[Ffmpeg-devel-irc] ffmpeg.log.20160327

burek burek021 at gmail.com
Mon Mar 28 02:05:01 CEST 2016


[00:01:37 CET] <_Timon> it started with ~2000kbps but then after 5 minutes it dropped down to 139kbps?
[00:01:41 CET] <_Timon> How does that work
[00:03:14 CET] <Mavrik> It adjusts according to complexity of the scene.
[00:03:19 CET] <Mavrik> To keep constant quality.
[00:03:40 CET] <Mavrik> If you have a static image then there's nothing to encode.
[00:07:59 CET] <_Timon> "ffmpeg -i 1.mp4 1_fix2.mp4 -vf "yadif=1:1,hqdn3d=2,format=yuv420p" -deinterlace -crf 23"
[00:08:15 CET] <_Timon> input file = 4:2:2 at L# output file is the same
[00:08:27 CET] <_Timon> seems like it's not catching the "format=yub420p" part?
[00:08:55 CET] <_Timon> yuv*
[00:09:31 CET] <_Timon> I think i'm double deinterlacing with that line aswell
[00:09:59 CET] <_Timon> yadif is also deinterlace it seems
[00:14:09 CET] <_Timon> neither format=yuv420p nor -pix_fmt yuv420p is changing the pixel format... im done for tonight, gn.
[01:02:35 CET] <krompus> hey all! I'm trying to record some screencasts, but the compressionis really high, and I could stand to have slightly larger filesize. I'm not sure what settings I should use. My bitmap fonts' colours look awful.
[01:03:10 CET] <krompus> I'm using Teiler, if it makes a difference
[01:03:24 CET] <krompus> https://github.com/carnager/teiler
[01:04:26 CET] <krompus> right now, the profile is set to encopts="-r 60 -vcodec libx264 -pix_fmt yuv420p -s $res -acodec libmp3lame"
[01:05:02 CET] <krompus> also, is there a way to get it to play nice with compton, or should I just disable it before? I get visual glitches.
[01:07:57 CET] <sfan5> i'd suggest adding a crf to that command line
[01:08:17 CET] <krompus> crf?
[01:09:23 CET] <sfan5> controls the quality x264 aims at
[01:09:29 CET] <krompus> ah
[01:09:40 CET] <krompus> example number for high quality?
[01:09:50 CET] <J_Darnley> default is 23
[01:10:02 CET] <sfan5> you have to try yourself
[01:10:02 CET] <krompus> higher = higher quality?
[01:10:04 CET] <J_Darnley> good/high quality is around 18
[01:10:12 CET] <sfan5> lower -> higher q.
[01:10:14 CET] <krompus> gotcha
[01:10:19 CET] <sfan5> 20 might be a good start
[01:10:28 CET] <krompus> yeah, i intend to tinker with it.
[01:15:00 CET] <krompus> sfan5: alright, it's much better, but the colours are still washed out
[01:15:22 CET] <krompus> probably because I'm using bitmap fonts; doesn't compress well
[01:15:43 CET] <krompus> considering that they are 1 pixel thick, lol
[01:16:05 CET] <sfan5> you could try -pix_fmt yuv444p
[01:16:20 CET] <sfan5> uses a higher resolution for some color channels, should look better
[01:17:10 CET] <krompus> sfan5: AH!
[01:17:12 CET] <J_Darnley> Better a bitmapped font rather than subpixel anti-aliasing.
[01:17:14 CET] <krompus> much better!
[01:17:41 CET] <krompus> yeah, i had have crf set to 0, lol
[01:17:52 CET] <krompus> ...i'll work it down a little bit.
[01:18:08 CET] <sfan5> 0 is lossless so it should look perfect
[01:18:11 CET] <J_Darnley> Although if the text isn't black/white then AA makes less difference.
[01:18:17 CET] <krompus> yeah, that's what I was testing
[01:18:33 CET] <krompus> the colours are still off a little bit. Light red looks orangey, for example
[01:18:39 CET] <krompus> it's acceptable though
[01:19:43 CET] <J_Darnley> ooh
[01:20:14 CET] <J_Darnley> now it sounds like you need to read about colour matricies
[01:20:49 CET] <J_Darnley> Video is a never ending rabbit hole if you actually care about quality
[01:22:19 CET] <krompus> haha
[01:22:22 CET] <krompus> uh oh
[01:22:40 CET] <krompus> yeah, I've already got one leg in the hole
[01:23:01 CET] <krompus> I've already switched to FLAC and BrRips
[01:23:34 CET] <krompus> and I converted my little brother who used to watch shitty 700MB movies
[01:23:47 CET] <krompus> :P
[01:24:20 CET] <krompus> He kept telling me he didn't care, and then I showed him. His reaction: "OHHHHHHHHHH!"
[01:25:14 CET] <krompus> hmm
[01:25:37 CET] <krompus> so, i'm chatting with the teiler dev, we're trying to work out some profiles
[01:25:53 CET] <krompus> he says: "a little warning: yuv444p wont work with every video size
[01:25:55 CET] <krompus> "
[01:26:05 CET] <krompus> "it needs a factor of 8"
[01:26:07 CET] <krompus> hmm
[01:27:21 CET] <krompus> any other suggestions to get the colours slightly better? It's still a little washed out
[01:27:40 CET] <krompus> MUCH better though! I'm just seeing how far I can take it.
[01:29:26 CET] <J_Darnley> 444 has no subsampling
[01:29:39 CET] <J_Darnley> it will work with more video sizes than 420
[01:30:06 CET] <J_Darnley> 420 reduces the size of both chroma planes by half vertically and half horizontally
[01:31:20 CET] <J_Darnley> My ultimate suggestion for colours is to know what ffmpeg is doing and make sure your player is doing the same in reverse
[01:31:26 CET] <J_Darnley> The first is hard
[01:31:55 CET] <J_Darnley> You have RGB input, right?
[01:32:00 CET] <J_Darnley> Some sort of screen cap?
[01:32:51 CET] <J_Darnley> In that case ffmpeg is probably doing a conversion to TV levels
[01:33:14 CET] <J_Darnley> you must ensure that the player converts back to full range
[01:33:33 CET] <krompus> oh gosh. I know some of these words
[01:33:35 CET] <J_Darnley> The usual symptom with this is poor contrast
[01:34:02 CET] <J_Darnley> Follow the white rabbit, Neo.
[01:34:06 CET] <krompus> :P
[01:34:14 CET] <J_Darnley> low contrast in typical video
[01:34:38 CET] <J_Darnley> with bold colours I guess "washed out" might describe that too
[01:35:07 CET] <krompus> well, my display is pretty high contrast https://ptpb.pw/FmOF.png
[01:35:32 CET] <krompus> i'm not sure what you're referring to
[01:35:52 CET] <krompus> here, I'll scrot the video during playback
[01:36:14 CET] <krompus> https://ptpb.pw/KsBj.png
[01:36:32 CET] <krompus> A/B those.
[01:36:35 CET] <krompus> see what I mean?
[01:36:37 CET] <krompus> it's close
[01:37:16 CET] <krompus> it's like it's "rounding" the colours
[01:37:41 CET] <J_Darnley> Yes, I see it in the pink text in the top-right
[01:37:50 CET] <krompus> yep yep. it's pretty obvious there
[01:38:21 CET] <J_Darnley> I don't think this is TV vs PC levels
[01:38:23 CET] <krompus> anyways, not a huge deal, but it would be nice if I could "enhance" so to meme
[01:38:38 CET] <J_Darnley> I think it is a difference in colour matix being used.
[01:38:53 CET] <krompus> (disclaimer: I'm not very familiar with ffmpeg, so please ELI5)
[01:38:58 CET] <krompus> well
[01:39:01 CET] <J_Darnley> i'm getting there
[01:39:09 CET] <krompus> not ELI5, but talk to me like a layperson
[01:39:11 CET] <krompus> hahaha
[01:39:20 CET] <krompus> you don't have to explain everything
[01:39:23 CET] <krompus> :)
[01:39:47 CET] <J_Darnley> In the very old days ffmpeg would always 601 for yuv<->rgb conversion.
[01:40:01 CET] <J_Darnley> ( If you really want to know: https://en.wikipedia.org/wiki/YCbCr )
[01:40:21 CET] <J_Darnley> 601 is the usual matrix used for SD content
[01:40:33 CET] <J_Darnley> 709 is usually usef ror HD content
[01:40:37 CET] <J_Darnley> *used
[01:41:01 CET] <krompus> ahh
[01:41:14 CET] <J_Darnley> I can't remember whether the choice of colour matrix was ever added to ffmpeg (specifically swscale)
[01:41:31 CET] <J_Darnley> I would guess that ffmpeg is using 601
[01:41:47 CET] <J_Darnley> Then you player is using 709 because it is HD
[01:42:00 CET] <krompus> ahh
[01:42:10 CET] <J_Darnley> libx264 has a methos for signalling which matrix should be used by the player/decoder
[01:42:14 CET] <J_Darnley> *method
[01:42:26 CET] <krompus> yeah, i'm all hd; it would be nice to use 701
[01:42:31 CET] <J_Darnley> let me look that up
[01:42:37 CET] <krompus> awesome; brb
[01:44:18 CET] <J_Darnley> Perhaps the scale filter does let you control it: http://ffmpeg.org/ffmpeg-filters.html#scale-1
[01:45:29 CET] <J_Darnley> specifically the in/out_color_matrix options
[01:47:40 CET] <J_Darnley> Oh god.  Why is there both x264opts and x264-params?
[01:49:09 CET] <J_Darnley> Anyway, I think you want to pass the colormatrix option to one of those.
[01:49:51 CET] <krompus> how do?
[01:51:03 CET] <J_Darnley> Possibly: -x264opts colormatrix=bt709
[01:52:33 CET] <J_Darnley> Well that does not result in an error
[01:53:35 CET] <krompus> http://img2.wikia.nocookie.net/__cb20100427134246/half-life/en/images/b/b8/Error.jpg
[01:54:08 CET] <J_Darnley> Basically, yes
[01:54:22 CET] <J_Darnley> We even red text in ffmpeg
[01:54:31 CET] <J_Darnley> (Which people still ignore)
[01:56:10 CET] <krompus> y u no read error
[01:56:55 CET] <J_Darnley> ah.  "This option is functionally the same as the x264opts, but is duplicated for compatibility with the Libav fork"
[01:57:04 CET] <J_Darnley> That explains that.
[04:28:23 CEST] <squarecircle> ohai
[04:28:45 CEST] <squarecircle> Does someone can tell me where I'm lost?
[04:29:03 CEST] <squarecircle> I try to understand the difference between height and coded_height
[04:40:32 CEST] <J_Darnley> At a guess: the height of the video/content vs the height produced by the encoder/decoder
[04:41:32 CEST] <squarecircle> so if I decode it, I get the coded_height?
[04:41:58 CEST] <squarecircle> because actually I'm creating a raw stream
[04:42:26 CEST] <squarecircle> and its quite important to know exactly how many bytes I have to read
[04:46:55 CEST] <squarecircle> well I'll try
[04:47:00 CEST] <squarecircle> tomorrow
[04:47:05 CEST] <squarecircle> thanks
[05:01:00 CEST] <Guest_24552> does anyone in here have any clue on how I could detect steganography in a recompressed video?
[05:01:33 CEST] <Guest_24552> I have an issue right now where I want to tell if a video has hidden information within its pixel data
[05:01:52 CEST] <Guest_24552> and I have an ad-hoc approach that works for slow moving videos
[05:02:00 CEST] <Guest_24552> if the algorithm is known
[05:02:18 CEST] <Guest_24552> but generally what features of a video are "normal"?
[05:04:05 CEST] <Guest_24552> like I know people add intentional modifications to videos; special effects, blurring, lighting and video composition
[05:04:25 CEST] <Guest_24552> but how do I separate such intentional effects from unintentional ones
[05:04:38 CEST] <Guest_24552> ones that may be malicious
[05:04:53 CEST] <Guest_24552> blindly
[05:05:43 CEST] <jmack> Is there a specific version of FFMPEG for Mint: Rosa?
[05:06:21 CEST] <thebombzen> jmack: I'd just compile it from source. that's what I did on my mint box
[05:07:27 CEST] <jmack> So v3.0 just needs to be compiled ans then run?
[05:11:28 CEST] <jmack> Should I install in main filesystem or in portable mode i.e. /home/username/sources ?
[05:24:03 CEST] <thebombzen> jmack: just run ./configure <options>, make, sudo make install
[05:24:06 CEST] <thebombzen> you know
[05:24:07 CEST] <thebombzen> the normal way
[05:24:31 CEST] <thebombzen> keep in mind that stuff from the repos linked to libav will not see ffmpeg's stuff. so if you want mplayer for example I'd build that from source too
[05:25:43 CEST] <jmack> thebombzen: Thanks, this is my first time on any kind of chat
[05:34:55 CEST] <jmack> thebombzen: I'm so tired of trying to capture /Desktop with avconv, I get nothing but errors (segmentation fault) and frustration
[05:36:08 CEST] <llogan> jmack: you could just download a binary http://johnvansickle.com/ffmpeg/
[05:36:27 CEST] <llogan> oh, wait, that probably doesn't have x11grab support
[05:36:32 CEST] <llogan> i can't remember
[05:36:52 CEST] <llogan> or see http://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu
[05:38:22 CEST] <jmack> llogan: your right it doesn't have non-free support
[05:44:39 CEST] <jmack> llogan: I saw that page and book marked it I think it'll come in handy
[05:45:41 CEST] <jmack> thebombzen: Where would I get the source for mplayer?
[05:46:26 CEST] <jmack> thebombzen: Do you have a link on hand?
[05:47:20 CEST] <llogan> you can probably just use the repo "ffmpeg" to provide the dependencies for other repo junk, and use your compiled ffmpeg binary for capturing the desktop
[05:49:55 CEST] <thebombzen> just google "mplayer" and poke around a bit
[05:50:01 CEST] <jmack> thebombzen: Is there an easyway to link stuff like mplayer to FFMPEG?
[05:50:19 CEST] <thebombzen> yea. if you compile it from source then it'll look in /usr/local/lib before /usr/lib
[05:50:41 CEST] <thebombzen> so if you compiled ffmpeg from source (make sure you do the dynamic build, not static) then build mplayer it should automatically use that one
[05:51:01 CEST] <thebombzen> if you want to be extra careful, don't install the development files for libav. that way pkg-config will only see ffmpeg
[05:52:30 CEST] <llogan> jmack: maybe Mint isn't the best fit for you if you want newer stuff and are wanting to compile.
[05:53:46 CEST] <jmack> thebombzen: But what if I want to use libmp3lame, don't you have to install libmp3lame-dev files?
[05:54:45 CEST] <jmack> llogan: What would be a better linux for newer stuff?
[05:55:42 CEST] <thebombzen> llogan: although he seems like a Linux beginner, so I'd say mint is alright
[05:56:03 CEST] <thebombzen> jmack: Arch Linux always has the latest stuff. but it's not for Linux beginners
[05:56:22 CEST] <thebombzen> and yes, you'd have to install lame-dev
[05:57:02 CEST] <thebombzen> but LAME is not libav. I meant don't install libavcodec-dev and related packages.
[05:58:56 CEST] <jmack> thebombzen:  Ok, I've been tinkering with Linux Ubuntu since 6.06 and before that Suse, Redhat, Slackware and some others
[05:59:15 CEST] <thebombzen> odd that you haven't built stuff from source then
[06:00:48 CEST] <llogan> jmack: Arch keeps up with upstreams well. but it takes work to setup and slightly more work to maintain.
[06:01:33 CEST] <jmack> thebombzen: I'm not a MOTU or Mr. Moneybags, (or I'd have a $200 Windows 10 maybe) I just need some minor stuff...
[06:02:18 CEST] <thebombzen> either way, Arch requires you to build stuff from source all the time
[06:02:26 CEST] <jmack> thebombzen: I never said I haven't compiled stuff, it's just been a while since I needed to...
[06:03:09 CEST] <thebombzen> anyway you can build ffmpeg the way you build everything else then
[06:03:23 CEST] <thebombzen> just do yourself a favor and disable the static libraries
[06:03:39 CEST] <jmack> thebombzen: I looked into Arch but it didn't seem as organized as the rest of the distro's out there
[06:04:11 CEST] <jmack> thebombzen: Disable static, roger that!
[06:05:16 CEST] <jmack> thebombzen: Arch seems to be very hands on and right now I don't have a lot of free time to
[06:05:54 CEST] <jmack> thebombzen: Do everything everybodu wants me to do.
[06:09:32 CEST] <jmack> thebombzen & llogan: Thanks guys...I really appreciate your help.
[06:19:34 CEST] <jmack> thebombzen: Well it looks like I'm going to have to get my hands dirty if I hope to get anything done, this stuff has changed so much since I first started doing this. This is what I get for getting to used to that stupid "Software Manager" Aarg!
[06:24:56 CEST] <deweydb> I can't figure out what i'm doing wrong here, i'm trying to strip the audio from this file and re-encode as mp4.  my command line output is: http://pastebin.com/mDjKFHRv and the input file is: https://www.dropbox.com/s/vbrtjodlo7fjqj4/C31-8x14.AVI?dl=1  and my output is always a broken (unplayable) video: https://www.dropbox.com/s/8ukfrcwfhzjiklg/1459052554-no-audio.mp4?dl=1
[06:28:31 CEST] <deweydb> hmmm added -pix_fmt yuv420p and seems to play now. ok.
[06:28:35 CEST] <deweydb> i guess i'm just dumb
[06:28:56 CEST] <c_14> Then I guess your player is just shitty
[06:29:08 CEST] <deweydb> ehhh i was testing with just spacebar in osx
[06:29:15 CEST] <deweydb> which i guess is kinda a shitty player
[06:29:25 CEST] <c_14> Most non-ffmpeg based players only support yuv420p
[06:29:32 CEST] <c_14> Also most hardware players/mobile etc
[06:29:56 CEST] <deweydb> good to know. thanks
[06:30:12 CEST] <llogan> i'm not aware of any non-FFmpeg that do support other than 420
[14:59:24 CEST] <kszere> Hi, I need convert multiple file from *.ts to *.mkv with multiple source audio and video. Output file must be cut start and end time.  How I do it?
[15:05:43 CEST] <gnome1> I've been transcoding to 768x576, but in some places people mention 720x576 for PAL, which is not 4:3. Does it matter that much if I use 768 instead, is there some reason not to use 768x576? it's mostly to view in a computer, although it'd be nice if it could be played in a TV somehow.
[15:07:53 CEST] <J_Darnley> 720x576 is not 4:3 thanks to the glory of anamorphic video.
[15:35:07 CEST] <kszere> Hi, I need convert multiple file from *.ts to *.mkv with multiple source audio and video. Output file must be cut start and end time.  How I do it?
[15:48:07 CEST] <J_Darnley> I still have no clue what half of that means.
[15:48:28 CEST] <J_Darnley> to convert ts to mkv: ffmpeg -i input.ts -c copy output.mkv
[15:51:30 CEST] <furq> gnome1: 720*576 is anamorphic pal dvd
[15:52:01 CEST] <furq> if you're ripping dvds then you're best off not resizing at all and setting the aspect ratio in the container
[15:54:10 CEST] <Filarius> how to fix "non-monotonous DTS in output stream" ?
[15:59:35 CEST] <DHE> Filarius: using stream copy? or re-encoding?
[15:59:44 CEST] <gnome1> furq: nah, I'm downscaling wider videos, mostly "HD" frame sizes
[16:00:53 CEST] <gnome1> perhaps the rescaling should be seen in a case-by-case basis for each source, to make sure it's done in the best way possible, but I doubt it will cause a big problem unless I'm watching less than 1m from the screen?
[16:01:24 CEST] <furq> shrug
[16:01:30 CEST] <furq> i just use spline16 for everything
[16:01:42 CEST] <furq> or bicubic for blocky sources
[16:02:00 CEST] <gnome1> hm, I'm using the default, whatever that is
[16:02:12 CEST] <gnome1> IIRC the help said it'd be 4, and that seemed to be "experimental"
[16:02:39 CEST] <gnome1> experimental or nearest neighbor, depending if you start counting from 1 or 0
[16:09:30 CEST] <furq> the default with sws is bilinear or bicubic, i forget now
[16:09:39 CEST] <furq> zscale is better if you have that compiled in
[16:11:18 CEST] <gnome1> is that the zimg-based scaler? I could enable that. how does it perform speed-wise and quality-wise compared to bilinear and bicubic?
[16:12:58 CEST] <furq> it seems roughly as fast as swscale
[16:13:18 CEST] <furq> swscale has spline/lanczos etc but zscale is apparently better
[16:13:30 CEST] <furq> according to the people in here who can tell the difference
[16:18:23 CEST] <Filarius> DHE, "non-monotonous DTS" error while re-encoding HLS stream, what was recorded by ffmpeg with "copy" encoder, and its also have some related error
[16:18:41 CEST] <Filarius> also here "past duration too large" errors
[16:25:50 CEST] <gnome1> I can't find zscale, so I probably need to enable something, now I wonder what, the zimg option I saw doesn't seem to be there for this version
[16:30:03 CEST] <kszere> @J_Darnley Thanks.
[16:33:36 CEST] <kszere> How files cut start and end time also with convert to *.mkv from *.ts? What I can add to this command: "ffmpeg -i input.ts -c copy output.mkv"?
[16:36:58 CEST] <relaxed> kszere: read about the -ss, -t, and -to
[17:53:17 CEST] <andrey_utkin> Hi! we have some hardware which generates h264 stream which is sometimes broken, and we generate frame headers manually. Sometimes the garbage sent by hardware confuses mplayer/ffplay/whatever, and decoder likely doesn't recover from that (despite we send I-frames quite frequently). What could be done to stream to make decoder to recover? Maybe declare IDR point? How to do that? By sending SPS&PPS again?
[18:29:05 CEST] <jkqxz> andrey_utkin:  Are your I frames actually IDR frames there or not?  (They aren't the same thing; an I frame on its own isn't necessarily sufficient to recover.)
[18:30:53 CEST] <jkqxz> If they are actually IDR frames they should generally be immediately preceeded by SPS and PPS NAL units; if they aren't then adding them will make it definitely-recoverable at that point.
[18:31:44 CEST] <andrey_utkin> jkqxz, thanks
[18:32:20 CEST] <andrey_utkin> jkqxz, that's why i asked - i don't quite know what's diff between I and IDR. The code which generates headers is this, you may take a look https://github.com/bluecherrydvr/linux/blob/tw5864/drivers/staging/media/tw5864/tw5864-h264.c#L80
[18:32:43 CEST] <andrey_utkin> jkqxz, is it correct that SPS+PPS+Iframe = IDR?
[18:32:57 CEST] <andrey_utkin> or SPS+PPS+Iframe-with-something-special?
[18:39:01 CEST] <jkqxz> IDR frames are made up of IDR slices (NAL unit type 5), all other frames are made up of non-IDR slices (NAL unit type 1).  The type there affects the slice header encoding.
[18:41:19 CEST] <jkqxz> Writing the NAL unit type there, you have "**buf = (frame_seqno_in_gop == 0) ? 0x25 : 0x21;".  So you begin every GOP with a IDR
[18:51:18 CEST] <jkqxz> The stream from this device is fixed-QP?
[18:52:05 CEST] <jkqxz> (If not, you'll need to set slice_qp_delta to something other than zero.)
[18:55:05 CEST] <andrey_utkin> yes, i set QP once and don't change the value
[18:55:26 CEST] <andrey_utkin> so i have IDR type NAL?
[18:56:09 CEST] <andrey_utkin> i think i send SPS+PPS just once
[18:57:47 CEST] <jkqxz> The code looks like it does the right thing, as far as I can tell.  But yes, you should send the SPS+PPS before every IDR frame.
[18:58:35 CEST] <andrey_utkin> jkqxz, thank you so much
[19:05:10 CEST] <jkqxz> For the other part of your problem, where something causes it to go wrong initially, I suggest capturing the stream from the beginning to the point where it fails.  Then you can examine it offline to work out what the problem is.
[19:11:23 CEST] <andrey_utkin> jkqxz: that is good advice, but i have repeatedly asked during last year for help with debugging h264 stream issues - for stream analysis tools and for personal paid support, with no success. I always had stream dumps when i was asking :-)
[19:14:23 CEST] <jkqxz> If you wrote the code you just showed yourself, you should have sufficient knowledge to do it.  ffmpeg and the reference decoder are the only tools you actually want here.
[19:19:58 CEST] <andrey_utkin> jkqxz: i disagree. At last wireshark-alike parser/presenter would help a lot
[19:39:55 CEST] <jkqxz> If your stream is obviously broken, then reading the trace output of the reference decoder at the point where it fails is exactly what you want to do.
[20:04:22 CEST] <Mavrik> jkqxz, is there a reference decoder implementation somewhere in ffmpeg source tree?
[20:09:03 CEST] <jkqxz> No.  It would be certainly nice if the reference decoder could be called directly from ffmpeg, but right now you can just dump the elementary stream and then feed that to the reference decoder (and if you can't get that far then you have more obvious problems to pursue).
[20:18:24 CEST] <andrey_utkin> jkqxz, are you talking about this software? http://iphome.hhi.de/suehring/tml/download/
[20:20:29 CEST] <jkqxz> Yes.
[21:17:35 CEST] <fturco> hello. i'm trying to back up my dvds with ffmpeg.
[21:17:45 CEST] <fturco> my current command is: ffmpeg -i alien-1.vob -map 0:0 -map 0:2 -metadata:s:a language=ita -codec copy test.vob
[21:18:16 CEST] <fturco> the problem is the resulting audio track is not marked as italian when i play it with mplayer
[21:18:43 CEST] <fturco> it says 128 unknown
[21:25:49 CEST] <andrey_utkin> fturco, what's the point on coverting if you can just copy? you are not changing the format as I see
[21:26:12 CEST] <andrey_utkin> of you are trying to enhance it?
[21:27:02 CEST] <fturco> i don't want to re-encode my dvds. just copy video and audio streams as-is and eventually remux them in a different container format
[21:27:10 CEST] <fturco> i would choose mkv but i have problems with it
[21:28:00 CEST] <fturco> original vob file contains extra audio tracks and subtitles that i don't want
[21:30:34 CEST] <andrey_utkin> fturco, i meant why not "cp alien-1.vob backup/alien-1.vob"
[21:31:07 CEST] <fturco> andrey_utkin, that is not good because alien1.vob contains extra audio tracks and subtitles that i want to remove
[21:31:10 CEST] <andrey_utkin> throwing away unneeded stuff, ok, understand
[21:32:11 CEST] <fturco> i'm not sure about how to properly use the -metadata option
[21:35:20 CEST] <fturco> i originally tried mkvmerge but it causes audio-video desync
[21:48:37 CEST] <J_Darnley> fturco: I would guess that you don't set language with the generic metadata option.
[21:49:43 CEST] <fturco> J_Darnley, i also tried with -metadata:s:a:0 language=ita, to no avail
[21:50:19 CEST] <J_Darnley> Although I also wonder whether the vob muxer supports metadata/language at all
[21:50:50 CEST] <furq> i'm pretty sure that metadata is in the ifo
[21:51:32 CEST] <fturco> J_Darnley, i would prefer to use matroska, but i get the following error message with it: Can't write packet with unknown timestamp
[21:51:54 CEST] <fturco> that error goes away if using the -fflags +genpts options
[21:52:01 CEST] <fturco> but then the process is very slow
[21:52:58 CEST] <fturco> i can't think of any other container i could use
[21:54:55 CEST] <J_Darnley> Sorry.  I don't have any useful suggestions.
[21:55:13 CEST] <fturco> J_Darnley, no problem
[21:58:13 CEST] <J_Darnley> Well the massive ffmpeg help page does suggets that using metadata is the right way
[21:58:38 CEST] <J_Darnley> Either it is unused by the muxer or it belongs in the ifo as furq suggested.
[21:59:56 CEST] <furq> fturco: mpegts should work, although i don't really recommend it
[22:00:57 CEST] <furq> i'm not sure why -fflags +genpts is causing a slowdown for you, it doesn't here
[22:01:20 CEST] <fturco> the output file size is not growing constantly
[22:01:26 CEST] <fturco> it is stuck
[22:01:35 CEST] <furq> with mkv?
[22:01:36 CEST] <furq> weird
[22:01:40 CEST] <fturco> yes
[22:02:05 CEST] <fturco> i had to stop it
[22:03:07 CEST] <fturco> it is stuck at 240mb
[22:03:31 CEST] <furq> that sounds like a bug if it works with other muxers
[22:04:25 CEST] <fturco> now it stopped at 784mb
[22:04:43 CEST] <fturco> it doesn't continuously update progress
[22:31:10 CEST] <jasonsu> Hi.  I want to use ffmpeg to single-line generate sine wav test tones into mp3 files.  I need a given frequency, 30 secs long, 44.1KHz sampling rate, normalized to 0Db. I've been trying to piece together various bits and keep stumbling :-(
[22:31:11 CEST] <jasonsu> Right now I try:
[22:31:21 CEST] <jasonsu> ffmpeg -f lavfi -i "sine=frequency=1000:sample_rate=44100:duration=30:volume=0dB" -acodec mp3 -b:a 192k out.mp3
[22:31:27 CEST] <jasonsu> But it complains about "Option 'volume' not found".  Can anyone lend a hand? just want to get this working.
[22:32:40 CEST] <J_Darnley> It doesn't have one.
[22:33:09 CEST] <jasonsu> J_Darnley: Sorry, what doesn't have what?
[22:33:20 CEST] <J_Darnley> sine does not have a volume option
[22:33:33 CEST] <jasonsu> How do I normalize to 0db?
[22:34:04 CEST] <J_Darnley> Juding from the docs: apply a gain of 8x
[22:34:33 CEST] <jasonsu> J_Darnley: ?  Would you mind point me at that?
[22:34:37 CEST] <jasonsu> ing
[22:34:51 CEST] <J_Darnley> http://ffmpeg.org/ffmpeg-filters.html#volume
[22:38:31 CEST] <jasonsu> sorry, don't see anything about gain = 8x there.  I do see "level
[22:38:31 CEST] <jasonsu>     Auto level output signal. Default is enabled. This normalizes audio back to 0dB if enabled. "
[22:41:31 CEST] <J_Darnley> volume=8 if I read it correctly
[23:47:23 CEST] <stemid> hey I'm in a weird situation with some old mp3 files that show up as just "Audio file with ID3 version 2.3.0" and work in my android, and then I have some other files show up as "Audio file with ID3 version 2.4.0, contains: MPEG ADTS, layer III, v1,  64 kbps, 44.1 kHz, Stereo" and not work.
[23:47:28 CEST] <stemid> how can I replicate the first type?
[23:47:50 CEST] <stemid> doing just ffmpeg -i file.wav -f mp3 gives me the second type
[23:48:33 CEST] <Mavrik> What does ffprobe say on both?
[23:48:38 CEST] <stemid> let me see
[23:48:40 CEST] <andrey_utkin> i'd check encoder & muxer settings
[23:48:49 CEST] <J_Darnley> I think there's an option for controlling the ID3 version
[23:51:43 CEST] <stemid> https://paste.fedoraproject.org/345867/91154531/ this is the file that my android cannot discover in spotify/google play/sonos apps. https://paste.fedoraproject.org/345868/15456145/ this one works everywhere.
[23:54:06 CEST] <Mavrik> Strange.
[23:54:23 CEST] <stemid> yeah I'm not an audio wiz but I can't see any weird differences
[23:54:58 CEST] <stemid> Ghost Commander app can play the first mp3 file btw. so maybe it's DRM
[23:55:01 CEST] <stemid> I wouldn't rule that out
[23:55:17 CEST] <stemid> because I've only tried major apps, not vlc for example.
[23:57:04 CEST] <stemid> yup works in vlc
[23:57:19 CEST] <stemid> this is sad, I wanted to play it on my Sonos speaker. must be stupid DRM
[00:00:00 CEST] --- Mon Mar 28 2016


More information about the Ffmpeg-devel-irc mailing list