[Ffmpeg-devel-irc] ffmpeg.log.20130530
burek
burek021 at gmail.com
Fri May 31 02:05:02 CEST 2013
[00:04] <Keshl> What can I do if I believe a bug I reported isn't being taken seriously by developers, oÉo?
[00:09] <llogan> Keshl: which ticket is it?
[00:09] Action: Keshl finds.
[00:10] <Keshl> llogan: http://ffmpeg.org/trac/ffmpeg/ticket/2607 oÉo
[00:10] <Keshl> A few people agrees that cehoyos seemed to be on "automatic mode", so to speak, when he looked at my ticket. Now it's stuck in undetermined, even though I did most of the work for him to show that there /is/ an issue. he just won't log into the FTP server to check. <.<
[00:11] <ubitux> Keshl: there are 620 opened tickets, for maximum 10 developers already working on different bugs/features/random-other-job
[00:11] <ubitux> so... you'll have to wait a little i guess :)
[00:11] <Keshl> Oh D: Kinda thought there were more devs, my bad .É.
[00:12] <llogan> we are just volunteers.
[00:12] <Keshl> I know, but so are Blender's peoples, and they got a few hundred <É< I think.
[00:13] <Keshl> For some reason I just assume that open source projects always have hundreds of devs. Might have to do with me being in #blender and KDE-related channels a lot.
[00:14] <llogan> and an update to your ticket by a developer 35 hours ago does not seem like you're being ignored.
[00:16] <Keshl> Lemme explain my side of this then, it could just be me; When I opened that ticket, he replied within a minute and basically said "I didn't read your entire post, because you said that you uploaded exactly what I wanted to the FTP server but I don't know that since I didn't read your post.". I uploaded the attachment again (a second time technically), and he repeats the same thing: "I'm asking a question you already answered, but I don't
[00:16] <Keshl> know that because I still never actually read your ticket"
[00:16] <Keshl> Hence, I think he's just kind of ignoring me, and since there's more than two comments other devs are saying "Oh, someone else is handling it, I won't bother reading it."
[00:18] <rickbol> is there any means to inject metadata into the webm format (libvpx)?
[00:21] <llogan> Keshl: maybe i'm looking at the wrong ticket because I don't get that impression. he asks for console output. that's a required piece of info. then he asks for another using a different command. i don't see any of the quotes you displayed here.
[00:21] <Keshl> llogan: I did mention I tried it with mpeg4 in my ticket, right?
[00:21] <Keshl> (Or did I just say "any encoder or container"?)
[00:22] <llogan> rickbol: -metadata title="FFmpeg Dubstep Band"
[00:24] <Keshl> Kay, my bad, appearntly I never did say I did mpeg4.
[00:24] <Keshl> ubitux did later in the comments, though .É.
[00:26] <llogan> Keshl: to help narrow down the cause it is useful to know if the behavior also occurs with a native encoder; otherwise it can be a bug with the external library.
[00:27] <Keshl> Ish native, oÉo. Happens on everything.
[00:27] <Keshl> Any encoder, any codec, any container.
[00:29] <rickbol> llogan: hmm, just tried the data I use when I generate a mp4, and the fields don't show in vlc for a webm like they do for mp4.
[00:30] <rickbol> llogan: is there a list of supported keywords, or does it take anything you throw at it?
[00:51] <llogan> Keshl: a console output is required, regardless.
[00:52] <llogan> Keshl: but ubitux beat you to it.
[00:52] <Keshl> llogan: But I put'd! .É.
[00:52] <Keshl> In two places, FTP server and as an attachment to the ticket .É.
[00:58] <relaxed> llogan: Most people who want to hardsub ass will be dealing with matroska. You should show an example of how to demux them, or can the ass filter use the video's internal ass stream?
[00:59] <Keshl> ...Uh.
[00:59] <Keshl> Is "rumpbutacuss" an abbreivation for something?
[01:01] <relaxed> does the ass filter require external subs? what about embedded fonts?
[01:10] <rickbol> llogan: from the pelr script... http://pastebin.com/Xz1SWLsi
[01:10] <rickbol> perl
[01:13] <rickbol> the script works, and the metadata appears in the console, but it doen't appear in vlc. What else can I use to view it? I tried firefox. It plays but no metadata
[01:13] <rickbol> These same tags do appear in vlc when creating the mp4
[01:14] <rickbol> I can't remeber which of them appeared when I messed around with mkv. Some, i think, but not all.
[01:14] <llogan> relaxed: i'm not sure. i don't deal with subtitles often.
[01:14] <llogan> or i should say, "ever".
[01:16] <llogan> rickbol: sorry, i'm not sure what webm supports for metadata.
[01:18] <rickbol> I read, as of last June, the plan was to use the mkv structures. But I don't know about now, and I'm using deb-multimedia ffmpeg, so I don't know where my version is.
[01:18] <rickbol> thanks though. "title" worked!!! ;-)
[01:18] <llogan> ffmpeg shows the version with each command, including simply "ffmpeg".
[01:20] <rickbol> llogan: do you know if there's a command-line utility I can use to view metadata? grepping for text hasn't turned up much.
[01:20] <durandal_1707> you can save metadata to file
[01:20] <llogan> i just use "ffmpeg -i input"
[01:20] <durandal_1707> also ffprove have various outputs....
[01:21] <llogan> if you want to process it follow durandal_1707 suggestions.
[01:21] <durandal_1707> *ffprobe , not ffprove :)
[01:22] <rickbol> looks like "title" is the only metadata that was saved into the file.
[01:24] <durandal_1707> that is only metadata no?
[01:24] <durandal_1707> if you need more use ffprobe -show_format
[01:25] <durandal_1707> for even more info there is documentation
[01:25] <llogan> rickbol: does the webm container spec indicate what metadata is supported?
[01:26] <llogan> *are
[01:27] <durandal_1707> webm supports bunch of metadat like matroska does
[01:29] <llogan> maybe in theory, but i cant get it to support as much as matroska
[01:29] <llogan> but also i haven't looked at any code.
[01:31] <durandal_1707> its just matroska with only vp8/vorbis/opus/vp9 in it
[01:32] <llogan> has opus in matroska/webm been finialized yet?
[01:32] <llogan> he does that often.
[01:32] <relaxed> ffmpeg/libavformat/matroskaenc.c shows title, creation_time, and language
[01:32] <relaxed> and a few for specific streams
[01:37] <llogan> http://wiki.webmproject.org/webm-metadata/global-metadata
[12:42] <Keshl> SQUEE ¿É¿
[12:43] Action: Keshl 's bug was confirmed OÉO!
[12:43] <Keshl> Well, squee for me, not-so-squee for you guys. .É. -Hands appology cupcakes- >É<
[12:51] <Bor0> what are all these smileys/IRC ]emoticons
[12:52] <ubitux> squirrel on lsd
[13:02] <ubitux> japanese squirrel even
[13:34] <chrisbuchholz> Hey guys
[13:34] <chrisbuchholz> what does exit status 5 mean?
[13:34] <chrisbuchholz> ...as in, the exit status i get when i run my ffmpeg command
[13:59] <elkng> really, is there a list of exit codes for ffmpeg anywhere ?
[16:45] <Emmanuel_Chanel> Hello!
[16:53] <smt> hello everyone.. I'm having trouble encoding audio when resampling from 44.1khz to 22kz because the codec does not expect the given nb of input samples to be different than frame_size
[16:53] <smt> but I can't get find the correct math to convert the nb of input samples at 44.1khz down to 22khz
[16:54] <smt> I'm using the formula like this: int out_samples = av_rescale_rnd(swr_get_delay(swr, 44100) + in_samples, 22050, 44100, AV_ROUND_UP);
[16:55] <smt> but when I do out_samples = swr_convert(swr, &output, out_samples, input, in_samples), out_samples is much lower than codec->frame_size
[16:56] <smt> any help regarding what am I blatantly missing?
[17:45] <bunniefoofoo> smt you are missing the fact that the resampler (I assume you use swr_convert) has an internal buffer, it won't return the amount you expect based on conversion ratio
[17:47] <smt> bunniefoofo, I think I found the issue and it's pointing to pad_last_frame
[17:48] <bunniefoofoo> swr_get_delay returns the *current* delay in the resampler which is a moving target. if you haven't called swr_convert I think it will return 0
[17:48] <smt> I get this message, but I shouldn't be receiving it
[17:48] <smt> av_log(avctx, AV_LOG_ERROR, "nb_samples (%d) != frame_size (%d) (avcodec_encode_audio2)\n", frame->nb_samples, avctx->frame_size);
[17:49] <smt> if (frame->nb_samples < avctx->frame_size && !avctx->internal->last_audio_frame) { ret = pad_last_frame(avctx, &padded_frame, frame);
[17:49] <smt> but it's not doing what it is supposed to do
[17:51] <bunniefoofoo> i think most codecs require nb_samples == frame_size except for the last frame, there is a capability flag if they except variable sizes
[17:51] <bunniefoofoo> what you need to do is increase your input samples to resampler so you always get a full frame out when calling swr_convert
[17:53] <smt> right, but I don't why av_rescale_rnd does not provide me the correct sample number
[17:53] <smt> *don't understand
[17:55] <smt> int out_samples = 2048 * 22050 / 44100 = 1024
[17:55] <smt> but when I call swr_convert, I get 1008 samples
[17:55] <smt> but the codec needs 1024
[17:55] <bunniefoofoo> yep, this is how swr_convert works believe it or not
[17:55] <smt> so I'm not sure if I should create an intermediate buffer or increase out_samples
[17:55] <smt> or pad the frames with silence
[17:56] <smt> which strategy is the best (or works)?
[17:58] <bunniefoofoo> are you resampling directly into AVFrame->data ?
[18:00] <smt> yes
[18:01] <bunniefoofoo> I would try increasing the input size, and use frame_size as the output size (to swr_convert). swr_convert will buffer if you feed it too much input
[18:01] <smt> avcodec_fill_audio_frame
[18:17] <talcat> Would someone be willing to help me convert a sequence of tif (gray16be) to a video (mp4/avi/anything really). Even when encoding the video with pix_fmt yuv420p, the output video looks as though there is pixel wrap around (the grayscale values are all wrong...). Thanks very much
[18:29] <elkng> is it possible to use ffmpeg or any other tool to make encrypted video ? like when you want to watch it you need to enter a password
[20:40] <hubx> I have some .ogg files encoded with ffmpeg (ffmpeg -i in.flac -ac 2 -vn -ab 64k -ar 44100 -f ogg out.ogg) but none of the major browser(Firefox, Chrome can play it in the browser). However VLC, mplayer are fine
[20:40] <hubx> See http://213.133.99.233/media/encoded/ag-proea_2013-05-16/ag-proea_2013-05-16.ogg
[20:42] <llogan> hubx: by default flac is used for output in ogg container. add "-acodec libvorbis" as an output option.
[20:48] <hubx> llogan, oh thanks. Now I see that in mplayer `[lavf] stream 0: audio (flac), -aid 0`
[20:50] <bunniefoofoo> is the aac encoder reliable or should I be using faac? The reason I ask is because my app crashes when I use ffmpeg aac and doesn't if I run i tthrough the debugger. I have not tried faac yet
[20:53] <bunniefoofoo> The aac encoder in 0.8 seemed to work fine, now crashing in 1.2.1
[20:54] <durandal_1707> api changed, so it may crash with others too
[20:55] <bunniefoofoo> I updated code to use the new apis, works on linux x64 but not win32, but sometimes will (heisenbug)
[20:57] <bunniefoofoo> valgrind on linux x64 only spews x264 errors
[22:23] <admalledd> ok, trying to capture video from both my desktop and webcam at the same time, running into an issue in that my ffmpeg (from the PPA) seems to lack `-filter_complex`, see http://pastebin.com/jz0yyGSL for CLI and error. Any ideas?
[22:24] <admalledd> and the PPA used was https://launchpad.net/~jon-severinsson/+archive/ffmpeg and im on linux mint 13 maya 64bit in case that matters...
[22:27] <llogan> admalledd: you'll have to use a newer ffmpeg, or use -vf with "movie" multimedia source filter if you want to use graybeard ffmpeg from PPA.
[22:28] <ubitux> this ppa is unfortunately extremely old now
[22:28] <admalledd> blarg, so any source of newer pre-compiled builds or am I now off on an adventure to compile my own ffmpeg?
[22:29] <llogan> none of the third-party builds support x11 AFAIK
[22:29] <llogan> other than the PPA
[22:29] <ubitux> admalledd: you also have the possibility to complain to your distro :p
[22:30] <llogan> what's your ubuntu version?
[22:30] <llogan> oh, mint
[22:30] <admalledd> llogan: :< that makes me a sad panda... blarg. welp thanks for the point in the right direction. maybe in the docs point out what version an option was added at?
[22:31] <llogan> it's easy to compile and shouldn't take too long. i even updated the guide recently so it performs a non-intrusive local install
[22:31] <admalledd> ubitux: meh, too slow would prefer to have this working by the end of the day. as well as i would be told to update to mint 15 or such
[22:31] <ubitux> it's likely mint 15 is still packaging the fork anyway, but well
[22:32] <admalledd> ubitux: yea, welp time for me to run and compile ffmpeg. I guess ill get some noms whilst my computer melts...
[22:32] <llogan> 13 is based on 12.04 and 15 is based on 13.04
[22:33] <llogan> i should do something about those old ubuntu guides...
[22:33] <llogan> or merge them. or make one big linux guide
[22:35] <llogan> the PPA author never got back to me any time i ever tried to contact him
[23:40] <admalledd> llogan: <3 thanks got my stuff working now with the overlay and a custom build of ffmpeg.
[00:00] --- Fri May 31 2013
More information about the Ffmpeg-devel-irc
mailing list