvhuff is fast
[18:52] <iive> the next in speed is mjpeg, but that one is not lossless. huffyuv is fast, but is produces huge files.
[19:34] <voip_> Hello guys, how can I check inbound video stream rate ? "ffmpeg - i streamSRC" gave me :  Duration: N/A, start: 82119.524033, bitrate: 141 kb/s
[19:34] <voip_> <voip1__>   Program 1
[19:34] <voip_> <voip1__>     Stream #0:0[0x44](rus): Audio: aac ([15][0][0][0] / 0x000F), 48000 Hz, stereo, fltp, 141 kb/s
[19:34] <voip_> <voip1__>     Stream #0:1[0x45]: Video: h264 (Constrained Baseline) ([27][0][0][0] / 0x001B), yuv420p, 720x576 [SAR 16:15
[20:38] <voip_> ?
[21:42] <bjoe2k4> how to do lossless x265 encoding with ffmpeg?
[21:53] <kepstin-laptop> bjoe2k4: you could probably enable it via the -x265-params avoption, e.g. "-x265-params lossless=1" (completely untested by me, I don't have x265 installed)
[22:06] <active8> ffmpeg -codecs question: the 2nd column says h264 ( the format according to wikipedia, but I asked for -codecs, not -formats) and the 3rd column lists the names (? is that what they're called? ) and the encoders like (encoders: libx264 libx264rgb .) Now when I use one like -c:v libx264, it's because it's what was reccomended for what I was doing at the time. Is there a place I can determine the (most likely) best choice of a/v formats and codecs for a specif
[22:06] <active8> ic application? I have notes on things that work up to this point, but I'd like to learn more about this powerful awesome tool, ffmpeg and to be able to implement its features quickly without having to refer to a lot of separate articles. Of course, having figured out which codecs and formats to use, if I need to learn more about a certain codec or format's specifics, I think I know where to find that.
[22:12] <llogan> what?
[22:15] <active8> I suspect that either ffmpeg.org and/or trac. (?) pages, but still haven't found a table with a quick breakdown of what to use like: libx264 supports x bitrate and works with these framerates; or pcm24le to .wav and mp4 will contain a wav, mp3, or these other audio formats
[22:18] <active8> these TOCs / indexes tell me a name of a format or codec and then i have to chase down more info. where is the "if you want..." use " this" table / flowchart?
[22:19] <kepstin-laptop> active8: the only format that really needs such a table is aac audio, and that's at https://trac.ffmpeg.org/wiki/Encode/AAC ; all the other formats are either fairly esoteric or only have one encoder...
[22:20] <shevy> I got a weird problem
[22:20] <shevy> ffmpeg merge yields audio files (mp3) that will have this problem:
[22:20] <shevy>   mpg123 failed to reopen stream: Error reading the stream. (code 18)
[22:22] <shevy> hmm
[22:23] <llogan> shevy: you're using ffmpeg to make MP3 audio files and then mpg123 can't open these files?
[22:23] <shevy> the commandline I use is:
[22:23] <shevy>   ffmpeg -i concat:foo1.mp3\|foo2.mp3 -acodec copy output.mp3
[22:23] <shevy> (In short version, the real names are longer)
[22:23] <shevy> llogan, oh no, it is a bit different
[22:23] <shevy> I use a light ruby wrapper over ffmpeg to cut + merge audio files, mostly .mp3
[22:23] <shevy> a user complained that the file duration of the then-merged .mp3 file is reported incorrectly in vlc
[22:24] <shevy> e. g. 68 seconds (1:08) rather than 30 seconds
[22:24] <bjoe2k4> kepstin-laptop: thanks, worked
[22:24] <shevy> he sent me that file, and I played it in mplayer ... and mplayer reports that error above through mpg123
[22:25] <shevy> I guess basically I am trying to find out where the error happens :)
[22:25] <llogan> shevy: perhaps you should use the concat demuxer instead of the concat protocol
[22:25] <shevy> it worked fine for me... I am on linux though, he is on mac osx
[22:25] <shevy> hmm ok, have that noted down
[22:30] <active8> kepstin-laptop, i also have a bookmark for the trac.ffmpeg page on h.264 encoding, which is a nice resource. What I don't have is a way to decide what I want to use for different tasks. I could be doing a screencast (i have reasonable notes on that thanks to this channel and another tab or three open in a browserr session) I could have an mp4 webm flv, or 3gp video that I want to do different things with. I might want to break out the video and audio for edi
[22:30] <active8> ting and then sync and mpx into mp4 and flv and it just seems to take a lot of time to find a way from point A to point B or maybe the prob is that I just don't know the possibilities. I guess I'll survive. I've learned a fair bit and - enough to maybe luck out on trial and error with few errors
[22:31] <kepstin-laptop> active8: well, deciding which codec to use is more a matter of what the people who will be watching are able to play, more than anything else...
[22:31] <active8> like someone yesterday had to ask what was best for raw video, YUVxxx or something else. That could have easily been me. I have to look up notes and mkv just to know why i used it once and whether I want it again
[22:33] <active8> ok. I know i can find another page on youtube formats - some chart on meduim, flv, 480p 720p 1080p
[22:36] <llogan> the easiest thing to do is to probably just ask here for each case if you're having trouble
[22:38] <active8> maybe. thanks again to those who've helped in the past.
[22:59] <active8> kepstin-laptop, i think the first paragraph of that link you gave sheds a bit of light on my quandry. And it says that AAC succeeds mp3 in the mp4 container, which I've noticed that AAC pops up a lot in VLC when I look at the codec info. Can I additionally glean from that that for audio only, mp3 is popular and since my android records in AMR, then I might want to support that and for other platforms just look and see what's in use?
[23:03] <faiden> Hi I'm trying to upgrade my arch linux and ffmpeg-compat2 gives me a error "ERROR: freetype2 not found. I have freefont2-infinality installed any ideas how to fix this?
[23:03] <faiden> I mean freefont2-infinality installed
[23:10] <tempUser> Hello. I'm trying to pass the parameter "border_masking" while encoding a video in ffmpeg using libavcodec. I see in documentation that it is type float, but I don't know what the range of values are and what they mean. Can someone please explain that parameter? Thank you.
[23:26] <kepstin-laptop> tempUser: what codec are you encoding with?
[23:27] <kepstin-laptop> tempUser: basically, if you don't know what the -border_mask parameter means, you shouldn't use it (leave it as 0)
[23:29] <tempUser> kepstin-laptop: using h264
[23:30] <kepstin-laptop> yeah, since h264 encoding via x264 is an external library rather than a built-in codec, I don't think that option does anything anyways.
[23:31] <tempUser> kepstin-laptop: I want to experiment with this parameter. I know I have static video data at the border regions and I want to increase compression on the border.
[23:31] <kepstin-laptop> x264 handles static data perfectly fine on their own
[23:33] <tempUser> kepstin-laptop: Then maybe you can suggest an optimization to make a region of interest Less Lossy and the remaining region More Lossy.
[23:33] <kepstin-laptop> if by "more lossy" you mean "less detail / more blurry", then applying a quantizer offset would do that, yes
[23:34] <kepstin-laptop> but to do that with x264, you'd have to program using the x264 library directly - this lets you provide an array of quantizer offsets. I don't believe the functionality is available via libavcodec.
[23:36] <kepstin-laptop> if the section of the video just stays the same constantly, but you want it to be as sharp as the rest of the video, do nothing. The codec will handle that just fine on its own.
[23:36] <tempUser> kepstin-laptop: If border_masking is defined in avcodec.h, why wouldn't it be an effective parameter? See http://pastebin.com/b0ndFHpj
[23:37] <kepstin-laptop> tempUser: h264 encoding isn't implemented in libavcodec, it's done via the x264 external library.
[23:37] <kepstin-laptop> so not all libavcodec options are mapped to the external library.
[23:38] <tempUser> how do I find out if this option is mapped to the external library?
[23:41] <kepstin-laptop> hmm. I dunno if there's a list anywhere. You could read the code for the interface at http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavcodec/libx264.c;hb=HEAD - it's not all that long. Look for references to the members of 'avctx'
[23:41] <kepstin-laptop> hmm. I think there is a doc somewhere...
[23:42] <tempUser> kepstin-laptop: Let me explain my video source for you and then you'll understand what I'm trying to do. I have medical imaging data that has a static black border that is always in the same location. I have dynamic, medical imaging data in a small region of the video. I want high quality in the medical imaging region, and I want low quality/less sharp encoding in the border/static regions. I am concerned with processor consumption
[23:44] <kepstin-laptop> tempUser: you should use the libx264 api directly if you want that much control; then you can provide a map of quantizer offsets to control the relative quality of different areas of the frame.
[23:47] <tempUser> kepstin-laptop: thanks for your help
[23:47] <tempUser> cheers
[23:49] <kepstin-laptop> to answer the other part of your question, the range of the quantizer offsets is codec dependent, but in x264 I think you can use -51 to +51, where 0 means no change, higher values reduce quality.
[23:49] <kepstin-laptop> (if you're using 10-bit x264, I think the range might be different, not sure)
[23:52] <tempUser> kepstin-laptop: is that quantizer the same as the CRF, or are there different quantizers?
[23:52] <kepstin-laptop> crf isn't quantizer, although the numbers look similar.
[23:53] <kepstin-laptop> crf is just a number which, given a particular set of encoding settings, will create a file of roughly constant quality with a proportional size.
[23:59] <kepstin-laptop> the quantizer number basically indicates how much detail is preserved in a macroblock, where (in h264, other codecs vary) 0 is lossless, and higher numbers reduce the detail more and more until you get to the max (51 in 8-bit, 64 in 10-bit) which preserves no detail - you just get a flat block out, basically.
[00:00] --- Wed Oct  1 2014