Ffmpeg-devel-irc

Download

ffmpeg-devel-irc@ffmpeg.org

April 2014

1 participants
60 discussions

ffmpeg-devel.log.20140424
by burek 25 Apr '14

25 Apr '14

[00:10] <michaelni> maybe someone should add a check to configure to test for 4.9.0 and print a big warning or fail [00:13] <kurosu> michaelni, regarding your change on fic idct, you may as well merge loops for horizontal idct and clip+put [00:13] <kurosu> not sure how to do that without code duplication and cleanly [00:14] <michaelni> yes, thought about it too and thats was also why i didnt do it [00:15] <michaelni> but if speed over clarity is preferred for fic i can surely do it [00:15] <kurosu> no opinion [00:15] <kurosu> this was more rethorical than of practical use [00:15] <kurosu> I don't even know that codec and how useful it is [00:17] <michaelni> the idct could also be optimized with conditional branches similar to the jpeg idct [00:17] <michaelni> the higher coeffs should often be 0 [00:18] <kurosu> iirc, BBB optimized larger vp9 transforms by providing codepaths for some max count of coeffs [00:19] <kurosu> makes even more sense, considering inter blocks [00:19] <michaelni> and as you mention mmx, the idct could of course be done in mmx [00:19] <michaelni> with idct + transpose + idct [00:20] <kurosu> at that point, of higher priority would be to provide >=mmx implementation of vc1 transforms [00:23] <kurosu> I'm not completely decided on the vc1 mc functions - main purpose would be to yasmify it but time + lambda*benefit suggests selecting skip instead [00:23] <michaelni> lol [00:25] <kurosu> anyway, good night [00:26] <michaelni> good night to you too [01:51] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:291d464161a5: swresample: fix AV_CH_LAYOUT_STEREO_DOWNMIX input [03:24] <BBB> kurosu: yeah it helps, particularly for bigger transforms... not sure if fic is worth optimizing extensively for (then again, is vp9? :-p), I guess I don't know what fic is... [03:44] <jamrial> BBB: speaking of vp9, vp9dsp.h states that dst/left/top pointers are aligned by transform-size, but then the latter two are defined in vp9.c using LOCAL_ALIGNED_16() [03:44] <jamrial> intra_pred, i mean [03:45] <BBB> maybe they should've been aligned by 32 and I forgot [03:45] <jamrial> I tried to write a quick avx2 version for some 32x32 functions and it crashes because they are not aligned [03:45] <BBB> please do send a patch, I believe that's kinda important [03:45] <BBB> right [03:45] <BBB> avx2 work is awesome btw [03:46] <BBB> anyway, consider it a bug, and feel free to send a patch, I'm obviously fine with changing it to LOCAL_ALIGNED_32, just an oversight [03:46] <cone-551> ffmpeg.git 03YuDenzel 07master:443936d8b999: configure: Fix ld flags when rpath is enabled. [03:47] <jamrial> ok [03:54] <jamrial> it also crashed with dst for the same reason [03:54] <jamrial> couldn't find where that one is defined, though [03:55] <jamrial> then again i may be doing something wrong [05:28] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:31f2357fd0e2: avdevice/qtkit: fix include [09:09] <cone-441> ffmpeg.git 03Michael Niedermayer 07release/2.1:8d6de877094f: swresample: fix AV_CH_LAYOUT_STEREO_DOWNMIX input [09:09] <cone-441> ffmpeg.git 03Michael Niedermayer 07release/2.2:4f41717d0100: avformat/mux: Check for and remove invalid packet durations [09:09] <cone-441> ffmpeg.git 03Michael Niedermayer 07release/2.2:4a479fd3e6a7: swresample: fix AV_CH_LAYOUT_STEREO_DOWNMIX input [09:41] <ubitux> BBB: does make -j5 fate-vp9-00-quantizer-01 THREADS=2 works for you? [09:41] <ubitux> i have a systematic fail here [09:42] <ubitux> seems it's present since a while [12:12] <wm4> where does crap like "avpicture_fill((AVPicture*)in_frame, data, PIX_FMT_YUV420P, width, height);" come from? [12:12] <wm4> I see it all the time and I know it can cause actual problems [12:12] <wm4> some tutorial? an example? [12:13] <wm4> it was probably the standard way pre-TEP [13:33] <BBB> ubitux: will have to recompile, that might take a while [13:38] <BBB> ubitux: yes that fails [13:43] <BBB> ubitux: the visible pixels seem identical according to glyuvplay [13:43] <BBB> I heard somebody on some ML (maybe libav-devel) complaining also that vdpau tests fail because we're comparing invisible pixels - are we doing that? [13:44] <BBB> or maybe it was dxva2, can't remember [13:44] <BBB> anyway [13:44] <BBB> the visible pixels are identical, so there's not much I can do beyond that [13:45] <BBB> hm hexdiff says they're different [13:45] <BBB> wait a minute [13:45] <BBB> is glyuvplay buggy? [13:45] <BBB> :( [14:18] <BBB> hm it's chroma-only, that's odd [18:56] <Mista_D> what web server modules are needed to accept "-progress http://..." please? The mailing list and manual have no info on that yet. [21:35] <cone-441> ffmpeg.git 03Luca Barbato 07master:374fdc8c071d: flv: Improve log messages [21:35] <cone-441> ffmpeg.git 03Michael Niedermayer 07master:277ae5a7b304: Merge commit '374fdc8c071dcd96422378b0a1a0d453336d8a01' [21:45] <cone-441> ffmpeg.git 03Luca Barbato 07master:5d983fdbca55: flv: Warn only once [21:45] <cone-441> ffmpeg.git 03Michael Niedermayer 07master:7da5ff5c18ae: Merge commit '5d983fdbca5570a1545a892583a372cfb3fffe92' [22:05] <cone-441> ffmpeg.git 03Luca Barbato 07master:152b797cd687: flv: Do not mangle dts values for negative cts [22:05] <cone-441> ffmpeg.git 03Michael Niedermayer 07master:ca9cc6fdcc78: Merge commit '152b797cd687e96a582a1cb908dddf3d330d7637' [23:01] <cone-441> ffmpeg.git 03Derek Buitenhuis 07master:d66de5006bbc: fate: Add fic-in-avi test [00:00] --- Fri Apr 25 2014

1 0

ffmpeg.log.20140424
by burek 25 Apr '14

25 Apr '14

[00:00] <sacarasc> Bray90820: -ss 2 as either an input or output option (forget which does the accurate seeking) should do it. [00:00] <blippyp> yes, but it should still be keeping the right aspect no? [00:00] <xreal> blippyp: no, it *has* to be distorted. [00:00] <Bray90820> blippyp: what do you mean [00:00] <blippyp> it's not the same as resizing the video to 16:9 [00:00] <blippyp> yes - distorted, but not stretched [00:00] <xreal> blippyp: but that's what they did, I think. [00:01] <blippyp> the concept works better if you had a device like my camera you might understand what I mean more [00:01] <xreal> blippyp: they just have put the 16:9 signal on a 4:3 DVD. [00:01] <blippyp> probably because you already degraded it I guess [00:01] <blippyp> I figured it had an algorithm to help resize without destroying the image [00:01] <xreal> blippyp: it's not my fault :) [00:02] <Bray90820> sacarasc: i don't really know what you mean by that [00:02] <blippyp> no - maybe you need to switch your setsar and setdar - did you try that - I'd give it a go just to make sure [00:02] <blippyp> it's a confusing concept [00:02] <c_14> Bray90820: https://trac.ffmpeg.org/wiki/Seeking%20with%20FFmpeg [00:02] <c_14> Read, learn, profit. [00:03] <xreal> Nice... handbreak doesn't support distortion/stretching. So I *need* to use ffmpeg anyway :) [00:04] <blippyp> Bray90820: keep in mind though with that link c_14 gave you - as soon as you implement accurate seeking - you will not simply splice the video - it will need to re-encode it [00:07] <Bray90820> blippyp: could I just put it in a different container [00:07] <blippyp> the container doesn't matter - as soon as you implement accurate seeking it will re-encode the entire video your making [00:08] <blippyp> some codecs might not even allow for quick splicing regardless [00:09] <blippyp> geez guys - went through the trouble to make a video of my cartoon example and not a single comment??? I promise - I take constructive criticism well.... [00:10] <Bray90820> Nevermind i found a different way to do it [00:10] <blippyp> Bray90820: how's that? [00:10] <Bray90820> When I import it i can choose custom start and ending times [00:11] <blippyp> import it into what? what software are you using? [00:11] <Bray90820> I am actually using dvd shrink on windows [00:11] <c_14> blippyp: the degradation makes the snow in the bottom vid turn into cobwebs [00:12] <blippyp> oh - so you were going to re-encode it anyways then... [00:12] <Bray90820> Well no dvd shrink was going to copy it over with no compression [00:12] <Bray90820> as a vob [00:14] <blippyp> c_14: not sure I'm understanding what you mean? [00:14] <Bray90820> Basically dvd shrink is just removing the encryption then copying the files over [00:15] <blippyp> like it just looks like cobwebs instead of snow? [00:15] <Bray90820> Basically dvd shrink is just going to remove the encryption then copy the files over [00:16] <c_14> Due to the distortion, the snow takes on the appearance of cobwebs. L [00:17] <c_14> With the way it's spread out and the holes. [00:17] <sacarasc> I don't l;ike DVD shrink. I use a combination of RipIt4Me and DVD Decrypter. [00:17] <blippyp> yeah, there isn't much I can do about that... with a filter like this you kind of need to 'tweak' it as you use it - basically use different settings for each scene. [00:18] <Bray90820> sacarasc: dvd shrink seems to be the best i can find [00:18] <blippyp> I also seriously decreased the video size to save on time - If I apply the filter to a higher resolution, it generally turns out much better - then I take that higher resoultion image and resize it down - I haven't tested it yet, but I'm sure that resizing the cartoon down ends up in a much better video... The more quality you use to begin with makes a much better drawing [00:19] <sacarasc> Get RipIt4Me and DVD Decrypter set up and it's 1 button ripping. [00:19] <Mista_D> can't see http traffic when using `-progress http://127.0.01/` http://pastebin.ca/2701710 [00:19] <Bray90820> sacarasc: what format are the videos in [00:19] <sacarasc> IFO and vobs. [00:20] <Bray90820> Can it keep the man feature in one file or does it split it into 1GB files [00:20] <sacarasc> It splits them, this is for archival, I guess. [00:21] <sacarasc> I think you can make it merge them, though. [00:22] <Bray90820> sacarasc: with dvd shrink i can keep them in one file and i need that because plex doesn't recognize Video_TS [00:22] <sacarasc> You can rip it as an ISO if you want, with it, too. [00:22] <blippyp> Mista_D: I've never used that - but I gotta ask the obvious - you must have something serving on port 80 ready to read that data? [00:23] <Mista_D> blippyp: its just a listenning port, there's nginx. Listener is not setup yet, just want to see the format of the data. [00:23] <Bray90820> sacarasc: Well i rip tv shows and it's much easier to select indevigual episodes when it's not one file and the metadata is pulled as well [00:24] <blippyp> I can't find any documentation on that parameter except in the man page... [00:24] <Mista_D> blippyp: same here... [00:24] <blippyp> Well there would need to be a server ready to receive that data on port 80 to read it.... and then spit it back out onto the page.... [00:25] <blippyp> Unless it's suppose to serve itself? I know ffserver will do that... [00:25] <Mista_D> blippyp: server it there, using NGINX. FFmpeg wont start unless it handshakes with HTTP server. [00:25] <blippyp> so there is a 'connection' happening that you're aware of? [00:29] <blippyp> like you're just feeding data to the sever, but the server has to have 'code' there ready to receive and spit that data back out.... something that obvioulsy can use sockets (like php) [00:29] <blippyp> *obviously [00:29] <Mista_D> blippyp: got it, had to use "-i any" in tcpdump, there's nothing setup on nginx but traffic was going to it... [00:30] <blippyp> so you're seeing the progress information on http://127.0.0.1 now? [00:30] <blippyp> that's weird? I don't get it [00:31] <xreal> man, I just figured it out! [00:31] <Mista_D> need to setup the web server to handle it now. But FFmpeg progress works. [00:31] <xreal> avconv is completly behaving different than ffmpeg. [00:31] <blippyp> oh I see [00:31] <blippyp> where did you read about the 'any' option? [00:31] <xreal> avconv: scale=iw*2:ih/2 => distortion as expected | ffmpeg: just scaling [00:31] <Mista_D> blippyp: had too many interfaces on the box, was not monitoring the right interface... [00:32] <blippyp> oh you're using avconv - I don't think ffmpeg has that - I don't see it in the man page anyways... [00:32] <Mista_D> blippyp: `tcpdump -D` [00:33] <xreal> blippyp: no, I am using ffmpeg. ffmpeg *had* it, there are many tutorials. [00:33] <xreal> I think, they removed it after forking. [00:33] <xreal> setdar/setsar doesn't work in avconv :) [00:33] <xreal> but it works in ffmpeg [00:33] <sor_> anyone no how to get fflpay to keep updatting source file (refreash) [00:34] <blippyp> Mista_D: you're a networker? [00:34] <blippyp> never used tcpdump before - neat [00:34] <Mista_D> blippyp: not really. but it is a neat tool. [00:35] <blippyp> well clearly you know a few things - do you sniff packets? and protocols on the network? [00:35] <Bray90820> Would this work to convert the file to flac [00:35] <Bray90820> ffmpeg -i '/home/aaron/Desktop/VIDEO_TS/file.VOB' -metadata TITLE=file file.flac [00:36] <Mista_D> blippyp: sometimes use Wireshark, but its not FFmpeg related, usually live streaming troubleshooting. [00:37] <blippyp> cool - I've use wireshark (I suck at it) but I've used it... [00:39] <Bray90820> So would the script i sent work? [00:41] <blippyp> not sure - try it - you might need to add -c:a flac before the final filename??? (not sure, never used it - but I'm looking at an example, and that's all they've done) [00:42] <Bray90820> It keeps telling me ffmpeg not found but i correctly used the script a week ago [00:42] <blippyp> well - it's not in your path now - you did something with it [00:43] <blippyp> either find it and add it to your path or go to wherever it is and run your script there [00:43] <Bray90820> Well pretty much the only thing i did was update ubuntu from 13.10 to 14,04 [00:43] <blippyp> oh, then it should be in your path I would think [00:44] <Bray90820> Then i have no idea what's going on [00:44] <blippyp> try this [00:44] <blippyp> run sudo updatedb [00:45] <blippyp> it will take a second if you have it installed [00:46] <blippyp> it will either error or seem like nothing happened, just kind of paused [00:47] <Bray90820> that seemed to not do much of anything [00:47] <Bray90820> http://pastebin.com/raw.php?i=v1rP8eGV [00:47] <blippyp> did it error though? [00:47] <llogan> sor_: you've asked that at least 5 times. try the ffmpeg-user mailing list. [00:47] <Bray90820> The link has the terminal output [00:47] <blippyp> yup that's good [00:48] <blippyp> you have it [00:48] <blippyp> now run: locate ffmpeg [00:49] <blippyp> locate ffmpeg|grep bin - that might be better [00:49] <blippyp> find it? [00:50] <blippyp> it's probably in /usr/bin [00:50] <blippyp> or at least should be [00:51] <llogan> Bray90820: 14.04 does not supply ffmpeg or the "ffmpeg" counterfeit, AFAIK. [00:52] <blippyp> so it didn't find it? [00:52] <blippyp> oh [00:52] <blippyp> that's right you're on ubuntu [00:52] <blippyp> you want avconv I think [00:52] <llogan> no [00:53] <llogan> or http://trac.ffmpeg.org/wiki/UbuntuCompilationGuide [00:53] <blippyp> haha [00:53] <blippyp> the man has spoken [00:53] <Bray90820> llogan: Even if the libav tools package? [00:54] <llogan> if you want to use third part tools and forks you'll have to get help at their help resources [00:54] <llogan> *third-party [00:54] <Bray90820> Hum [00:55] <Bray90820> then do you have another script that will convert to flac without loosing audio quality [00:55] <llogan> the static build is easy. just download, extract, and run. in your script supply the full path to the new ffmpeg [00:55] <llogan> why do you want flac? [00:55] <llogan> (sorry, i didn't completely follow the discussion) [00:55] <blippyp> it's all the 'rage' with these kids these days... ;) [00:56] <Bray90820> llogan: i am converting a dvd to a loseless audio file [00:56] <llogan> "ffmpeg -i input output.flac" is the most basic command [00:57] <Bray90820> llogan: ffmpeg isn't available in 14.04 as you mentioned [00:57] <blippyp> so basically you had it right Bray - but you don't have ffmpeg, which is your problem [00:58] <Bray90820> Well the libav tools package is installed [00:58] <llogan> what's that have to do with ffmpeg? [01:00] <blippyp> Bray did you download the static build he linked you to yet? [01:00] <Bray90820> Doesn't the libav tools package include ffmpeg [01:00] <Bray90820> blippyp: was just about to [01:00] <llogan> not anymore. it used to provide a fake ffmpeg from a fork. [01:01] <llogan> http://blog.pkh.me/p/13-the-ffmpeg-libav-situation.html [01:01] <Bray90820> ok [01:01] <Bray90820> How would i install teh static ffmpeg [01:02] <llogan> you don't need to install it [01:02] <llogan> just download, extract, and execute [01:02] <Bray90820> Well then how would i put it in my path so the script will work [01:03] <llogan> /full/path/to/ffmpeg -i input ... output [01:03] <blippyp> just copy it to your /usr/bin folder [01:03] <Bray90820> ok [01:07] <Bray90820> Thanks [01:07] <Bray90820> It works now [01:28] <xreal> Does anyone know a demo DVD, which contains subtitles? [15:13] <superc2_> hi. just trying to install a very basic avserver and followed the howtos... but I'm not able to see anything when opening a stream. [15:13] <superc2_> how can I see whats going wrong or missing? [15:15] <klaxa|work> already sounds like you are using avconv and not ffmpeg, in that case see #libav [15:15] <superc2_> so avserver is running... and I used 'avconv -i file.mp4 http://...:8090/test1.ffm' to stream the file to avserver [15:16] <klaxa|work> see #libav for avconv support [15:16] <superc2_> ah, I was not sure there was a difference... I tried the same with ffserver first... but didn't work either [15:17] <klaxa|work> unless you compiled it yourself, installed a different package actually packing ffmpeg or downloaded a static build, you are still using avconv [15:34] <blippyp> klaxalwork: Good read - I had no idea they were such dicks, I've been aware of the confusion of the fork for a while - But I can't believe they still took the name Libav after all that - That's just mind-boggling.... [15:35] <superc2_> mh. actually I was just looking for a small streaming solution... avserver was installed with the debian package 'ffmpeg'... very confusing [15:36] <blippyp> superc2_: it is confusing to those who aren't aware - it's understandable... [16:56] <droetker> Hello. I am trying to download a ism/manifest. How do I download the second best quality and not the best. (qualitylevel index 1 instead of 0) [16:57] <sacarasc> I may be missing something, but how is this related to ffmpeg? [16:57] <droetker> i am trying to convert the ism manifest to mp4/mkv using mkv [16:58] <droetker> using ffmpeg* [16:58] <klaxa|work> looking at google it yields some silverlight results [16:59] <droetker> yes, ism manifests are microsoft iis streams. [16:59] <droetker> Playing in silverlight/moonlight. [17:04] <klaxa|work> droetker: what have you been doing so far? [17:13] <droetker> ffmpeg -i ismlink -vcodec copy -acodec copy output.mkv [17:14] <droetker> ismdownloader downloads the file, but can only do so in the highest quality available. [17:14] <klaxa|work> and that works for the highest quality? [17:14] <droetker> Yes. [17:14] <klaxa|work> is ismdownloader a different tool? [17:14] <droetker> yes [17:16] <klaxa|work> so ffmpeg generally supports the manfest files? [17:19] <droetker> hm [17:20] <droetker> the other file was ismv [17:20] <droetker> this is .ism/manifest [17:20] <droetker> MAybe ffmpeg just doesnt support manifests. [17:25] <klaxa|work> ismv is with payload already, it contains an actual video stream [17:25] <klaxa|work> at least it appears so [17:25] <klaxa|work> yes pretty sure [17:25] <klaxa|work> i downloaded some files from http://playready.directtaps.net/smoothstreaming/ [17:26] <klaxa|work> the .ismv files contain the videos, the Manifest most likely contains "links" to those .ismv files [17:35] <klaxa|work> also, apparently ffmpeg cannot parse the manifest file (i would have been quite surprised if it could) [17:35] <klaxa|work> >data/Manifest: Invalid data found when processing input [18:26] <Voicu> hello, can someone explain how pts' and timebases work? [18:27] <Voicu> if I set the timebase of a video stream to 1/30 (i.e. 30 fps) what do I need to set the pts values to ? [18:27] <Voicu> is it microseconds or microseconds / 30 or what? [18:39] <Jonas_> Does somebody know how I can speed my video? [18:44] <c_14> Jonas_: https://trac.ffmpeg.org/wiki/How%20to%20speed%20up%20/%20slow%20down%20a%20… ? [18:47] <Jonas_> I get the error: unrecognized option 'filter:v' [18:58] <Brolan> I am trying to convert a '.AA' file into something useable. I have tried [-i book.aa book.mp3] but that has only done pic related: http://i.4cdn.org/adv/1398358300240.jpg [18:59] <Brolan> something about unable to find codec parameters [19:07] <c_14> Brolan: I'm pretty sure ffmpeg does not support the audible audio format. [19:07] <c_14> Especially since it usually has DRM. [19:08] <Brolan> ah. Thanks I'll figure it out elsewhere then. Thanks though! [19:08] <c_14> And Jonas_ the pastebin was for you. [19:08] <c_14> s/pastebin/pastebin request [19:09] <Jonas_> heres my console output: http://pastie.org/9108687 [19:10] <c_14> You are using libav. libav support is in #libav [19:10] <blippyp> Jonas_ use -vf instead [19:11] <Jonas_> -vf instead of -filter:v ? [19:11] <blippyp> yup [19:12] <Jonas_> ok this works [19:12] <Jonas_> thanks [19:13] <blippyp> np [19:26] <blippyp> c_14: just curious - but why would you send Jonas to #libav for a legitimate ffmpeg question? am I missing something? [19:27] <c_14> His pastie lines 2-4 [19:28] <blippyp> ah - I get it - it's a falsie... ;) [19:29] <blippyp> should have have noticed the obiouvs deprecated notice - my bad [19:30] <c_14> Nah, it's fine. I would have helped him, but -filter:v worked on my end. [19:31] <blippyp> never used it that way - always -fv or -complex_filter (although there's a much shorter version for that one but I never remember it for some reason) [19:31] <blippyp> is -filter:v simple or complex? [19:56] <ross_> Hey everyone [19:56] <ross_> I have a newbie question if anyone can help..im running ffmpeg on windows, I'm no sure how to add the current time to the output filename [19:58] <Mavrik> you'll have to probabl use batch variable / script for that [19:58] <Mavrik> something like http://stackoverflow.com/questions/1192476/windows-batch-script-format-date… [19:58] <ross_> didint ffmpeg had option like %t for time? [19:59] <klaxa> -t [19:59] <ross_> The problem is that I'm segmenting live stream into chunks of 1hr each, and I need the time for that [19:59] <klaxa> oh dunno bout that [20:00] <Mavrik> ross_, hmm [20:01] <ross_> Mavrik, my other option is to loop and call ffmpeg everytime and set the time myself using batch, but I thought ffmpeg has something built in [20:02] <ross_> plus it can result in data loss between transition which I dont wanna do [20:02] <Mavrik> um [20:02] <Mavrik> I don't really understand your issue [20:02] <Mavrik> you want to change filename... without restarting ffmpeg_ [20:02] <Mavrik> ? [20:03] <ross_> i'm using -f segment to save a livestream, i want to save it as play_2014_24_4_2_03_44.mpg insteaed of play_004.mpg [20:03] <Mavrik> ah, ok [20:03] <ross_> instead of counters, I want to use system time as filenames [20:03] <Mavrik> I've never used HLS segmenter, so you'll have to wait for someone else :) [20:04] <ross_> kk, thanks [20:08] <Mavrik> looking at source it seems timestamp is only implemented in image output [20:08] <Mavrik> but I haven't looked at HLS :- [20:08] <Mavrik> / [20:11] <ross_> ohh interesting [20:12] <ross_> the segment doucmentation states that it uses the same output filename pattern as image2 [20:12] <ross_> do you know how I would set it in case of an image? [20:12] <ross_> https://ffmpeg.org/ffmpeg-formats.html#segment_002c-stream_005fsegment_002c… [20:42] <blippyp> ross: I think you want to do something like this? ffmpeg -i video.mp4 test/img%4d-${DATE=$(date +"%Y_%m_%d_%H_%M")}.png [20:44] <blippyp> assuming you're in linux... [20:48] <ross_> Hi blippyp [20:48] <ross_> im on windows [20:49] <blippyp> not sure if you can do it in windows - give ma second [20:50] <blippyp> check this out: http://stackoverflow.com/questions/7727114/batch-command-date-and-time-in-f… [20:51] <blippyp> same idea [20:56] <ross_> hmmm [20:56] <ross_> thanks for helping out blippyp, but the problem is that i want to use this with -f segment [20:57] <ross_> what I mean is, i want the segmented 1 hour long videos to have the corresponding time values [20:58] <blippyp> then why did you post the date and not just the time? I don't get it? [21:02] <klaxa> blippyp: the problem is different from what you think [21:02] <klaxa> your command will change the name on shell level [21:02] <klaxa> he needs to change the name according to the time/date on ffmpeg's output-file level [21:03] <blippyp> doesn't the segmenter break apart the files in chunks - and he wants to name the files according to the time that segment ended I guess...??? [21:03] <blippyp> what would a date have anything to do with that? [21:04] <klaxa> > i want to save it as play_2014_24_4_2_03_44.mpg insteaed of play_004.mpg [21:07] <blippyp> I don't know - I'm confused - I gotta go though - good luck ross, sorry I couldn't help ya... [21:07] <ross_> haha np thanks [21:07] <ross_> i wanna do something like play_%H_%m_%s.mpg as output file [21:07] <ross_> for a segmented output [22:02] <xreal> "-vf scale=" always keeps ratio in ffmpeg... what's wrong here? [22:07] <klaxa> nothing? [22:13] <xreal> klaxa: it keeps ratio here... but it shouldn't. [22:14] <klaxa> >The scale filter forces the output display aspect ratio to be the same of the input, by changing the output sample aspect ratio. [22:14] <klaxa> http://ffmpeg.org/ffmpeg-filters.html#scale-1 [22:16] <xreal> force_original_aspect_ratio=disable :D [22:53] <albator> why do i have segfault when i concat with files including a ' in the filename? [22:54] <c_14> Are you escaping correctly? [23:07] <haptiK> Hello. I'm sure this has been asked a million times, but after searching google for a few hours I'm drawing a blank: Unknown encoder 'libx264'. Could someone point me in the right direction? (centos6) thanks. [23:10] <haptiK> I followed the following guide: https://trac.ffmpeg.org/wiki/CentosCompilationGuide [23:11] <iive> haptiK: when you run ffmpeg configure, do you see libx264 as successfully detected ? [23:12] <c_14> Or what does ffmpeg -encoders | grep libx264 give you? [23:15] <haptiK> http://pastebin.com/K4EKQfTp [23:15] <haptiK> i compiled with --enable-libx264 [23:15] <haptiK> and x264 is installed [23:18] <c_14> ffmpeg isn't registering x264 as an encoder for whatever reason. Do you have configure output? [23:18] <haptiK> yes one second [23:29] <haptiK> http://paste.ubuntu.com/7325391/ [23:29] <haptiK> thats the output from the entire build script (sorry for delay, couldn't find a pastebin that would accept all that text) [23:36] <c_14> Do you have the script (I think this is a script managing it) somewhere? Because if you can see at line 7117 Enabled encoders: it never lists libx264. [23:37] <haptiK> you're asking for the script i threw together ? [23:38] <c_14> Ye. [23:38] <haptiK> its basically just all the commands throwin in from https://trac.ffmpeg.org/wiki/CentosCompilationGuide [23:38] <haptiK> thrown in* [23:38] <haptiK> one sec [23:39] <haptiK> http://pastebin.com/G54H5KJ3 [23:40] <haptiK> i cut off some stuff from the header, assume $DIR means something [23:40] <c_14> Add a \ after ./configure [23:40] <c_14> Line 81 [23:41] <haptiK> oh ffs are you kidding me [23:41] <haptiK> *facepalm* [23:41] <blippyp> haha [23:41] <c_14> And I'd get rid of the \ on line 98 [23:41] <blippyp> happens to the best of us... ;) [23:41] <c_14> 93 [23:41] <haptiK> okay rerunning the script [00:00] --- Fri Apr 25 2014

1 0

ffmpeg-devel.log.20140423
by burek 24 Apr '14

24 Apr '14

[00:01] <cone-647> ffmpeg.git 03Derek Buitenhuis 07master:40b331e1f41c: fic: Use proper quantization matrix index [00:01] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:83412f00e98f: Merge commit '40b331e1f41cf118bd1b0807cac801437255636f' [00:03] <michaelni> Daemon404, what issues ? [00:04] <Daemon404> https://lists.libav.org/pipermail/libav-devel/2014-April/059146.html [00:05] <Daemon404> its not essential [00:05] <Daemon404> i did something wrong there, so ill fix it up and send as a patch tomorrow [00:05] <Daemon404> to both lists [00:21] <michaelni> Daemon404, ill leave the cursor change to you to fix up, the libav commit is missing things that you replied to with "done" on ffmpeg-devel [00:22] <Daemon404> yes [00:22] <Daemon404> that was the accident [00:22] <Daemon404> my branches got mangled and i accidentally pushed it [00:23] <cone-647> ffmpeg.git 03Derek Buitenhuis 07master:9155c595f1a9: fic: Support rendering cursors [00:23] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:4b6bb7bfe18f: Merge commit '9155c595f1a9c5a089a59e2c0aefdfabd0a6c59b' [00:24] <michaelni> Hi saste [00:25] <michaelni> if you want / have time there are at least 2 ffprobe patches on ffmpeg-devel [00:25] <saste> michaelni, hi [00:25] <saste> michaelni, yes, i'm on it [00:25] <michaelni> 0421 13:06 To FFmpeg devel (1.1K) [FFmpeg-devel] [PATCH] ffprobe: fix scaling of vali [00:25] <michaelni> 0421 18:38 Nicolas George (5.8K) [FFmpeg-devel] [PATCH] ffprobe: add -show_data_hash option. [00:25] <michaelni> ahh ok, thanks [00:33] <Daemon404> meh, i guess i wont update gcc to 4.9 [00:33] <Daemon404> how lame. [00:47] <cone-647> ffmpeg.git 03Derek Buitenhuis 07master:502512ea0f17: avutil: Add av_get_time_base_q() [00:47] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:c2cd0945c9ae: Merge commit '502512ea0f179165782777265f441d4061ff6012' [00:59] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:b0bdc2a7123a: avutil: move av_get_time_base_q() after include rational.h [00:59] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:d3f3035f0857: Merge commit 'b0bdc2a7123abd0287de435726ab68e9e475a8d8' [00:59] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:e79b15f5f1ae: avutil/avutil.h: remove duplicate rational.h include [06:57] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:b9017fdd426d: avformat/flvdec: also include file position in debug output [06:57] <cone-703> ffmpeg.git 03Michael Niedermayer 07master:dc6a17cf74a9: avformat/mux: Check for and remove invalid packet durations [08:19] <anshul_> what is page composition in dvb [10:58] <cone-551> ffmpeg.git 03Stefano Sabatini 07master:bf411fffd453: examples: rename avcodec.c to decoding_encoding.c [10:58] <cone-551> ffmpeg.git 03Stefano Sabatini 07master:1ba59b1cbeaf: ffprobe: fix scaling of vali in value_string() in case -prefix is selected [15:44] <cone-551> ffmpeg.git 03kyh96403 07master:f21c263c8979: avformat/rtmppkt: Fix random crash [16:41] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:c161134035a0: avcodec/snow_dwt: use av_malloc(z)_array() [16:41] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:7de2cea8a9ea: avcodec/motionpixels: use av_mallocz_array() [16:41] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:1aa388d4397e: avcodec/jpeg2000dwt: use av_malloc_array() [17:26] <cone-551> ffmpeg.git 03Michael Niedermayer 07release/1.2:1530fbccac86: avformat/avidec: Speed up keyframe detection code [17:26] <cone-551> ffmpeg.git 03Stefano Sabatini 07release/1.2:0759bdc31549: ffprobe: fix scaling of vali in value_string() in case -prefix is selected [17:26] <cone-551> ffmpeg.git 03Michael Niedermayer 07release/2.1:116b959565d3: avformat/avidec: Speed up keyframe detection code [17:26] <cone-551> ffmpeg.git 03Stefano Sabatini 07release/2.1:af04f0d1ce70: ffprobe: fix scaling of vali in value_string() in case -prefix is selected [17:26] <cone-551> ffmpeg.git 03Michael Niedermayer 07release/2.2:0385c824f101: avformat: Fix decoder search in find stream info [17:26] <cone-551> ffmpeg.git 03Michael Niedermayer 07release/2.2:abd6decd554d: avformat/pmpdec: remove cur_dts timestamp hack [17:26] <cone-551> ffmpeg.git 03Michael Niedermayer 07release/2.2:e9e42beed2ac: avformat/asfenc: dont allow non interleaved packets [17:26] <cone-551> ffmpeg.git 03Stefano Sabatini 07release/2.2:14404170b959: ffprobe: fix scaling of vali in value_string() in case -prefix is selected [17:26] <cone-551> ffmpeg.git 03Michael Niedermayer 07release/2.2:6896dcbf5f1f: avformat/avidec: Speed up keyframe detection code [18:31] <Daemon404> does ffmpeg actually build on riscos? [18:55] <nevcairiel> Daemon404: there appears to be a "port", so not natively i guess [19:13] <jamrial> michaelni: it seems the fate tests for linear interpolation resample are using the wrong codepath [19:13] <jamrial> you can break the asm code in LINEAR_CORE_FLT_SSE and fate will still pass [19:14] <jamrial> looks like linear_interp=1 is in the wrong place in the test's filter chain [19:15] <nevcairiel> position of options is always just so wtfy [19:31] <cone-551> ffmpeg.git 03Derek Buitenhuis 07master:14da7f9eb7d7: fic: Support rendering cursors [19:40] <jamrial> yeah, -af and -vf filter chains tend to be a headache :P [20:32] <michaelni> Daemon404, are you fic maintainer or not? [20:32] <Daemon404> yes [20:32] <Daemon404> why? [20:32] <Daemon404> i even added myself to MAINTAINERS [20:32] <michaelni> then why are there tons of changes in libav that arent in ffmpeg ? [20:33] <Daemon404> they are [20:33] <Daemon404> theyre squashed into the final patch [20:33] <Daemon404> because they were supposed to be in teh first place [20:33] <Daemon404> it's all the stuff you requested [20:34] <Daemon404> how doesnt that test pass [20:34] <Daemon404> it passes with ffmpeg HEAD here [20:35] <michaelni> hmm, seems i forgot a git pull --rebase [20:35] <michaelni> sorry [20:41] <michaelni> about the fic patch(es) its a bit confusing to have 4 patches in libav and them all sqashed in one in ffmpeg [20:41] <michaelni> doesnt matter much though [20:42] <llogan> what does FIC stand for? I didn't find answer in my 3 seconds of lazy searching [20:44] <cone-551> ffmpeg.git 03Derek Buitenhuis 07master:e299cb2cd373: fic: Simplify alpha blending [20:44] <cone-551> ffmpeg.git 03Derek Buitenhuis 07master:7596fc3d4b61: fic: Remove redundant clips [20:44] <cone-551> ffmpeg.git 03Derek Buitenhuis 07master:d7eb8f903338: fic: Make warning message more accurate [20:44] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:ecb815e5117f: Merge commit 'd7eb8f903338048c0b222d92357d67f5d3b54295' [20:46] <Daemon404> [19:41] <@michaelni> about the fic patch(es) its a bit confusing to have 4 patches in libav and them all sqashed in one in ffmpeg <-- they were supposed to be squashed in libav too [20:46] <Daemon404> but i screwed up and pushed it too early [20:46] <Daemon404> and had to fix it [20:46] <Daemon404> llogan, no idea. its used my mirillis [20:56] <cone-551> ffmpeg.git 03Kostya Shishkov 07master:e2834567d73b: On2 AVC decoder [20:56] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:56efd98a6574: Merge commit 'e2834567d73bd1e46478ba67ac133cb8ef5f50fd' [21:35] <jamrial> michaelni, what i mentioned above [21:56] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:0bfb1e9238d8: avdevice/jack_audio: use av_malloc_array() [21:56] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:2c4862e7a131: avfilter/avcodec: Use av_mallocz_array() [21:56] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:fc145e576a44: avformat/h263dec: Fix h263 probe [22:07] <michaelni> jamrial, issue confirmed, ill push a fix in a moment [22:09] <jamrial> cool, thanks [22:15] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:c26b4b6f55b0: fate/libswresample: use linear interpolation on both aresample filters [22:37] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:23cd5cb18f55: avcodec/fic: avoid 2 additions per idct row [22:47] <wm4> wat [22:57] <jamrial> btw, fix for the flac golomb miscompilation is out http://gcc.gnu.org/git/?p=gcc.git;a=commitdiff;h=bb3a8839965c29e7f83fcba20d… [22:57] <Daemon404> jamrial, 1 day after gcc release [22:57] <Daemon404> how very useful [22:57] <jamrial> Unfortunately [22:58] <Daemon404> god, gcc code scares me [23:00] <jamrial> what sucks is that we have had a fate system with 4.9.0 prerelease failing these tests for a good while now, and nobody noticed until it was too late [23:00] <Daemon404> we tend not to put too much effort int oprerelease gcc [23:00] <Daemon404> since it is pretty much always broken [23:02] <wm4> Daemon404: what scares me about this commit is there changelog duckery [23:02] <wm4> Daemon404: and they seem to have a separate changelog for the testsuite [23:02] <wm4> although it's in the same repo [23:02] <wm4> ...wat [23:11] <cone-551> ffmpeg.git 03Kostya Shishkov 07master:7cade8ea2bb1: on2avc: change a comment at #endif to match actual define [23:11] <cone-551> ffmpeg.git 03Michael Niedermayer 07master:417ae6281b9b: Merge commit '7cade8ea2bb19e78dae42b29720535a70fb2ae84' [23:14] <jamrial> fwiw, i can confirm that fate passes using gcc 4.9.0 compiled with that fix backported [00:00] --- Thu Apr 24 2014

1 0

ffmpeg.log.20140423
by burek 24 Apr '14

24 Apr '14

[02:13] <i336_> hi. I want to delay the start of a video (ie, video+audio stream) being run through ffplay by an arbitrary amount. how might I do this? [02:14] <llogan> sleep 3 && ffplay input [02:15] <i336_> oh, nice :P but that doesn't exactly work [02:15] <i336_> for whatever reason (probably old-ish hardware :P ...very probably :P) ffplay has a small tendency to jitter/stutter at the start of videos [02:15] <llogan> oh, you want to video to play offset from audio? [02:15] <i336_> once they start playing, they're fine [02:16] <i336_> so I want ffplay to play 5-10 seconds of blankness so it sort of "settles", hopefully playing the video perfectly once it starts [02:16] <i336_> I have no idea if it'd work, but I know ffmpeg has a few kitchen sinks in it so I thought I'd give this idea a try [02:16] <i336_> (note, the video and audio both stutter) [02:16] <i336_> (so I want to pause both) [02:17] <blippyp_> use a pad [02:17] <i336_> hm? [02:17] <blippyp_> man ffmpeg-filters [02:17] <i336_> oh, okay, thanks [02:17] <blippyp_> search for pad [02:17] <blippyp_> np [02:18] Action: i336_ gets a "No manual entry for ffmpeg-filters" and minces off in the general direction of google [02:18] <i336_> xD [02:18] <blippyp_> go to the ffmpeg site [02:18] <i336_> oh, ok [02:18] <blippyp_> they have the same info in their documentation section [02:18] <i336_> right [02:19] <i336_> um, padding seems to be primarily geometric, not chronological: https://www.ffmpeg.org/ffmpeg-filters.html#toc-pad [02:19] <blippyp_> you could also use a drawbox... [02:19] <i336_> unless I'm reading something wrong [02:20] <blippyp_> it's all in the filter [02:20] <blippyp_> something like [02:21] <blippyp_> give me a sec... [02:21] <i336_> mk :D [02:23] <blippyp_> I think you'd have to create the blank video first and then concatenate it [02:24] <i336_> o.o [02:24] <blippyp_> ffmpeg -loop 1 -i blank.png -t 5 -c:b libx264 -qp 0 -preset ultrafast blank.mkv [02:24] <blippyp_> something like that would make a 'blank' 5 second clip (assuming blank.png is a black picture) [02:24] <i336_> ah, I see [02:25] <blippyp_> then concatenate that with your video [02:25] <i336_> ha, right. :/ [02:25] <blippyp_> using the concat filter [02:25] <i336_> oh ok [02:25] <blippyp_> there might be an easier way though [02:25] <blippyp_> but that will work [02:26] <blippyp_> whether if solve your problem or not, you'll have to test to see... ;) [02:26] <blippyp_> if=it'll [02:27] Action: i336_ boggles at doing that to all the random videos he downloads from youtube [02:27] <i336_> hrm :s lol [02:27] <blippyp_> yeah, not really worth the re-compiling... [02:27] <blippyp_> re-encoding* [02:28] <blippyp_> try pausing the video when it starts for a second? [02:29] <i336_> hm [02:30] <i336_> that was amusing :D [02:31] <i336_> I got instance #6212 of "ALSA lib pcm.c:7339:(snd_pcm_recover) underrun occurred" when I paused the video, and when I unpaused it, the video ran 1.2 times faster than normal and finished before the audio stream :D [02:31] <i336_> lol [02:40] <llogan> blippyp_: no need for a blank png if you ever wanted "blank" video: ffmpeg -f lavfi -i color=s=640x480:color=blue ... [06:02] <blippyp_> i336: If you're still around, I just remembered - You can use a bitstream filter if your files are x264's - Convert the streams to .ts or m2ts and then concatenate them without re-encoding. I do this all the time actually - You might be able to do something similar with other codecs as well. Point being, you may not need to re-encode depending on the codec (and if you're getting them from you-tube, then they should qualify for this I think [06:14] <i336_> oh ok [06:15] <i336_> I get the files at, um, some kind of low quality MP4 setting, because anything wider than 720 pixels just shows up as black here [06:15] <i336_> (old graphics card) [06:15] Action: i336_ remembers it's format 18, but doesn't remember what that is [06:31] <blippyp_> i336: What are you using to download them? [06:31] <i336_> the awesomeness of youtube-dl :D [06:31] <blippyp_> you should be getting the best quality video from them by default then [06:32] <blippyp_> are you specifically specifying which format you want? [06:32] <blippyp_> I wouldn't do that... [06:32] <blippyp_> I just pull the direct link - no other settings, and I usually get the highest quailty mp4 from them [06:33] <blippyp_> sometimes they don't have mp4 versions though - not sure why, but whatever - by default they do send the highest quality video they have to you as far as I know [06:35] <blippyp_> with those mp4's you'll want to look at this page if you haven't found it yet: http://www.ffmpeg.org/ffmpeg-bitstream-filters.html [06:35] <blippyp_> specifically you'll want to use the h264_mp4toannexb bitstream [06:36] <blippyp_> when you're done your splicing, join them back together.... [09:43] <misterno> hey guys is there any other channel devoted to encoding/transcoding and videos [09:44] <misterno> because i'm looking to ask for advice if anyone has used popular encoders/transcoders beside ffmpeg [09:44] <misterno> which one has cuda support/good quality [09:46] <JEEB> misterno, "GPU encoding" in general sucks. As in, those that use the actual GPU hardware instead of having a specific hardware encoder on board. Of course, for such hardware encoders, the intel stuff is better than what any GPU maker has around, so... [09:46] <misterno> by "sucks" you mean quality or speed? [09:47] <JEEB> both in the end [09:47] <misterno> so there is noticable difference in quality when i turn gpu acceleration? hmm didnt know that [09:47] <misterno> *turn on [09:47] <JEEB> if you actually mean encoding, yes. And if the encoding is done on the GPU [09:47] <misterno> thats why im looking fro advice [09:47] <JEEB> there's the "do motion estimation on the GPU/whatever" kind of stuff too [09:48] <JEEB> but really [09:48] <misterno> well yes im looking for encoding/transcoding [09:48] <JEEB> nah, I meant as in the chain of things [09:48] <JEEB> reading->decoding->scaling/whatever->encoding->muxing->writing [09:48] <misterno> here is one example i just used for test and why im confused [09:48] <JEEB> anyways, you're much better off using money for a better intel CPU than grabbing anything GPU-based [09:49] <JEEB> unless you like expensive heaters [09:49] <misterno> i used wondershare video converter to transcode flv video i downloaded from web into mp4 and quality was horrible at approximately same filesize [09:49] <misterno> most settings were "as original" [09:49] <misterno> basically it looked like framerate dropped, not smooth [09:50] <misterno> then i downloaded handbrake, used default settings and also got approx same file size and conversion looked great and smooth [09:50] <misterno> this wondershare has gpu acceleration, but i tried turning it off with same result [09:50] <misterno> so in both cases codec used was h264 [09:50] <JEEB> it's not a codec, it's a video format [09:50] <misterno> why would first one have such horrible result [09:51] <JEEB> a codec is a coder and a decoder [09:51] <JEEB> as in, an implementation [09:51] <misterno> isnt mp4 video format [09:51] <JEEB> it's a multimedia container [09:51] <misterno> dont we call containers basically formats [09:51] <JEEB> container contains streams, streams contain audio/video/subtitles/whatever [09:51] <JEEB> yes, it's a container format basically [09:51] <misterno> well then we understand each other [09:52] <misterno> what i dont undersatnd is why would video look like framerate has dropped [09:52] <JEEB> but yeah, you most probably weren't using the same implementation, and even if you were it's possible to completely mutilate the result if you use an implementation badly [09:52] <JEEB> well, most probably that random video converter just did some random filtering? [09:52] <JEEB> because that's such a good idea, right? :P [09:53] <misterno> im looking for simple software that doesnt have many options [09:53] <JEEB> anyways, don't waste your time with random video converters [09:53] <JEEB> handbrake is probably the sanest thing around for people who have no clue [09:53] <misterno> the point im trying to make is i used default handbrake options [09:53] <JEEB> as far as a GUI goes [09:53] <misterno> so i should expect even from software that doesnt have many options to just do it well by default [09:53] <misterno> i need to find best quality transcoder/encoder with simple interface for my mom to use [09:54] <JEEB> well first of all they are completely different software? even though both use ffmpeg's libraries in the background most probably (since most random converters use ffmpeg while ignoring the license) [09:54] <JEEB> well, handbrake's pretty much the sanest you'll find [09:54] <misterno> thats why im looking for advice if someone has already tested all popular transcoders [09:54] <misterno> is there a channel besides this one that talks about transcoding? [09:55] <JEEB> well, I'll just say that you will want to keep to handbrake, it doesn't have all the options I'd like, but for random transcoding for a random user it's a good one [09:55] <JEEB> and it has an active development team [09:55] <misterno> handbrake will be too confusing for my mom [09:55] <JEEB> well, nothing else comes even close to the sanity and not having random issues :P [09:55] <misterno> if app supports cropping video that is a plus! [09:55] <misterno> :) [09:56] <misterno> i mean cutting [09:56] <JEEB> you could wrap around handbrake somehow if you wanted [09:56] <misterno> i just tried freemake video converter... slower than wondershare but double the filesize and quality was ok [09:56] <misterno> hmm [09:57] <JEEB> the only non-free sane'ish thing I know is tmpgenc mastering works (I think that's how it's called nowadays?), but really... there just is no magical bullet in these things. The things that try to be a single-button thing usually have random things they fail at, and otherwise you get tons of crappy GUIs [09:57] <JEEB> you could just make a drag and drop batch file for handbrake or something [09:57] <JEEB> so your mom could just drag and drop files [09:57] <misterno> cant [09:58] <JEEB> anyways, most of those random converters _will_ fail you at some point, and many of them are ffmpeg license offenders [09:58] <JEEB> so stay away from pretty much anything that's not open source [09:58] <misterno> this freemake is opensource [09:59] <JEEB> no, it is not [09:59] <misterno> oh there is another one called mediacoder [09:59] <JEEB> that one's awful too [09:59] <misterno> oh sorry its free not opensource [09:59] <misterno> mediacoder is awful? [10:00] <JEEB> yes [10:00] <misterno> again quality or? [10:00] <JEEB> as an application [10:00] <misterno> ah UI [10:00] <JEEB> no [10:00] <JEEB> last thing I remember of it and many others is that it's pretty much malware [10:01] <JEEB> anyways, you just don't want to go there. Wrap around handbrake, or if you don't want that just wrap around ffmpeg or mpv or something >_< [10:01] <JEEB> what you find on the internet is mostly just crap, and otherwise will just contaminate your system [10:01] <misterno> just a question, having multiple transcoders installed at the same time wont affect one another? [10:01] <JEEB> not worth it [10:01] <JEEB> sane applications don't install system-wide components. Unfortunately not many are sane of the random shit you find on the internet :P [10:01] <misterno> i have to do it [10:02] <misterno> since i cant find advice i will just have to install each and every and test myself [10:02] <JEEB> I think I've already given you advice [10:02] <misterno> i was hoping to save time xD [10:02] <JEEB> stay the fuck away [10:02] <misterno> cant [10:02] <JEEB> of everything not open source [10:02] <JEEB> seriously [10:02] <misterno> because i need to find something nice and simple for my mom [10:02] <misterno> i dont know why you cant understand [10:03] <JEEB> I can understand, I'm just telling you that look for open source [10:03] <JEEB> if you can't find the source code around, it's bad [10:03] <JEEB> there are exceptions, but not in your class [10:03] <JEEB> (of software) [10:04] <JEEB> also you could just write an automatization wrapper with batch files or 3gpp encoder GUI or whatever to enable drag and drop encoding for a specific format or so [10:04] <JEEB> if you don't find anything [10:08] <misterno> to reduce filsize and maintain quality i should basically just use h264? [10:08] <misterno> no xvid divx and such? [10:08] <JEEB> yes, it's the thing with best implementations right now [10:08] <JEEB> and most hardware thingamabobs support it [10:08] <misterno> alright [10:09] <misterno> and what format you suggest? mp4? [10:09] <misterno> i mean container [10:09] <JEEB> yeah [10:09] <JEEB> mp4 is supported out of box with win7+ and most thingamabobs have at least broken support for it [10:09] <misterno> and all apple devices use mp4 right [10:09] <JEEB> yes [10:09] <JEEB> it's based on their container after all [10:10] <JEEB> mov [10:10] <misterno> oh boy i always hated mov [10:10] <misterno> mov along with real [10:10] <misterno> xD [10:12] <JEEB> also man, it just makes me sad when I see a listing of "five best video converters" on some relatively popular site, and I know that at least two or three of them bundle the universe from like 2005 to 2009 related to DirectShow in themselves [10:12] <JEEB> and register it globally [10:13] <JEEB> mostly because I deal with the aftermath of such shit related to a certain separate project [10:13] <misterno> are you on linux or windows [10:13] <JEEB> both [10:13] <misterno> what is your windows media player of choice [10:14] <JEEB> windows at home for various reasons (linux in a VM), mobile laptops are linux [10:14] <JEEB> media player? [10:14] <misterno> yea [10:14] <JEEB> if you mean for simplicity and JustWorks, VLC [10:14] <misterno> justworks/ [10:14] <misterno> ? [10:14] <JEEB> I use something else, but that's because I can handle the troubleshooting [10:14] <misterno> i use vlc as backup but since it doesnt have nice UI i use different one [10:14] <misterno> gom player [10:15] <JEEB> yay, the GPL fucking korean player (infringes on the license of at least ffmpeg and mpc-hc as well as a few other projects most probably) [10:15] <misterno> gom player is just amazing, really nice seeking too [10:15] <JEEB> I maintain a DirectShow playback package and use it mainly of course, but I hate DShow :P [10:15] <misterno> all other players have bugs in seeking [10:15] <JEEB> (GOM is one of many DShow players) [10:15] <misterno> but gom is one of best [10:16] <misterno> and ofc mpc-hc is supposed to be good but i just prefer gom [10:16] <JEEB> dunno, sounds like you were just using bad dshow filters with everything else :P [10:16] <misterno> i also use shark007 codec pack [10:16] <JEEB> anyways, just noting that gom/kmplayer are hurr durr license violators from way back [10:17] <misterno> then i can choose to play via internal gom codecs or just disable and use sharks [10:17] <misterno> gom is license violator? [10:17] <misterno> in what way [10:17] <JEEB> uhhh, yes? [10:17] <JEEB> ffmpeg and mpc-hc code has been used at least [10:18] <JEEB> ffmpeg is more or less OK if you use it as LGPL, but guess if anyone does when they just grab and use the code :P [10:18] <JEEB> mpc-hc code is GPL so welp [10:20] <misterno> media codecs, encoders, transcoders, etc is such a huge mess... i mean ofc it's been like that since forever but this is never gonna sort out is it [10:20] <misterno> and when you dive deeper into encoding, omg so many different formats, parametars one has to have in mind, to not screw up quality [10:20] <JEEB> nah [10:21] <JEEB> that latter part you're incorrect in [10:21] <JEEB> libx264 is very simple to use [10:21] <misterno> trust me, for average person - this is incredible mess [10:21] <misterno> for someone in the know everything is always easy [10:21] <JEEB> you can pretty much make a sane GUI for x264 with two sliders (in addition to "what shit you want this thing to play on") [10:21] <JEEB> one being "quality level" (CRF) [10:22] <JEEB> and the other being "preset" (which goes from ultrafast to placebo) [10:22] <JEEB> that's it [10:22] <misterno> it's not about potentiality but about what is out there [10:22] <JEEB> literally, that's it [10:22] <JEEB> the real hard part is the "what shit you want this thing to play on", but you can mostly hide that [10:23] <JEEB> except when the user wants to have some stuff to tweak, which is when your simple design starts coming crumbling down [10:23] <misterno> well that's exactly my point [10:23] <misterno> tweaking is brought to absurd levels [10:23] <misterno> for average user [10:23] <misterno> you know how fridge just works? [10:23] <JEEB> the closest I've ever seen to a sane encoding GUI is handbrake, and trust me - I've seen plenty :P [10:23] <misterno> all you tweak is temperature and even that is pretty much standard for most foods [10:24] <misterno> and you just throw stuff into fridge and you are fine [10:24] <misterno> this is exactly the opposite [10:24] <misterno> you have to tweak everything to get good results [10:24] <JEEB> no [10:24] <misterno> too much to learn [10:24] <JEEB> you don't [10:24] <misterno> too much too know [10:24] <misterno> you have to, if for no other reason then just to test and discard crap that is out there [10:24] <misterno> one cant immediately know the correct or best way [10:24] <JEEB> as I just said, x264 has those two parameters and even if some apps try to show you something else you can generally just ignore them [10:25] <JEEB> and if the app doesn't have those two parameters you can just "close the book" for that thing [10:25] <misterno> my head just hurts :( [10:26] <JEEB> anyways, when I used to do shit for a specific thing I just made a preset for a GUI that ran random commands in the background [10:26] <misterno> what im looking now is perfect example [10:26] <JEEB> I drag and drop files, press encode and go take some coffee [10:26] <misterno> i need something where my mom will just drop the file, reduce its filesize without completely ruining the video [10:26] <misterno> it has to be like 2-3 clicks [10:26] <JEEB> you can't always reduce the file size btw [10:27] <JEEB> also you can't know what kind of quality level a person likes, and we don't yet have AIs simulating visual quality [10:27] <misterno> that's why one has to tweak to get good result [10:27] <JEEB> which is why you need that CRF slider [10:27] <JEEB> the user pokes it a couple of times, and then selects the highest that still looks good [10:27] <JEEB> you really can't automatize it other than giving a sane default like the libx264's own default (CRF 23) [10:27] <misterno> when i speak of quality i speak of perceivable same quality on computer monitor [10:27] <JEEB> yes [10:28] <misterno> i dont mean watching movies on big TV [10:28] <JEEB> which is something you can't just automatize, unfortunately [10:28] <JEEB> you just can't [10:28] <misterno> and i need reduced filesize for youtube uploads [10:28] <misterno> but to still look same [10:28] <misterno> it is possible because ive done it [10:28] <misterno> but it is just so messy to get there [10:29] <misterno> so much time to research :( [10:29] <JEEB> "I have been able to get some samples to compress while not losing too much of perceptional quality" [10:29] <JEEB> this is what you are saying [10:29] <misterno> not just samples but yea [10:29] <JEEB> anything you encode, a single thing is a single sample [10:29] <misterno> videos are usually recorded by phone or similar [10:29] <misterno> my mom dancing :D [10:29] <misterno> i dont know what they aer doing [10:29] <misterno> but anyways, quality to begin with is not that great [10:30] <misterno> i just want if video looks smooth to look smooth after conversion [10:30] <misterno> and not like i just now got the effect of halfed framerate [10:30] <misterno> that was so obvious [10:30] <JEEB> anyways, you will not be able to _always_ (with all samples) compress something down from its original size without perceived visual quality loss [10:30] <JEEB> so just throw that idea out of your head [10:31] <misterno> i understand, thats why im looking for best compromise [10:31] <misterno> but from what ive seen her phone videos can be compressed alot [10:31] <JEEB> that's quite possible [10:31] <JEEB> esp. with libx264 you can often good get results [10:31] <JEEB> (and with just those two parameters I noted) [10:32] <JEEB> you start losing in compression capabilities as soon as you start having to cater for hardware playthings [10:32] <JEEB> because generally that means limiting the feature set in one way or another [10:32] <JEEB> anyways, you could ask on #handbrake or something if they know of something even more simple [10:32] <misterno> when resolution is lets say 480 or lower one should always keep orig res? but when it is higher one could change it? [10:33] <misterno> one lady had this camera that can record in 1080p [10:33] <JEEB> but really, the real answer should be giving your mom handbrake and telling her to ignore most of it. [10:33] <misterno> so to reduce filesize and maintain great quality it is recommended to reduce res? [10:33] <JEEB> (unless there's of course a better alternative that's sane) [10:34] <JEEB> also if I had the time and money I'd just make something simple myself as open source (you just have to make the interface after all), but I just don't have the need myself [10:34] <misterno> what about videos shot at 50fps [10:34] <misterno> they look really smooth i have seen [10:34] <misterno> havent tried compressing these yet [10:35] <misterno> i admire you guys who know how to code [10:35] <misterno> everything seems doable to you guys :) [10:36] <JEEB> basically if I'd do something for myself I'd just automatize stuff but leave the GUI gimmicks out of it, which then wouldn't really be that useful [10:37] <JEEB> drag and drop batch files on windows, shell scripts or python scripts on *nix [10:37] <misterno> you would just make a command line hehe [10:38] <JEEB> http://mobilehackerz.jp/contents/3GPConv [10:38] <JEEB> this is what I used to use at one point for automation on windows [10:38] <JEEB> you drag and drop files onto a list, pick the preset for them to be encoded, select the output directory and press the button [10:38] <JEEB> on the background you had ini files that just told it what apps to launch [10:39] <JEEB> (the presets, basically) [10:39] <misterno> :) [10:40] <misterno> i just dont see this ever resolving [10:40] <t4nk921> Hello guys, Why is transcoding an mpegts segment with -copyts much more expensive than with out the -copyts parameter? I was tying to transcode segments on a cluster and wanted to keep the time stamps while transcoding each segment. [10:40] <misterno> it seems there is natural tendency for entropy to increase [10:40] <misterno> and make things more convoluted [10:40] <JEEB> that's not really true [10:40] <JEEB> as I said, during the years x264 and its library have become much simpler to use [10:41] <JEEB> still doesn't stop people from providing you the kitchen sink and all, even if it has no real use for end users :P [10:41] <misterno> thats what i mean, people looking for profits will make things so convoluted over time [10:42] <JEEB> many new encoders also more or less follow the suit, it's generally the GUIs and such that have the slower turnover [10:42] <JEEB> because users LOVE things to tweak, even if they most probably just make it all worse [10:42] <JEEB> and/or because the GUI was made when those options were all that was to it [10:42] <misterno> t4nk921 im sorry i dont know this stuff im only here on bicycle xD [10:43] <JEEB> anyways, as I said there's plenty of really simple stuff, you just have to make them do what you want :P [10:44] <JEEB> like that thing I linked [10:44] <JEEB> make an ini fle for your mom, put it into an archive where it's the only preset [10:44] <JEEB> have her use that [10:44] <JEEB> and boom [11:00] <t4nk921> @misterno do u know any one that can help me on this [11:00] <misterno> dunno sorry, maybe jeeb knows [11:01] <JEEB> I would have replied to him if I knew :P [11:58] <misterno> is cuh264 cuda supported h264? [12:02] <misterno> yes it is [12:30] <misterno> does anyone know to explain what is "log verbosity level" [12:46] <klaxa|work> misterno: the level of verbosity? [12:47] <klaxa|work> higher verbosity level means more information printed on stderr [12:47] <misterno> ok [13:01] <dagerik> help me make screencast which i can put inside public_html and have browser play it nicely right away. [13:01] <dagerik> i have seen the webm container does this. [13:02] <klaxa|work> https://trac.ffmpeg.org/wiki/How%20to%20grab%20the%20desktop%20%28screen%29… [13:02] <dagerik> klaxa|work: why does author use flv container? [13:02] <klaxa|work> instead of libx264 use libvpx with vp8 [13:02] <klaxa|work> i don't know [13:03] <klaxa|work> using webm instead should make no difference [13:03] <dagerik> should i just convert afterwards? [13:03] <klaxa|work> if your computer has enough computation power you can encode on the fly [13:03] <dagerik> i got intel i7 [13:05] <klaxa|work> depending on what you want to screencast, that's probably fast enough [13:05] <dagerik> ffmpeg -video_size 1920x1080 -framerate 25 -f x11grab -i :0.0 -f alsa -ac 2 -i pulse output.webm gave really bad quality [13:05] <dagerik> how do i increase quality [13:06] <klaxa|work> i think you can use crf with libvpx [13:07] <klaxa|work> try adding -crf 20 if the quality is still not good enough, decrease the value [13:08] <dagerik> CQ level must be between minimum and maximum quantizer value (4-63) [13:09] <dagerik> klaxa|work: even at 5 the quality is bad [13:10] <klaxa|work> maybe have a look at this: https://trac.ffmpeg.org/wiki/vpxEncodingGuide [13:10] <dagerik> yes thanks for googling for me. sorry :P [13:10] <klaxa|work> i rarely encode to vp8 [13:10] <klaxa|work> so i'm not sure about what settings to use [13:10] <dagerik> i want to deliver the video in HTTP via browser as easy as possible [13:10] <dagerik> no flash [13:10] <klaxa|work> i *think* there was a quality switch that had "best" "good" and "realtime" [13:11] <klaxa|work> yeah i really like that idea [13:11] <klaxa|work> if you want to do video in a browser, use webm [13:17] <dagerik> i tried using -crf 10 -b:v 1M -framerate 25 but the framerate is more around 5 [13:18] <dagerik> one cpu core goes up to 97% [13:23] <Turingi> I am trying to chop off the last part of an mp3 file and trying this: ffmpeg -t position 10:57 -acodec copy -i input.mp3 output.mp3 and I'm getting "unknown decoder copy" error [13:24] <Turingi> should chop off the trailing part of the mp3 after 10 minutes, 57 seconds [13:24] <Turingi> and without transcoding [13:28] <klaxa|work> put the -acodec copy after the -i input.mp3 [13:29] <klaxa|work> general syntax is: ffmpeg [-f <format>] [input options] -i <input> [output options] <output> [13:30] <Turingi> klaxa|work: ah, I see [13:31] <Turingi> klaxa|work: thanks, so in principle input and input decoders go before outputs and encoders [13:32] <klaxa|work> yes, input decoders have to be specified before the input, output encoders have to be specified between input and output [13:39] <blippyp_> t4nk921: Just a shot in the dark (I am literally talking out of my ass here) but I think your issues has to do with keyframes and (B Frames??) I think to be more precise... Google that and you might see what I mean... [13:57] <blippyp_> dagerik: What resolution is the video you're making? A bitrate of 1M is very low?? [13:58] <Mavrik> also [13:58] <Mavrik> setting CRF and bitrate is sill [13:58] <Mavrik> y [13:58] <Mavrik> so don't do that. [14:09] <dagerik> blippyp_: 1080p [14:09] <blippyp_> you should be using at least 9M I think [14:10] <blippyp_> here's youtubes 'suggested' - it might help you as a guide: https://support.google.com/youtube/answer/1722171?hl=en [14:11] <blippyp_> obviously for better quality you'll want to use higher bitrates... but that's useless if you're uploading to a service like youtube which will downgrade your video anyway... [14:12] <blippyp_> t4nk153: thought you were still here - you were t4nk921 I assume? [14:15] <t4nk153> this is a random name given by webchat :) [14:15] <blippyp_> oh - so you're someone else then? [14:16] <_odie> probably i'm not the same guy you were talking to earlier [14:16] <blippyp_> no - i never spoke to him - just offered an answer to his question [14:16] <blippyp_> but didn't realize he left [17:13] <xreal> Do I need to create x264 files with a resolution dividable by 16 ? [17:13] <RenatoCRON> 2 [17:14] <xreal> RenatoCRON: I thought, Macroblocks are 16 [17:15] <Plorkyeran> the video is padding to mod16 and cropped on decode [17:16] <Plorkyeran> this does not have any meaningful negative effects [17:16] <xreal> Plorkyeran: Thanks. I've read many tutorials, which for me to use 16 :( [17:16] <RenatoCRON> It's true that you might get some tiny bit of extra compression efficiency using a mod16 resolution, but the difference in x264 today is so small that it doesn't really matter in practice. The warning was removed for a reason. [17:16] <RenatoCRON> from: http://forum.doom9.org/archive/index.php/t-157072.html [17:17] <xreal> I already wondered because 1920 x 1080 isn't correct with 16 :) [17:18] <RenatoCRON> LoRd_MuldeR explains what happens in this case [17:19] <Plorkyeran> the warning never really should have existed in the first place with x264 [17:19] <Plorkyeran> it mattered more with xvid [17:20] <Plorkyeran> especially low-resolution xvid, since the padding was larger relative to the size of the image [17:29] <xreal> oh, why does it say "Audio: Vorbis (deprecated)" ? [17:29] <xreal> I'm using c:a libvorbis [17:40] <xreal> Damn, why can't I find any information to keep chapters :( [17:41] <xreal> Do I need mplayer for this? [17:41] <sacarasc> xreal: What are you going from and to? [17:41] <xreal> sacarasc: a demo bluray (ISO file) to mp4 and mkv [17:42] <sacarasc> Didn't know MP4 could do chapters... [17:42] <xreal> sacarasc: I tried handbrake, it works... [17:42] <xreal> sacarasc: But I want ffmpeg. [17:43] <xreal> sacarasc: so, I need just mkv as target and it'll work ? [17:43] <sacarasc> Apparently, it's -map_metadata c:chapter_index. [17:44] <xreal> ah ok [17:44] <sacarasc> Wait, no. [17:44] <sacarasc> That's per chapter metadata. [17:44] <sacarasc> -map_chapters I guess. :p [17:44] <sacarasc> http://ffmpeg.org/ffmpeg-all.html [17:45] <xreal> sacarasc: Expected number for map_chapters [17:45] <sacarasc> So read the docs. [17:48] <xreal> sacarasc: I just wonder, how to get this index file. [17:49] <sacarasc> AFAIK, ffmpeg can't take disk input, so maybe when you rip the disk you get it? [17:50] <xreal> sacarasc: [x] done already. Have direct access to the file. [17:51] <xreal> it's not protected, it's a CC-BY-SA demo DVD. [17:55] <xreal> oh, why does it say "Audio: Vorbis (deprecated)" ? I'm using -c:a libvorbis [18:11] <xreal> "Only SUBTITLE_ASS type supported" ... I can't understand this :( [18:12] <xreal> Is this a problem of source or target? [18:13] <sacarasc> Depends where it says it. [18:13] <xreal> [ass @ 0e0c9a20] Only SUBTITLE_ASS type supported. Subtitle encoding failed [18:14] <xreal> I've chosen "-scodec ass" [18:14] <xreal> ok, takes some time [18:15] <lmfao> how to use ffmpeg with kaffaine? [18:19] <xreal> sacarasc: http://sebsauvage.net/paste/?5212ff953eb65506#8YIQ6dsDGWrq5PqojpSbTobS3iqA8… [18:20] <sacarasc> Maybe ASS output only works for ASS input? [18:21] <sacarasc> IIRC, though, PGS subs are images, so you'd have to do some OCRing on them. [18:22] <xreal> sacarasc: I should get a better demo bluray :( [18:22] <xreal> I tried "Sintel" [18:22] <sacarasc> PGS is about all you'll get on a BluRay. [18:23] <xreal> sacarasc: oh, okay. [18:23] <xreal> http://exar.ch/suprip/ :) [18:23] <xreal> SupRip lets you OCR and detect text within subtitles as used by HD-DVD and Bluray movies. You can then save the text as a SRT file to be included in reencoded movies. [18:25] <sacarasc> You can put PGS in Matroska anyway. [18:25] <sacarasc> When you OCR stuff, you have to read through it all to make sure there's no errors. [18:25] Action: sacarasc is way too lazy for that. [18:29] <xreal> sacarasc: -c:s copy ? [18:29] <sacarasc> I think so. [18:29] <xreal> sacarasc: Can I select subtitles? [18:29] <xreal> sacarasc: copy first subtitle only [18:30] <sacarasc> You can use -map to pick which streams you want. [18:30] <xreal> sacarasc: ok [18:30] <xreal> sacarasc: -c:s:0 copy might also work! [18:39] <xreal> sacarasc: works, but subtitles are auto activated now :( [19:26] <dannyzb> my ffmpeg is defaulting x264 "profile high 5.1" .. i thought it goes upto 41 -_- [19:39] <sacarasc> dannyzb: Did you mean 4.1? [19:39] <dannyzb> nope .. 5.1 [19:39] <dannyzb> it says 5.1 .. thats the wierdest default ever [19:40] <sacarasc> No, instead of 41. [19:40] <dannyzb> sacrasc : what? [19:41] <sacarasc> "i thought it goes up to 41" [19:43] <dannyzb> ah yea [20:03] <llogan> xreal: does your ffmpeg build have support for libvorbis? check configure line for --enable-libvorbis [20:03] <llogan> if not then it will use the crappy, native "vorbis" toy. [20:03] <llogan> i guess [20:04] <xreal> llogan: 1 sec [20:04] <xreal> llogan: --enable-libvorbis [20:05] <xreal> llogan: I'll check this later again. [20:06] <xreal> llogan: Is there a way to enable/disable subtitles in MKV by default or is it a setting on the player's side? [20:08] <llogan> i dont know [20:09] <llogan> i believe i've seen some users complain about ffmpeg adding "(default)" to languages or subtitles, but I don't remember much or have an answer [20:23] <dannyzb> when i try to use tee to output HLS ([bsfs/v=h264_mp4toannexb,dump_extra]/var/www/media/video/hd_hls/72760ed097ebe68dfaaf23377be167d2cff457c8/72760ed097ebe68dfaaf23377be167d2cff457c8.m3u8) I get the error: [20:23] <dannyzb> Packet header is not contained in global extradata, corrupted stream or invalid MP4/AVCC bitstream [20:32] <dannyzb> pretty sure it's a bug .. where do i report? [20:38] <dannyzb> llogan:http://pastie.org/9105750 [20:38] <llogan> where is the rest? [20:41] <dannyzb> the rest? this is the command [20:41] <dannyzb> llogan: thats the whole command [20:52] <xreal> Can mkvmerge also be discussed here? [20:53] <JEEB> the #matroska channel is generally used for that [20:56] <xreal> oh ok, thanks [21:08] <xreal> JEEB: Do you know how to copy chapters? or do I need mkverge for that? [21:09] <dannyzb> xreal: -map_chapters input_file_index (output) [22:19] <xreal> dannyzb: but how do I fill the input_file_index? What's that exactly? [22:21] <blippyp_> xreal: https://trac.ffmpeg.org/wiki/How%20to%20use%20-map%20option [22:21] <blippyp_> is that what you're talking about? [22:23] <xreal> no: I want to copy chapters. [22:25] <blippyp_> yeah - the chapters should be in the same format [22:26] <blippyp_> run medianfo (or simply run ffmpeg -i input.mkv [22:26] <blippyp_> it will show you the indexes [22:26] <xreal> blippyp_: or mkvmerge.exe ? [22:26] <blippyp_> no idea, never used it [22:27] <sacarasc> Use mmg.exe instead. :p [22:28] <sor_> is there a way to have ffplay constantly update or refresh video -- i am running durring capture and it stops after about a min - my guess is it thinks itls the end of the file [22:29] <xreal> sacarasc: yeah, I was use this, but I like ffmpeg [22:29] <xreal> sacarasc: Also, I want to transcode. [22:29] <sacarasc> You also didn't want to use handbreak, which is pretty much just a front end for ffmpeg. [22:30] <xreal> sacarasc: I like to fight until I get good results without external tools. It's a hobby. [22:30] <xreal> :) [22:30] <sacarasc> Then maybe you should go reading ALL the docs, and the source? [22:30] <xreal> sacarasc: I did, but the docs aren't that good about chapters. [22:31] <sacarasc> Then the source. [22:31] <xreal> I wonder, why there isn't -c:c copy [22:32] <blippyp_> my guess would be because it's metadata, not a stream [22:32] <xreal> sacarasc: by the way: handbrake often doesn't get the correct values for crop. I wrote my own tool for that using a numeric histogram. [22:32] <xreal> blippyp_: but you can even copy metadata... [22:32] <blippyp_> yeah I know [22:33] <blippyp_> but not with a -copy??? or am I wrong about that? [22:33] <blippyp_> doesn't it just automatically copy the metadata, and if you want to remove it you have to remove each one separately? [22:34] <xreal> blippyp_: then the chapters aren't metadata of the video. perhaps there's something like a "data" stream? [22:34] <blippyp_> why are you have such a hard time with the chapter information? I thought those transferred over just like the metadata? [22:34] <xreal> no :( [22:35] <xreal> the result doesn't have chapters. [22:35] <blippyp_> I ripped a bunch of movies last year - I didn't have to do anything, I'm 99% positive it copied all that stuff [22:35] <blippyp_> I remember because I was annoyed about not wanting to go thru and remove them, and instead just left it [22:35] <xreal> blippyp_: ew, ffmpeg -i <file> doesn't show the chapters. [22:36] <blippyp_> hold on - let me see if I can find one [22:36] <blippyp_> yea it does [22:37] <blippyp_> I think I ripped my videos with cdrtools - but I can't remember at all tbh... [22:37] <blippyp_> and my videos have all the chapter information in them [22:37] <blippyp_> and they're displayed whenver I run ffmpeg -i on them [22:37] <xreal> I am using ffmpeg version N-62569-gf57ac37 directly on a m2ts file. Is this the problem? [22:38] <blippyp_> not sure - I don't remember how I converted my videos in the end - sorry I wish I did, obviously whatever I did is exactly what you're looking for [22:39] <blippyp_> I think I just concatenated all the VOB files and then encoded them into mkv's [22:39] <blippyp_> but not with ffmpeg [22:40] <blippyp_> like cp file1.vob+file2.vob+file3.vob movie.mpg or something like that [22:40] <xreal> ah ok [22:41] <xreal> -f ffmetadata exports metadata [22:42] <xreal> it's empty here. very interesting. [22:42] <blippyp_> sorry if I'm misguiding you - like I said, it was well over a year ago (maybe even two or three) - I don't remember what I did exactly - But I still have the finished mkv's - and like I said I just checked them and they have the chapter information in them - And I definately didn't go out of my way to get them. [22:44] <blippyp_> the hardest part I remember was making sure that I only exacted exactly the video/sound/subtitle information I wanted out of the dvd with none of that extra crap... [22:44] <blippyp_> but that was an issue with cdrtools or whatever I used for that... [22:47] <xreal> I tried 12 versions of ffmpeg - doesn't work. [22:47] <blippyp_> trust me it does - at least from a year ago forward... [22:49] <blippyp_> http://sprunge.us/aEdA [22:50] <xreal> http://pastie.org/9106078 [22:50] <blippyp_> brb [22:50] <xreal> k [22:51] <xreal> JEEB: could you give me a hint [23:00] <blippyp_> xreal: you clearly don't have the chapters in the m2ts files - you lost them somewhere along the way... [23:01] <blippyp_> is this from a blu-ray disc? [23:01] <xreal> blippyp_: nope. This shows them: mkvmerge.exe --output-charset UTF-8 --identify-for-mmg I:\BDMV\STREAM\00000.m2ts" [23:01] <xreal> blippyp_: Yeah, Blu-Ray. [23:01] <blippyp_> maybe that's the difference [23:01] <xreal> blippyp_: Shall I move the file somewhere and test it again? [23:01] <blippyp_> maybe they're stored differently now - I've never used a blu-ray reader/burner [23:01] <JEEB> the chapters are not in the m2ts file [23:01] <xreal> blippyp_: It's an emulated ISO. [23:02] <JEEB> mkvmerge is just checking for and uses the whole directory structure [23:02] <blippyp_> ffmpeg may not have support for it [23:02] <xreal> blippyp_: ffplay supports ISOs. [23:02] <JEEB> the mpls (IIRC) files contain the chapters (playlists) [23:02] <xreal> JEEB: okay, 1 sec [23:02] <JEEB> but no idea if you can easily grab the chapters from those :P [23:02] <JEEB> with just ffmpeg [23:02] <blippyp_> it's not a dvd - it's a blu-ray completely different formats... [23:03] <blippyp_> you need to find someone who can actually confirm that this works with blu-ray [23:04] <xreal> Nope, mpls can't be read. [23:04] <JEEB> IIRC you might be able to build ffmpeg with libbluray [23:04] <JEEB> but not sure if you want to go through all that just to do this .-. [23:04] <xreal> So, I need to create an uncrompressed MKV via makemkv and put this into ffmpeg ? [23:05] <blippyp_> yeah - I'd try what jeeb's suggesting - rebuild with libbluray [23:05] <xreal> blippyp_: I'm on Windows using the official binaries. [23:05] <xreal> I'm not able to compile them on my own. Pretty much knowledge is needed. [23:06] <blippyp_> use a linux livecd [23:06] <xreal> blippyp_: even there it's pretty hard :) [23:06] <blippyp_> cross build your ffmpeg in there for windows [23:07] <blippyp_> not really... a little effort and you'll get it [23:07] <xreal> blippyp_: I've read some tutorials about it, but hell... that's complicated. [23:07] <xreal> I wish they 'd provide a simple shellfile which does everything. [23:07] <xreal> wget, configure, make, make install ... [23:08] <blippyp_> trust me, I understand your woes - it took me forever to get off my ass to re-compile ffmpeg with frie0r enabled - finally got it working and I was less than impressed... :( [23:08] <blippyp_> but I got it - it's not that hard, there are plenty of tutorials online for it [23:08] <xreal> I think, this is best right now: Blu-Ray => mkvmerge => mkv => ffmpeg => transcoded mkv [23:08] <blippyp_> I don't know what you need to do exactly to cross build it for windows, but there's tutorials for that as well [23:09] <xreal> blippyp_: there must be a reason, libbluray isn't integrated. [23:09] <blippyp_> not standard I guess - many people stil aren't using them... [23:09] <sacarasc> Integrated in what? [23:10] <xreal> sacarasc: in zeranoe's builds [23:10] <blippyp_> it's the same if you install something like debian - you have to install every other library that you want yourself... [23:10] <sacarasc> Is it in any ffmpeg? [23:10] <xreal> sacarasc: there's a patch: http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2012-February/120963.html [23:10] <blippyp_> I use arch, so I get a 'standarized' pre-compiled version of ffmpeg with much of what people tend to want already enabled in it [23:10] <sacarasc> I don't think you can even read DVDs with ffmpeg, so blurays are a bit out. [23:10] <xreal> sacarasc: okay, then I'll use the more complicated way :) [23:11] <xreal> DVD/Blu-Ray => mkvmerge => mkv => ffmpeg => transcoded mkv [23:11] <llogan> blippyp_: additionally you're using a version that isn't a million years old [23:11] <blippyp_> no, we're not talking about reading the dvd's or blurays just the chapter information... [23:11] <blippyp_> yes, mine is completely up to date and I even compiled it myself - I'm not even using the one from the arch repository at this point [23:12] <llogan> ABS or AUR? [23:12] <blippyp_> neither [23:12] <blippyp_> git [23:12] <llogan> ffmpeg-git is in AUR [23:13] <blippyp_> well, I cloned the git and built it myself [23:13] <blippyp_> the one from the aur is also designed to have certain things enabled in it that aren't standard (like libass or whatever) - At least I think that's right... [23:14] <llogan> so you didn't use a PKGBUILD? [23:14] <blippyp_> x264, x265, vorbis, etc.... [23:14] <blippyp_> no [23:14] <blippyp_> why? [23:15] <llogan> just curious. i like using PKGBUILDs [23:16] <llogan> not that i use them everytime. [23:16] <blippyp_> yeah me too - I use the aur all the time - i want to learn how to make my own and start contributing... [23:16] <blippyp_> no - neither do I - I actually try to avoid them [23:16] <blippyp_> but the aur is what separates arch from nearly every other distro out there [23:17] <blippyp_> it's beautiful - it's a mess also, but it's beautiful - I have personally never had a problem with it except that the package doesn't necessarily provide what I specifically wanted (like ffmpeg with frei0r enabled) [23:17] <blippyp_> I'm gonna re-compile it again with the opencl or whatever enabled as well - I think I'll enjoy having that more [23:19] <blippyp_> anyone here make any cool complex filters that they use? I'd love to hear about them... [23:21] <blippyp_> for example: I've been working on a 'cartoon' filter - it' works pretty good so far, better than any cartoon filter I've ever seen - I'm still not completely happy with it yet, but it works pretty well. [23:23] <llogan> blippyp_: here's a fun one https://trac.ffmpeg.org/wiki/EncodeforYouTube#Usingfilters [23:24] <llogan> oh, make a filter...not a filtergraph? [23:24] <blippyp_> yeah I was talking about a filtergraph [23:25] <blippyp_> give me a few minutes and I'll post a link to an example of mine [23:26] <blippyp_> but stuff like you just posted if fine though - that's the kind of thing I'm talking about - doesn't necessarily need to be a 'filter' - maybe a layout that you like or whatever... [23:26] <llogan> http://superuser.com/questions/736388/crop-triangular-area-with-ffmpeg/7365… [23:27] <blippyp_> which, btw - if you guys haven't checked out alphamerge yet - get on that - that's some cool stuff [23:27] <llogan> that link i just spammed uses it [23:28] <blippyp_> yeah - that's the kind of stuff I'm talking about... ;) [23:30] <blippyp_> damn, my system went wonky on me - I gotta reboot - brb [23:32] <xreal> My DVD recorder records 16:9 streams in 4:3 to get higher quality. Do I need scale or pad to correct this? [23:34] <llogan> how does recording 16:9 as 4:3 result in higher quality? [23:35] <xreal> llogan: that's what Panasonics manual says:) [23:36] <xreal> -aspect 16:9 doesn't exist anymore, does it? [23:36] <sacarasc> llogan: Maybe because 16:9 only works for 720x576 in PAL DVDs? :D [23:36] <llogan> i believe it does [23:37] <llogan> you can even use it with -vcodec copy, IIRC [23:37] <xreal> llogan: oh, I wanted to try ffplay to see if the result is correct [23:41] <xreal> damn [23:43] <sor_> ok can someone explain why setting -refs 1 allows my video to contiue playing when before it would stop [23:44] <xreal> found it! -vf "setdar=16:9" [23:44] <blippyp> I hate setsar/setdar - they give me headaches [23:45] <xreal> blippyp: it gives the correct result :) [23:45] <blippyp> that's because the src is 'wonky' [23:46] <blippyp> I dont' fully understand it myself - all I know is that I'm constantly having to setsar=1:1 in my filtergraphs - it drives me nuts [23:47] <blippyp> my camera will record with a wacky setdar/set sar setting as well - when I want to record in 16:9 it still displays at a 4:3 until I change the setdar i think (might be setsar) I have no idea, but it stores the video all whacky [23:48] <xreal> blippyp: but what's another way to compress the images on the y-axes ? [23:48] <xreal> scale doesn't seem to work. [23:49] <blippyp> no, it's my own stupidity - I literally don't understand it - I've gone over it a few times in the man pages and it kind of makes sense to me, but I just don't get it at all... [23:50] <blippyp> like setsar is literally how the video is stored I believe [23:50] <blippyp> and setsar is the display aspect ratio - by manipulating the actual data in setsar it can properly display the video [23:51] <blippyp> I think they made it to allow for automatic switching between 4:3 and 16:9 kind of scenario - but I don't really get it [23:52] <xreal> so is it okay to use setdar ? [23:53] <blippyp> yeah, that's what they're there for - normally I don't think you would mess with setdar - unless you were dealing with a situation like my camera [23:53] <blippyp> I don't know what you want it for [23:54] <xreal> 16:9 TV got recorded into 4:3 [23:54] <xreal> Now everything is scaled on the y-axes. [23:54] <xreal> I need to compress it back to make the people's faces smaller :) [23:55] <Bray90820> Would there be anyway to clip a video in the linux terminal without loosing any audio or video quality [23:55] <sacarasc> Bray90820: What do you mean by clip? [23:55] <Bray90820> I want to remove like a second or 2 from the beginning [23:56] <sacarasc> No, you would have to reencode (and thus lose video quality) to do that. [23:56] <sacarasc> Probably. [23:56] <blippyp> well, like I said - i dont' fully understand it - but the only reason why switching the setdar would help is if the original 16:9 data was still stored in the file... ??? otherwise you want to re-format the video and would instead want to switch the setsar to go back to 16:9 - but you'd lose quality -if that makes sense??? assuming I'm even right [23:56] <sacarasc> If your GOP is like 1 or 2, then you might able to. [23:57] <blippyp> Bray90820: depends on the keyframes - you can still mux it - but you might not splice it exactly where you want it - might also depend on the original codec [23:57] <blippyp> GOP - that's it [23:57] <xreal> blippyp: "setsar=4:3" and "setdar=16:9" are working. [23:58] <blippyp> it's a shitty example - but here ya go if you wanted to see it: http://a.pomf.se/jfngrx.mkv [23:58] <Bray90820> What my end goal is to remove 1 second from this video then convert it to audio with this small script [23:58] <Bray90820> ~/Desktop/VIDEO_TS$ ffmpeg -i '/home/aaron/Desktop/VIDEO_TS/file.VOB' -metadata TITLE=file file.flac [23:58] <blippyp> yeah, that's what I would have set - but you DID lose quality - you are 'stretching' the video without stretching it - again, assuming i understand it right [23:59] <Bray90820> so i don't really need to preserve the video quality just the audio [23:59] <xreal> blippyp: why am I not stretching it? I can see the streched result :) [23:59] <blippyp> Bray90820: it's about timing... you won't know if you can mux it unless you try [00:00] --- Thu Apr 24 2014

1 0

ffmpeg-devel.log.20140422
by burek 23 Apr '14

23 Apr '14

[00:59] <cone-977> ffmpeg.git 03Timothy Gu 07release/1.1:6294d84dca46: configure: use pkg-config to detect libbluray [00:59] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:e770488b5044: swscale/x86/swscale: fix missing xmm clobbers in yuv2yuvX_sse3() [00:59] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:af1bf7f2772a: avcodec/h263dec: Fix use of uninitialized memory from the bitstream buffer [00:59] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:f741c39752a0: avcodec/x86/mpegvideoenc_template: fix integer overflow [00:59] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:8130449f32bb: avcodec/h264_mp4toannexb_bsf: prepend global headers before any in stream parameter sets [00:59] <cone-977> ffmpeg.git 03Lukasz Marek 07release/1.1:88ec8021ce47: lavu/opt: validate range before dereference [00:59] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:93fe53da91f2: avcodec/wma: use av_freep(), do not leave stale pointers in memory [00:59] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:0143eb91098f: swresample/resample: Limit filter length [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:2880de35b53d: swresample/dither: use av_malloc_array() [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:eea1c5daa050: swresample/resample: use av_malloc_array() where appropriate [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:8f596a45998e: swscale/x86/swscale_template: loose hardcoded dstw_offset [01:00] <cone-977> ffmpeg.git 03Anthoine Bourgeois 07release/1.1:b546b7a029a6: avcodec/dirac_arith: Fix build with PIC and stack-check options [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:15f122316eef: avcodec/x86/idct_sse2_xvid: fix non C99 inline function [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:e459c80fdd13: swscale/swscale: fix srcStride/srcSlice typo [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:f30bd7c21569: avformat/mpegts: Remove redundant check [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:d51e0dd9b4d4: avcodec/diracdec: fix undefined behavior with shifts [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:47c23f93925d: avcodec/g723_1: add assert to help static code analyzers [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:353ff5b4ebf4: avfilter/vf_deshake: fix loss of precission with odd resolutions [01:00] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:b24e68c9c313: avfilter/filtfmts: Support dynamically allocated in/outputs [01:11] <cone-977> ffmpeg.git 03Michael Niedermayer 07release/1.1:5f390ee944f5: Update for FFmpeg 1.1.10 [01:31] <cone-977> ffmpeg.git 03Michael Niedermayer 07fatal: ambiguous argument 'refs/tags/n1.1.10': unknown revision or path not in the working tree. [01:31] <cone-977> Use '--' to separate paths from revisions [01:31] <cone-977> refs/tags/n1.1.10:HEAD: Merge commit '1c7b71a5bdb88ebb69734100405bbb5441b871e8' [01:32] <smarter> > hevc in flv [01:32] <wm4> lol [01:33] <wm4> multimedia... [02:21] <JEEB> them chinese [02:21] <JEEB> and their hacked up shit [02:21] <JEEB> there's like X versions of it, too [02:21] <JEEB> so nope, it should not be supported :P [02:21] <JEEB> (not to mention it's not official) [02:23] <Plorkyeran> if supporting it means they stop using rmvb... [02:24] <JEEB> they won't change that [02:24] <JEEB> also jesus christ [02:25] <JEEB> that's a german app that made that file [02:25] <JEEB> it must burn in hell! [02:25] <wm4> better than all the broken ffmpeg produced mkvs I stumble over... [02:26] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:47ec0b0b0f21: avutil/internal: add FF_ALLOC_ARRAY_OR_GOTO & FF_ALLOCZ_ARRAY_OR_GOTO [02:26] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:0dadbbbfd165: avcodec/snow: use FF_ALLOC(Z)_ARRAY_OR_GOTO [02:48] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:78fb5f6e2aaa: avcodec/cngenc: Use av_malloc_array() [02:48] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:dcad6ba68abb: avcodec/libxavs: Use av_mallocz_array() [02:48] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:658f5d1f63c9: avfilter/pthread: Use av_mallocz_array() [02:48] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:da621ffbcf19: avcodec/wma: use av_malloc_array() [03:28] <cone-977> ffmpeg.git 03Derek Buitenhuis 07master:84d6ba60db95: MAINTAINERS: Add myself as FIC maintainer [04:39] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:57fb570908df: avformat/avidec: Speed up keyframe detection code [06:35] <jamrial> I sure hope gcc fixes the miscompilation problem with flac/shorten/loco before releasing gcc 4.9.0... [06:35] <jamrial> Assuming of course it's on their end [07:24] <Daemon404> did you file a bug [07:34] <jamrial> cehoyos and the user that reported the issue did [07:36] <jamrial> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60902 [07:46] <nevcairiel> how unsurprising, it miscompiles the golomb code, i wouldnt be surprised if there was something "bad" in the code :p [13:35] <cone-447> ffmpeg.git 03Anton Khirnov 07master:a4ed995cabf2: txd: do not set the codec timebase. [13:35] <cone-447> ffmpeg.git 03Michael Niedermayer 07master:57123dc67e77: Merge commit 'a4ed995cabf220029f1d0e185a6fb45eed7b4091' [13:43] <cone-447> ffmpeg.git 03Anton Khirnov 07master:b19a5e51981b: lavc: improve AVCodecContext.delay doxy [13:43] <cone-447> ffmpeg.git 03Michael Niedermayer 07master:5e6be52bacb9: Merge commit 'b19a5e51981be5b69cf550a3bc17fe5300d0dbc9' [13:56] <cone-447> ffmpeg.git 03Anton Khirnov 07master:1ae8198bca74: avconv: always reset packet pts after decoding an audio frame [13:56] <cone-447> ffmpeg.git 03Michael Niedermayer 07master:3bf189bb0086: Merge commit '1ae8198bca749a0cff205196cc83d35b9962849b' [14:01] <cone-447> ffmpeg.git 03Anton Khirnov 07master:f9157463dbcd: lavf: do not use the parser duration for video [14:01] <cone-447> ffmpeg.git 03Michael Niedermayer 07master:8064c6ac5fe9: Merge commit 'f9157463dbcd2db8fe9504197c0c04d0d7d04f31' [14:07] <cone-447> ffmpeg.git 03Hendrik Leppkes 07master:5809c2a99d96: vc1dsp: fix build without inline asm [15:21] <dannyzb> when i have one input and 2 outputs , does ffmpeg use separate threads for each output and read the input once ? [15:29] <michaelni> ubitux, theres a subtitle patch (anshul (2.4K) [FFmpeg-devel] checked region version in dvbsubdec.c) [15:30] <michaelni> do you have time to review it ? [15:31] <ubitux> i never actually worked on dvb subs except as a user [15:31] <michaelni> dannyzb, if the encoders use frame threads that should be true [15:37] <dannyzb> michaelni: what does that mean ? i use h264 [15:44] <michaelni> ubitux, ok, reviewed it myself, though iam no dvb subs expert either [16:37] <cone-447> ffmpeg.git 03Graham Booker 07master:60fcc19b9068: avformat/mpegtsenc: Changed Video PES packet length to 0. [16:37] <cone-447> ffmpeg.git 03Michael Niedermayer 07master:f1c48c3a4de3: avformat/mpegtsenc: make the pes packet length omission optional [16:44] <ubitux> ah, 4.9 is out [16:47] <ubitux> (colors \o/) [16:48] <nevcairiel> too bad the flac golomb issue wasn't fixed in time [16:49] <nevcairiel> so, two or three months before it becomes usable [16:50] <JEEB> lately it's always been at least until the first or so dot release until both GCC and the projects get ready for a new GCC :) [16:50] <nevcairiel> 4.8 worked surprisingly well at release [16:50] <JEEB> was it 4.8 that derped with gradfun? [16:50] <JEEB> or was that 4.7? [16:51] <nevcairiel> 4.8 broke a few apps that abused C in bad ways, but who can blame a compiler for breaking code thats not standard compliant [16:51] <JEEB> yup [16:51] <JEEB> it's generally been like that [16:51] <JEEB> and then there's been the few cases of actually GCC bug being found [19:25] Action: Daemon404 points kierank at the ML [19:26] <kierank> for what [19:26] <Daemon404> libx264 change [19:26] <Daemon404> cant reember if you care about that or not [19:26] <kierank> I suggested doing that way [19:26] <kierank> the last implementation had hardcoded tables [19:27] <Daemon404> oic [20:10] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:b20d6cf60332: avcodec/dsddec: use av_malloc_array() [20:10] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:ea8e089ff7cd: avcodec/cook: use av_malloc_array() [20:10] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:994a582a02ac: avcodec/dvbsubdec: use av_mallocz_array() [20:45] <cone-647> ffmpeg.git 03Reimar Döffinger 07master:8cbf0827e1b2: Various small spelling fixes. [20:45] <cone-647> ffmpeg.git 03Reimar Döffinger 07master:3bd7a7149fbc: Add missing ';'. [20:45] <cone-647> ffmpeg.git 03Reimar Döffinger 07master:ab14bcb10420: Add "const" to avoid compiler warning. [21:39] <llogan> Daemon404: what would the message say? regarding the libx264 refs patch [21:39] <Daemon404> i dunno, just an info message sayng wht its been changed to? [21:40] <Daemon404> i can imagine being wtf'd over refs suddenly changing [21:40] <llogan> i was more wtf when refs were not changed [21:41] <Daemon404> well it's still a silent change in behavior [21:43] <llogan> i doubt most users even knew that refs was not changing previously and just assumed it would be similar to whatever x264cli does [21:43] <llogan> a guess a note wouldn't hurt, but i don't know what it would say [21:43] <JEEB> yeah, I always wondered why libx264 didn't do the same as what x264cli did regarding refs [21:44] <llogan> i'm no throbbing brain, but it even took me a while to notice. [21:45] <Daemon404> :P [21:56] <J_Darnley> If I remember correctly it was going to be added to libx264 at the same time as the mythical "device patch" [22:22] <cone-647> ffmpeg.git 03Derek Buitenhuis 07master:6d149d28183a: fic: Use proper quantization matrix index [22:39] <cone-647> ffmpeg.git 03Derek Buitenhuis 07master:dac7e8a94e49: avutil: Add av_get_time_base_q() [22:39] <Daemon404> michaelni, dont merge 9155c59 [22:39] <Daemon404> i accidentally pushed that without changes [22:40] <Daemon404> i will push the ffmpeg version myself in a bit [22:53] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:c7251fec39c5: avutil: move av_get_time_base_q() after include rational.h [22:59] <ruggles> just in case nobody has noticed yet, HEAD seems broken [22:59] <ruggles> http://fate.ffmpeg.org/?dsort=date [23:02] <Daemon404> yes [23:02] <Daemon404> patch is on ml [23:02] <Daemon404> oh wait wrong channel [23:02] <Daemon404> [21:53] < cone-647> ffmpeg.git Michael Niedermayer master:c7251fec39c5: avutil: move av_get_time_base_q() after include rational.h [23:02] <Daemon404> ^ [23:03] <ruggles> ah, ok. will pull and rebuild. thanks. [23:12] <jamrial> Just great, gcc 4.9.0 was released without the fix fox flac miscompilation [23:13] <Daemon404> yep. [23:27] <cone-647> ffmpeg.git 03Janne Grunau 07master:650c4300d94a: aarch64: NEON float FFT [23:27] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:13f4428915d6: Merge commit '650c4300d94aa9398ff1dd4f454bf39eaa285f62' [23:31] <ubitux> Justin cares about FFmpeg? [23:31] <Daemon404> he works with me [23:31] <llogan> vimeo? [23:31] <Daemon404> aye [23:32] <ubitux> i was wondering what was going on, okay [23:32] <Daemon404> thats why i occasionally send patches from him [23:32] <llogan> are they fun to work with? [23:32] <ubitux> fun ok [23:33] <cone-647> ffmpeg.git 03Janne Grunau 07master:ee2bc5974fe6: aarch64: NEON float (i)MDCT [23:33] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:ef1961e7fc09: Merge commit 'ee2bc5974fe64fd214f52574400ae01c85f4b855' [23:46] <cone-647> ffmpeg.git 03Janne Grunau 07master:f4d5a2cc35fc: aarch64: NEON float to s16 audio conversion [23:46] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:96a4d0c1c3cf: Merge commit 'f4d5a2cc35fcdf06ec031fabe8b0710e995fe924' [23:51] <cone-647> ffmpeg.git 03Janne Grunau 07master:8f9fe6ae3461: aarch64: NEON fixed/floating point MPADSP apply_window [23:52] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:59bfddfcdacf: Merge commit '8f9fe6ae3461ce270bce6b7083fda5ec314cdad4' [23:58] <cone-647> ffmpeg.git 03Janne Grunau 07master:3956a5e0ea46: aarch64: NEON vorbis_inverse_coupling [23:58] <cone-647> ffmpeg.git 03Michael Niedermayer 07master:cc17ff882635: Merge commit '3956a5e0ea46ed7e27ca888fe11c47986ad99261' [23:58] <Daemon404> michaelni, go aheaad and merge the cursor stuff from libav then [23:58] <Daemon404> the suggested change i forgot caused issues anyway. [00:00] --- Wed Apr 23 2014

1 0

ffmpeg.log.20140422
by burek 23 Apr '14

23 Apr '14

[00:25] <xreal> JEEB: ? [04:21] <ItsMeLenny> hello [04:22] <ItsMeLenny> how would one change the colour balance? particularly i want to multiply the red and blue channels by 2 [04:23] <ItsMeLenny> in ffmpeg [04:28] <klaxa> ItsMeLenny: maybe look at this? http://ffmpeg.org/ffmpeg-filters.html#colorchannelmixer [04:29] <ItsMeLenny> klaxa, i have that open in a text file, i never know how to adjust on 3 tone levels :P [04:29] <klaxa> neither do i, sorry :P [04:30] <ItsMeLenny> :P [05:37] <ItsMeLenny> klaxa, tried a few different values, dont think i got close :P [06:38] <sor_> is there a way to force ffplay to constantly update the video file? [08:49] <K4T> hi [08:49] <stalintrotsky> Hi, I have a question about converting parts of a video into a series of images [08:49] <stalintrotsky> IS there a way to make subtitles from a subtitle stream appear in these images? [08:51] <K4T> Can someone tell me how to pass x264 parameter "--stitchable" to ffmpeg? I tried following commands: [08:51] <K4T> ffmpeg.exe -i "teledysk.vob" -c:v libx264 -x264-params stitchable:keyint=75 output.mp4 [08:51] <K4T> ffmpeg.exe -i "teledysk.vob" -c:v libx264 -x264-params stitchable=1:keyint=75 output.mp4 [08:52] <K4T> but none of them work [08:52] <K4T> correct way using x264.exe is: x264.exe --stitchable --keyint 75 -o "teledysk.264" "teledysk.flv" [08:52] <K4T> like u can see there is no =VALUE after --stitchable in x264 [09:04] <blippyp_> stalintrotsky: ffmpeg -ss 00:00:10 -i input.mkv -vf "ass=subtitles.ass" -ss 23 -t 145 images/output%4d.png [09:12] <K4T> anyone? [09:12] <blippyp_> I'm looking right now K4T ;) [09:13] <blippyp_> I thought there was a way to pass an encoder's paramater diretly to it, but I guess I was wrong - I can't find it... [09:13] <K4T> now I`m sad ;[ [09:14] <blippyp_> be patient, someone with more knowledge about it may still answer [09:14] <stalintrotsky> blippyp_, how would I use the subtitles from inside an mkv file I'm using? [09:14] <stalintrotsky> Thanks for responding, by the way [09:15] <blippyp_> and if they don't, come back again later and ask again - (like 8am-8pm) as there are more people awake... ;) [09:15] <blippyp_> stalintrotsky: you have to export them to a .ass file first - one sec [09:15] <stalintrotsky> oh [09:15] <stalintrotsky> I think I can handle that [09:15] <stalintrotsky> Thanks blippy [09:15] <stalintrotsky> I was just wondering if there was a more convenient way [09:16] <jonascj> Hi all. What approach would you guys take to streaming images from a webcam to a webpage? How is this done for example: http://www.fishcam.com/ [09:16] <blippyp_> ffmpeg -i input.mkv -an -sn -s:c cop subs.ass [09:16] <blippyp_> np [09:16] <stalintrotsky> thanks [09:16] <blippyp_> jonascj: ffserver [09:17] <blippyp_> was that you I was talking to the other day about that? with mjpeg? [09:17] <blippyp_> it worked for me (using mjpeg) [09:18] <jonascj> blippyp_: yeah, I got ffserver and ffmpeg working :) Now I just need to figure out if I should use mjpeg, mp4 with html5 or similar. I was just wondering if you guys had any suggestions. [09:18] <blippyp_> but swf had much better video quality imo (only tried the two) [09:18] <blippyp_> I'd go with swf - it worked very nicely from my tets [09:18] <blippyp_> tests [09:19] <jonascj> i.e. just in a browser? [09:19] <blippyp_> yup [09:19] <blippyp_> hold on, let me find my files [09:19] <jonascj> My browsers just keep trying to download the mjpeg picture. If I use it as source in an <img>-tag they just display the first frame they receive [09:19] <blippyp_> I saved them as an example... [09:19] <jonascj> blippyp_: no need to - I just wanted some input :) [09:20] <blippyp_> I wouldn't use mjpeg - it really looked back compared to the swf video [09:20] <blippyp_> here is what I did: [09:21] <blippyp_> I used your ffserver.conf file - but modified the <stream webcam.mjpeg> to <Stream .webcam.swf> [09:21] <blippyp_> and also changed the format in the Stream section to swf instead of mjpeg [09:22] <blippyp_> I left everything else the same (although you might want to 'tweak' some of those settings better later) [09:22] <blippyp_> I started ffserver with: ffserver -f ./ffserver.conf [09:22] <jonascj> yeah :) [09:22] <blippyp_> then ran ffmpeg with: ffmpeg -f v4l2 -pix_fmt yuvj422p -i /dev/video0 -f alsa -i plughw:0 -af volume=1 http://localhost:8090/webcam.ffm [09:23] <blippyp_> my html page was very simple [09:23] <jonascj> http://thecheat.colding.com:8090/webcam.swf is what I have now using swf. Some settings needs to be tweaked :P [09:23] <blippyp_> <html><body><object width="320" height="240" data="http://localhost:8090/webcam.swf"></object></body></html> [09:24] <blippyp_> maybe, but your initial settings worked fine for me - but I connected locally - so across the internet in a different country or whatever might need some tweaking - you'd have to test it [09:24] <jonascj> of course my swf stream now is full height becaues I haven't wrapped it in <img>. [09:24] <blippyp_> but it worked great - I really liked it [09:24] <blippyp_> mjpeg always started at the beginning when I opened it [09:25] <jonascj> the mjpeg picture looks great for me but the swf looks horrible :) [09:25] <blippyp_> the swf was more 'real' time - it stay with what I was currently doing [09:25] <jonascj> but that must be a matter of settings. [09:25] <blippyp_> really? [09:25] <blippyp_> weird? [09:25] <jonascj> http://thecheat.colding.com:8090/webcam.swf [09:25] <blippyp_> checking it out [09:25] <jonascj> compared to http://thecheat.colding.com:8090/webcam.mjpeg [09:27] <jonascj> But I'll just have to fiddle the settings I think. Maybe enable some more libraries in my ffmpeg [09:27] <jonascj> right now it is compiled with "./configure --enable-libv4l2". Maybe having some other options would be good [09:28] <jonascj> blippyp_: thank you for your input - I'll try to make something round the .swf format or maybe .mp4 which is supported in html5 and newer browsers. [09:29] <blippyp_> yeah - not sure why - but if you tweak your settings it would alter how the video is displayed... [09:29] <stalintrotsky> blippy, is there a filter for ssa subs? [09:29] <blippyp_> like I said, because I did mine locally, it may have been much better for me - you'd have to test it - but I can certainly agree with you that you mjpeg does look better [09:30] <blippyp_> if you wanted to stay 'with the times' mp4 or webm would likely be the way to go [09:30] <blippyp_> ssa? no - just ass format - unless you can find an external filter for it [09:31] <blippyp_> jonascj: no problem, it was fun - I learned a few things from it - and I'm just as happy as you - I meant to figure out how to use ffserver again anyways, so it was a pleasure... ;) [09:32] <blippyp_> that fishcam is a cool idea.... [09:32] <blippyp_> I love fish [09:40] <K4T> where can I download FFMPEG build with libfdk_aac enabled? [09:40] <blippyp_> what os are you using? [09:40] <K4T> windows [09:40] <K4T> pls dont tell me that I have to compile it ;[ [09:40] <blippyp_> I'm guessing you'd have to re-compile it yourself [09:40] <blippyp_> sorry [09:41] <K4T> nooooooooooooooooooooooooooo :< [09:41] <blippyp_> if you can get on Arch, it's enabled by default [09:41] <blippyp_> either way - it's probably much easier to encode under any linux distro [09:42] <K4T> is it hard to compile it on Windows? [09:42] <K4T> or maybe is there any other encoder which is build into ffmpeg on windows? [09:42] <blippyp_> I just re-compiled mine to enable the frei0r plugins (was very disappointed with them so far though) [09:42] <blippyp_> I doubt it's any harder on windows than linux - but you need the right software and libraries [09:43] <K4T> hm, so you can compile it on Windows, yes? You have environment ready, yes? [09:43] <blippyp_> what do you mean by other encoder which is built into ffmpeg? [09:43] <blippyp_> yes you can compile on windows [09:44] <K4T> I though you can compile it now casue you have everytinhg prepared to do this :P I think I missunderstand you. [09:44] <blippyp_> there was someone on here the past couple of days who was doing exactly that - if you stick around and keep an eye out and ask, you'll likely find someone who is familiar enough with doing it on windows who can help give you pointers [09:45] <blippyp_> it's not that I have 'everything' prepared - but much of it [09:45] <K4T> I have to encode audio to AAC, is it possible with current ffmpeg build on windows? [09:45] <blippyp_> and the extra libraries are very easy to get with a package manager... ;) [09:45] <blippyp_> it should be [09:46] <blippyp_> but I think it's still experimental - I use ac3 (mostly because I couldn't be bothered by the extra typing required for the aac) [09:48] <K4T> I will try to compile :< [09:49] <blippyp_> it's not that hard - just takes a little time and effort at first... ;) [09:50] <K4T> https://trac.ffmpeg.org/wiki/CrossCompilingForWindows this looks good and easy [09:52] <K4T> which Linux distro should I choose to compile ffmpeg for windows? [09:53] <blippyp_> that probably doesn't matter - you might get away with using a livecd, and then you wouldn't need to install anything - Just burn it on DVD or put it on a USB and boot into it... ;) [09:54] <K4T> I will use Ubuntu on VBox [09:54] <K4T> thank you for advices [09:54] <blippyp_> no problem - hope it works out for ya :) [09:54] <K4T> same :D [09:54] <blippyp_> ubuntu is probably a good choice for you tbh [10:27] <brontosaurusrex> K4T, you can also use something lighter, like #! [10:31] <K4T> like what? [10:32] <brontosaurusrex> crunchbang < a lite debian respin [10:32] <K4T> I have installed Ubuntu on VBox right now, trying to compile ffmpeg [10:32] <K4T> but I think it will be a long journey for me... [10:32] <brontosaurusrex> i used to compile ffmpeg there all the time (only nix version thought) and it was working fine [10:35] <brontosaurusrex> nothing wrong with ubuntu either anyway ... [10:53] <K4T> brontosaurusrex, I discovered that I can use libvo_aacenc to encode audio with AAC with official ffmpeg build. Hope it will be enough. [12:47] <K4T> I`m trying to make 2 pass encoding with ffmpeg with following command: ffmpeg.exe -y -i "movie.vob" -c:v libx264 -pass 1 -an -f mp4 NUL && ffmpeg.exe -i "movie.vob" -c:v libx264 -pass 2 "ouput.mp4" [12:48] <K4T> but after 1 pass I`m getting that error: Error while opening encoder for output stream #0:0 - maybe incorrect parameters such as bit_rate, rate, width or height [12:49] <droetker2> I am trying to find a way, to download an ism/manifest. ISMDownloader can download the highest quality, but the highest quality, for some reason is bad. I want to download "QualityLevel Index="1" and mux into mkv. How? [12:49] <K4T> here is full log: http://pastebin.com/dtavXfF2 [12:49] <droetker2> Any ismdownloader available, that allows me to pick the quality level? [12:51] <JEEB> K4T, [libx264 @ 00000000047b1a00] constant rate-factor is incompatible with 2pass. [12:51] <JEEB> this is what derps your second pass :P [12:51] <JEEB> you didn't set any rate control -> CRF 23 is the default [12:51] <K4T> but I didnt set constant reate factor in my command [12:51] <JEEB> and CRF doesn't need multiple passes [12:51] <JEEB> yes, it's the default [12:51] <K4T> ow, so what should I do to remove CRF? [12:51] <JEEB> saner than setting some random bit rate [12:52] <JEEB> just set a rate control mode [12:52] <JEEB> if you need a specific size [12:52] <JEEB> basically set a -b:v [12:52] <JEEB> video bit rate [12:53] <K4T> thank you, I`m trying [12:53] <JEEB> but yeah, you only need multipass if you want the best results hitting a specific file size (average bit rate) [12:53] <JEEB> CRF is what you want to use if you only care about getting some level of quality [12:54] <blippyp_> K4T: are you limited in your file size? otherwise you should just use a crf (no second pass) [12:54] <K4T> I`m just trying to reconstruct encoding setting from VIMEO video file which is prepared for mobile devices [12:55] <K4T> settings* [12:55] <blippyp_> what's the resolution of the video? [12:55] <K4T> 480x270px [12:56] <JEEB> anyways, rate control doesn't really have much to do with things related to that :) [12:56] <JEEB> it's just that vimeo wants to make sure that no output is bigger than X [12:57] <JEEB> even if it means that quality can get derped (it really depends on the source clip) [12:57] <blippyp_> what's the maximum size limit? and how long is the video? [12:58] <K4T> blippyp_, it doesnt metter because I will have to encode many different video files, I just need encoding settings [12:58] <K4T> I think JEEB helped me [12:58] <blippyp_> K4T: determine your bitrate by this: MaxFileSize*8192/TimeOfVideo [12:59] <K4T> ok [12:59] <JEEB> generally if you want to support most mobile devices, -profile:v mainline , -level:v 3 and then keep refs within the level [12:59] <JEEB> and the more stuff you want to support the more lax your settings can be [12:59] <JEEB> newer devices support pretty much anything you can poke them with, up to high profile H.264, level 4 or more [12:59] <JEEB> and not mainline, baseline profile [13:00] <JEEB> and if you don't care about hitting an exact file size, just use crf, use the highest crf value that still looks good [13:00] <JEEB> start from 23 or so, and go down if it looks bad, or go up if it looks good [13:00] <JEEB> you can test with a few thousand frames or so, not the whole clip [13:01] <K4T> thank you for advices, they are very important for me [13:17] <Mavrik> hmm, I think you can just put Main on pretty much everything right now, haven't seen a device that wouldn't do it for awhile [13:21] <JEEB> yeah, it all depends on what you're aiming for [13:30] <Jeroi> hello [13:31] <Jeroi> I have problems to use fmmpeg libs with windows Qt Creator qt5 minge 4.8.1 [13:31] <Jeroi> complains about errrno_t don't define type [13:32] <Jeroi> size_t was not decalred in this scope [13:33] <Jeroi> and similar errors [13:33] <Jeroi> I have include to pro file mingw/include folder [13:35] <K4T> is it possible to change video resolution, but force width to be equal some const value so height will be automatic calculated by ffmpeg? [13:38] <blippyp_> K4T -vf "scale=-1:360" [13:39] <blippyp_> -1 on either w or h keeps the aspect ratio [13:39] <K4T> yes, but this is problematic when video height is not div by 2 ;/ [13:39] <JEEB> the scale video filter examples contain a way to do that right :P [13:39] <K4T> [libx264 @ 000000000432eb00] height not divisible by 2 (480x203) [13:39] <JEEB> for things that need mod2 [13:39] <JEEB> or whatever other mod [13:40] <JEEB> http://ffmpeg.org/ffmpeg-all.html#Examples-61 [13:40] <JEEB> "Increase the size, making the size a multiple of the chroma subsample values" [13:41] <JEEB> this is what you need, the chroma subsampling (hsub/vsub) usage [13:42] <K4T> what kind of black magic is this? :O [13:46] <K4T> should I resize in 2nd pass too? [13:47] <JEEB> of course [13:47] <JEEB> otherwise the pictures you'd be coding wouldn't be the same with both passes [13:48] <K4T> scale=480:trunc(ow/a/2)*2 - this is solution for height which have to be div by 2 [13:48] <K4T> JEEB, thank you one more time [14:39] <axelri> Hi! I've decoded a png to an AVFrame and want the final video to display that picture during several frames in the video file. My current solution is to do an "avcodec_encode_video2" for every frame I want to display, and then write the allocated packet to file. This seems ineffecient though, since I'm encoding the same picture multiple times. Can I write the same packet multiple times to the file, or will that mess up the video timi [15:00] <K4T> heh, that pad filter is so confusing ;/ [15:02] <klaxa> in how far? [15:02] <klaxa> i think the example makes it pretty clear [15:03] <dannyzb> when i encode a file ( mp4 output ) to 2 outputs : 1 standard MP4 and one HLS format , but same video track , does it only encode once ? [15:04] <dannyzb> ie. same encoding parameters in different containers for output [15:04] <JEEB> it encodes twice [15:04] <dannyzb> how do i avoid dual encoding? [15:04] <JEEB> if you are encoding the same kind of tracks you might want to separate encoding and final muxing [15:05] <dannyzb> how do i do that? [15:05] <JEEB> you could use mkv or mpeg-ts or nut to output something first into stdout, and then you grab that with another ffmpeg or so, and then you -c copy there to those two outputs [15:06] <JEEB> ffmpeg -i INPUT ENCODER_SETTINGS -f matroska - | ffmpeg -i - -c copy out.mp4 -c copy WHATEVER_YOU_DO_WITH_HLS [15:06] <JEEB> something like that? [15:07] <dannyzb> JEEB ; would that use a lot of ram ? [15:08] <JEEB> I don't see it using much more, of course you'd be running two ffmpeg processes now, but you are not running two encoders any more [15:08] <JEEB> and the stdin stuff is generally buffered [15:08] <klaxa> oh, you can use '-' for output now? [15:08] <JEEB> I think you've always been able to do that as long as you set the format? [15:08] <JEEB> there's no extension so ffmpeg can't guess [15:08] <klaxa> i think i had some problems with it and used pipe:1 [15:09] <dannyzb> JEEB: perfect , thanks ! [15:09] <JEEB> pipe:1 should be the same [15:09] <dannyzb> uuhhh now i remembered something [15:09] <klaxa> maybe i used it incorrectly back then [15:09] <dannyzb> i have one input , and i output 3 times : 1 highres mp4 , 1 highres HLS , 1 lowres MP4 [15:09] <JEEB> also the other ffmpeg should probably set -f before -i as well, to make stuff go faster [15:10] <dannyzb> can i split the output from the source to go to both the pipe and a different outpu ? [15:10] <JEEB> dannyzb, well then you will have to do two encodes with the first process and have that output one to stdout and the other somewher else :P [15:10] <dannyzb> yeaaa cool [15:10] <dannyzb> one last thing , i've seen online ffmpeg doesn't use separate threads for 2 outputs .. is that true ? [15:10] <dannyzb> that would make a serious waste of resources [15:11] <JEEB> it does use quite a few threads but you'd have to ask someone who knows lavf for that :P [15:11] <dannyzb> first i need somebody who can tell me what lavf is ;O) [15:11] <JEEB> shorthand for libavformat [15:11] <JEEB> the muxing/demuxing/input/output thing [15:11] <JEEB> although I guess it's more correct to say AVIO [15:12] <JEEB> or so [15:12] <JEEB> for the input/output part [15:12] <JEEB> since you give an AVIOContext to lavf if you want to open some custom IO of your own [15:12] <dannyzb> is uuhuhh i have an idea [15:12] <dannyzb> what if i make a double pipe [15:12] <dannyzb> ie .. input + copy first pipe - second pipe gets data as input so HDD is only read once [15:12] <dannyzb> then the third pipe is what you said [15:13] <JEEB> you are not making any sense :P [15:14] <dannyzb> ffmpeg -i file -f copy | -i - -o #firstoutput -o #secondoutput -f matroska | #output HLS+mp4# [15:15] <dannyzb> #secondoutput = - [15:15] <JEEB> I don't really see that making any more sense than just doing the encoding in the first process [15:15] <dannyzb> it doesn't for MP4+HLS [15:15] <dannyzb> it does when i encode two different files [15:15] <dannyzb> 2 output streams to 3 files would read twice [15:16] <JEEB> what [15:16] <JEEB> you're still not making sense [15:16] <JEEB> you've got one input, you encode that for two outputs [15:16] <JEEB> one goes into pipe to get muxed for two things [15:16] <JEEB> the other goes straight to output [15:16] <JEEB> done [15:16] <dannyzb> ah right .. [15:16] <dannyzb> and it would multithread perfectly [15:17] <dannyzb> one thread per output format [15:17] <JEEB> the fuck I know, but I'm pretty sure if you will get bottlenecks they'd be somewhere else [15:17] <JEEB> than just because a thread not being somewhere [15:17] <dannyzb> JEEB: not sure , what i prefer about a pipe is i create a second ffmpeg process [15:17] <JEEB> I don't know about lavf or the IOs [15:17] <JEEB> but I'm pretty sure that if what I just fails [15:17] <JEEB> it fails because of something else [15:18] <dannyzb> JEEB : I read everywhere online ffmpeg wastes a lot of resources for splitting output [15:18] <dannyzb> separating output to 2 separate ffmpeg processes negates that [15:18] <JEEB> I have no idea, ask actual lavf/AVIO knowing people [15:18] <JEEB> but don't assume [15:18] <dannyzb> what channel is that? [15:19] <JEEB> if you really want you can poke the -devel channel, but I think you're just worrying about things that don't exist [15:21] <dannyzb> JEEB: when 80% of my HDD/CPU is going to conversions i just wanna make sure (: [15:21] <JEEB> that sounds like stuff is properly threaded just fine :P [15:22] <dannyzb> well [15:22] <JEEB> also you're running two libx264 encodes [15:22] <dannyzb> everything online says otherwise but it's from 2011/2012 when separate outputs were just added .. it's likely not relevant anymore [15:22] <JEEB> as well as most probably resizing for the "low quality" stream [15:22] <JEEB> those are more probably doing any kind of bottlenecking if anything is [15:23] <dannyzb> JEEB: yea , thing is , i found out decode takes considerably more resources than the encode [15:23] <JEEB> then your input is either consuming to read or consuming to decode [15:23] <dannyzb> JEEB : SATA2 Raid10 with nothing running besides 1 encode [15:24] <dannyzb> i wouldn't call it a read bottleneck [15:24] <dannyzb> so it's decode [15:24] <JEEB> parsing a lot of shit can take cpu time and friends just fine :P [15:24] <JEEB> more likely then that it's the decoding, but still [15:24] <JEEB> assumption is the mother of fuck-ups [15:24] <dannyzb> maybe it's the resize? [15:25] <JEEB> resize can very well be a bottleneck as well [15:25] <JEEB> since swscale is not exactly great [15:25] <JEEB> anyways, herp derp [15:36] <dannyzb> JEEB : basically , i noticed that even though i have identical output formats , the higher resolution the source file - the longer it takes to encode [15:36] <dannyzb> is there a way to make it input faster ? [15:40] <c_14> I'm pretty sure it's not that the input is slower, but that ffmpeg has to work with more pixels. Which in turn takes longer. [15:42] <dannyzb> c_14 : the output is the same resolution - how would more pixels slow it down ? interpolation ? [15:43] <c_14> Pretty much, it can't just delete pixels because that would destroy the picture. [15:44] <JEEB> you were wondering about threading and you are so oblivious to how this shit actually works? color me surprised :P [15:44] <JEEB> s/wondering/oh so sure that your problem was the lack of threading in the IO/ [15:44] <JEEB> anyways, your input has to be fully decoded [15:44] <JEEB> then it gets scaled and possibly converted to another pixel format [15:44] <JEEB> then it gets fed to the encoder [15:45] <dannyzb> if for some reason it does any of those things twice , that would be a massive waste of CPU .. either that or running any of those on the same thread as output [15:45] <dannyzb> thats why i was interested (: [15:46] <JEEB> so both your outputs scale/change pixfmts? [15:46] <JEEB> and the scaling is the same? [15:47] <JEEB> if yes, then it might be doing that twice, once for each encode (if you have the setup you noted before) [15:47] <dannyzb> -x264opts crf=$crf:vbv-maxrate=800:vbv-bufsize=270000 -vcodec libx264 -movflags +faststart -preset veryfast -tune film -y /var/www/media/video/hd/{$hash}.mp4 [15:47] <dannyzb> two outputs with these settings , second one with a different CRF and resolution [15:47] Action: JEEB facepalms [15:47] <Hello71> {$hash} [15:48] <JEEB> yeah, sure -- ok, you have made it clear that you're ignoring what I write and you are just oblivious to shit [15:48] <JEEB> thanks for confirming that I'm OK to use my time in a more productive way [15:48] Action: JEEB is off [15:48] <dannyzb> JEEB: dude wtf .. resolution is different and pixfmt how do i know .. thought thats the encoder job [15:50] <dannyzb> and if you actually read my response you would see i said i use different resolutions for each output [15:54] <dannyzb> JEEB: ignoring the bickering -- thanks for the help man [16:40] <dannyzb> JEEB: https://trac.ffmpeg.org/wiki/Creating%20multiple%20outputs#Teepseudo-muxer [16:40] <dannyzb> apparently piping was integrated into ffmpeg internally [19:05] <t4nk594> hi, im trying to cross-compile ffmpeg from an arch linux machine for win64. i compiled opus/vpx/ogg in a separate directory (with this https://trac.ffmpeg.org/wiki/UbuntuCompilationGuide guide) but i get the error while executing the configure which results in: ERROR: opus not found [21:24] <dericed> is there a preferred way to convert PAL to NTSC. I'm using -r:v ntsc to duplicate frames, but would a pulldown process look better? [21:31] <blippyp_> dericed: http://lists.ffmpeg.org/pipermail/ffmpeg-user/2013-July/016098.html [21:36] <blippyp_> jonascj: did you get your camera working the way you want yet? [21:48] <droetker> I want to download a ism/manifest. The manifest, has several qualitylevels, where 0 is the best quality. I want the second best quality though. How do I convert/raw dump the "quality level 1" of the ism/manifest [22:17] <sor_> is there a way to force ffplay to constantly update the video file? [00:00] --- Wed Apr 23 2014

1 0

ffmpeg-devel.log.20140421
by burek 22 Apr '14

22 Apr '14

[01:45] <cone-230> ffmpeg.git 03Michael Niedermayer 07master:5b19fc7aae98: avformat/pmpdec: remove cur_dts timestamp hack [02:56] <cone-230> ffmpeg.git 03Michael Niedermayer 07master:369cdf917aaa: avformat/asfenc: dont allow non interleaved packets [11:04] <cone-699> ffmpeg.git 03Michael Niedermayer 07master:f3743901d7df: avformat: Fix decoder search in find stream info [11:51] <cone-699> ffmpeg.git 03Michael Niedermayer 07master:59c7615d58b5: avfilter/filtfmts: Support dynamically allocated in/outputs [14:02] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:cb95c1935f43: qt-faststart: Check offset_count before reading from the moov_atom buffer [14:02] <cone-699> ffmpeg.git 03Janne Grunau 07release/1.1:7a2254ab26b2: arm: hpeldsp: fix put_pixels8_y2_{,no_rnd_}armv6 [14:02] <cone-699> ffmpeg.git 03Diego Biurrun 07release/1.1:a32e45df8275: configure: Support preprocessor macros as header names [14:02] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:c40631baf77f: Merge commit 'cb95c1935f43ee6ae7fd15439df0f674e809f180' into release/1.1 [14:02] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:f6cb3827dfb8: Merge commit '7a2254ab26b24255a4d6ea476ba5fd38525a9bb3' into release/1.1 [14:02] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:fa29e89d4165: Merge commit 'a32e45df82751784f9461f01310e63f4bed721dc' into release/1.1 [14:18] <cone-699> ffmpeg.git 03Luca Barbato 07release/1.1:29b3ca743dd0: drawtext: Drop pointless header [14:18] <cone-699> ffmpeg.git 03Luca Barbato 07release/1.1:b00444c0bbbc: configure: Update freetype check to follow upstream [14:18] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:be9074debcfe: Merge commit '29b3ca743dd02efbe00565c0b997e735e4daa772' into release/1.1 [14:18] <cone-699> ffmpeg.git 03Alexander Strasser 07release/1.1:fb487e2f6668: configure: Special case libfreetype test [14:18] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:4ccde2d356e7: Merge commit 'b00444c0bbbc78ae477ec697d6569893609f420d' into release/1.1 [14:32] <cone-699> ffmpeg.git 03Reinhard Tartler 07release/1.1:f1a2364cebe4: Update Changelog for 9.12 [14:32] <cone-699> ffmpeg.git 03Martin Storsjö 07release/1.1:d791e8c990a7: doc: Point to the correct, actually maintained gas-preprocessor repo [14:32] <cone-699> ffmpeg.git 03Reinhard Tartler 07release/1.1:4c1e4bc6af15: Prepare for 9.13 Release [14:32] <cone-699> ffmpeg.git 03Johan Andersson 07release/1.1:fc53a09909ae: cmdutils: update copyright year to 2014. [14:32] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:1fabe4ae8cc0: Merge commit 'f1a2364cebe472a98a5ea157071f4eee4429ce9f' into release/1.1 [14:32] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:ed9ab30f4cca: Merge commit 'd791e8c990a752435d993df01516ebe469d94535' into release/1.1 [14:32] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:238532700e84: Merge commit '4c1e4bc6af1596995e895cd7da9f3b436051114c' into release/1.1 [14:32] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:6d33c8ec8490: Merge commit 'fc53a09909ae046dbe4d339bffc744cfc328253e' into release/1.1 [14:45] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:af40847681b9: h264: check buffer size before accessing it [14:45] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:30e58e65e543: h264_refs: make sure not to write over the bounds of the default ref list [14:45] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:c9f17afe4e3c: Merge commit 'af40847681b978441fdd25276685414f9e17820a' into release/1.1 [14:46] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:7434225ed15d: Merge commit '30e58e65e543d04d5e52bf188b31abf4dd41a104' into release/1.1 [15:09] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:1147d39bca13: h264: reset data_partitioning if decoding the slice header for NAL_DPA fails [15:09] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:b6a0f5cde808: h264: do not use 422 functions for monochrome [15:10] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:7e513d85e80d: h264: check that execute_decode_slices() is not called too many times [15:10] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:58bdec57fa92: Merge commit '1147d39bca135df8673031d16989ba486a52172a' into release/1.1 [15:10] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:b9550ea11617: Merge commit 'b6a0f5cde808d84626f07b11c381cc6ef0954940' into release/1.1 [15:10] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:58e2d529b3df: Merge commit '7e513d85e80d730718695d09fcaf0295ae24699e' into release/1.1 [15:25] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:7f604a048e9b: h264: reject mismatching luma/chroma bit depths during sps parsing [15:25] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:175b53d051cf: h264: limit allowed pred modes in ff_h264_check_intra_pred_mode() to 3 [15:25] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:9081b4ddddee: Merge commit '7f604a048e9b6128cdf9ce7e95f21d1a9822ba39' into release/1.1 [15:25] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:40e6c148f5dc: h264/ff_h264_check_intra_pred_mode: fix input value check [15:25] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:f294748cbce9: h264: fix size of arrays in ff_h264_check_intra_pred_mode() [15:25] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:f6c1fdba22dc: Merge commit '175b53d051cf9e17583106c828c35d169f335ea3' into release/1.1 [15:39] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:7fd6c9fb4957: h264: reset first_field if frame_start() fails for missing refs [15:39] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:7e8d27c8510a: h264: reset ref count if decoding the slice header fails [15:39] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:d8e89a37267f: h264: reset data partitioning at the beginning of each decode call [15:39] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:11dfd2a74a33: Merge commit '7fd6c9fb49573870b48a995947edc83eedd7e86e' into release/1.1 [15:39] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:6961f9f9f4a0: Merge commit '7e8d27c8510a024e6d66d2e1116bb8692d9cb98a' into release/1.1 [15:39] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:a23748a80429: Merge commit 'd8e89a37267f276afd404bd062e5112a336d1a36' into release/1.1 [16:08] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:c9be276f8b04: mpegvideo: set reference/pict_type on generated reference frames [16:08] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:ff409c7d80b6: adx: check that the offset is not negative [16:08] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:ea1806ce650f: sgidec: fix buffer size check in expand_rle_row() [16:08] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:8e5a02151227: Merge commit 'c9be276f8b040e7d441a77f774afe94b6f137be6' into release/1.1 [16:09] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:fcaf78f5f941: Merge commit 'ff409c7d80b6781e4b81dd860badb569061ddaa5' into release/1.1 [16:09] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:8401ed651b2c: Merge commit 'ea1806ce650f0502dd25939c335b9216fa4a955f' into release/1.1 [16:24] <cone-699> ffmpeg.git 03Janne Grunau 07release/1.1:b701e26a4e89: mpeg12: check scantable indices in all decode_block functions [16:24] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:fa73d13678c4: Merge commit 'b701e26a4e89e3816490bc74bc47f82e6b7f7002' into release/1.1 [16:32] <cone-699> ffmpeg.git 03Diego Biurrun 07release/1.1:2eb15cdeef29: pthread: flush all threads on flush, not just the first one [16:32] <cone-699> ffmpeg.git 03Pierre Lejeune 07release/1.1:1bccf68caeee: build: Use pkg-config for openjpeg [16:32] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:0162ad1a5997: Merge commit '2eb15cdeef29eb8a0a32658154decba94b4b89cb' into release/1.1 [16:32] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:47e18e2e03c9: Merge commit '1bccf68caeee787139f22679e1300a5eea048cf5' into release/1.1 [16:42] <cone-699> ffmpeg.git 03Luca Barbato 07release/1.1:b1432e905d46: avfilter: Add missing emms_c when needed [16:42] <cone-699> ffmpeg.git 03John Stebbins 07release/1.1:d375d6395c0a: matroskaenc: allow override of "writing application" tag [16:42] <cone-699> ffmpeg.git 03John Stebbins 07release/1.1:9f7119b7fe00: movenc: allow override of "writing application" tag [16:42] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:1552b850db54: Merge commit 'b1432e905d462198a8d7834b256848f2b9e4f8e6' into release/1.1 [16:42] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:8ca0d74d5945: Merge commit 'd375d6395c0a5203932d89c087f77a5757a08138' into release/1.1 [16:42] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:2ce46a959c20: Merge commit '9f7119b7fe003e94d72e81f3aa922568c2827dce' into release/1.1 [16:47] <j-b> 'morning [16:56] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:ee21d270f8dc: af_channelmap: fix ONE_STR mapping mode [16:56] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:03562c44c0c1: avi: Improve non-interleaved detection [16:56] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:cb29eb144b81: Merge commit 'ee21d270f8dc5fdb2263f5dd87407aae9fe1ecd1' into release/1.1 [16:56] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:6a356801952b: Merge commit '03562c44c0c1e59d2a598390c5b2181ac406611c' into release/1.1 [17:01] <michaelni> afternoon [17:02] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:d21bf0d27b54: resample: fix avresample_get_delay() return value [17:02] <cone-699> ffmpeg.git 03Paul B Mahol 07release/1.1:0c5e6c542f5e: bytestream: add functions for accessing size of buffer [17:02] <cone-699> ffmpeg.git 03Justin Ruggles 07release/1.1:d946e4c1bcfb: bytestream: add bytestream2_copy_buffer() functions [17:02] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:6e3830cca297: Merge commit 'd21bf0d27b547adcaabaa28d475e6b9f97dfe20a' into release/1.1 [17:02] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:da32e3c39819: Merge commit '0c5e6c542f5efffb4cd770e84967478b0076a6e5' into release/1.1 [17:02] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:260f37ee4d03: Merge commit 'd946e4c1bcfb6accd97f6d4cbfa324ef01913c8d' into release/1.1 [17:23] <cone-699> ffmpeg.git 03Justin Ruggles 07release/1.1:85b8b169175a: tiffdec: use bytestream2 to simplify overread/overwrite protection [17:23] <cone-699> ffmpeg.git 03Anton Khirnov 07release/1.1:dccac759d39e: lavr: allocate the resampling buffer with a positive size [17:23] <cone-699> ffmpeg.git 03Martin Storsjö 07release/1.1:b77819afbc44: rtmpproto: Make sure to pass on the error code if read_connect failed [17:23] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:63b38d4a9181: Merge commit '85b8b169175a55fc862e01ecc96f649374bc14d2' into release/1.1 [17:23] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:10b8ae5b108d: Merge commit 'dccac759d39e761398d3d4172ae0b98c0186efa5' into release/1.1 [17:23] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:238360665b26: Merge commit 'b77819afbc44016143b4a11b46109625203c4cef' into release/1.1 [17:29] <cone-699> ffmpeg.git 03Luca Barbato 07release/1.1:ca14a2d0e407: configure: Use the right pkgconf file for openjpeg [17:29] <cone-699> ffmpeg.git 03Luca Barbato 07release/1.1:ce364088ef53: configure: Support older version of openjpeg1 [17:29] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:fdcdfb36b74b: Merge commit 'ca14a2d0e40720d018421e06995737e5682c33c9' into release/1.1 [17:29] <cone-699> ffmpeg.git 03Michael Niedermayer 07release/1.1:a68d4f55debc: Merge commit 'ce364088ef53dbaf90e74a471d12b732b697e0cd' into release/1.1 [17:30] <microchip_> wow, a commit fest :D [18:48] <Daemon404> oh right [18:48] <Daemon404> i go to add my stuff to fic.c i nffmpeg [18:48] <Daemon404> and it has a bnch of shit michael patched with no cc to the maintainer [18:48] <Daemon404> goddammit [18:50] <nevcairiel> even if he had, its not exactly obvious that you consider yourself that, since kostya is the author of the patch and all [18:51] <Daemon404> i am in MAINTAINERS for it [18:51] <Daemon404> at michael's request even [18:51] <Daemon404> your argument is invalid. [18:51] <Daemon404> nevcairiel, and the patches i am applying (curor rendeing) are mine [18:51] <nevcairiel> thats irrelevant, isnt it [18:52] <Daemon404> what the fuck is the point of being added as a maintainer IF IT DOESNT MEAN ANYTHING [18:52] <nevcairiel> i see no mention of it in MAINTAINERS [18:53] <Daemon404> inded i musnt have pushed [18:53] <Daemon404> but also my name is in copyright on the top [18:53] <Daemon404> of the actual file [18:54] <Daemon404> indeed i even wrote ost of it [18:56] <Daemon404> (also this happened with stuff i am listed for regularily) [20:34] <cone-699> ffmpeg.git 03Kostya Shishkov 07master:93e15a323871: fic: use correct IDCT [20:34] <cone-699> ffmpeg.git 03Michael Niedermayer 07master:451c78740ae3: Merge commit '93e15a323871613fd26f1f1e317029a50b5b24ca' [20:34] <michaelni> Daemon404, if you want to maintain fic.c or some other files, please make sure that you are listed in MAINTAINERs for them [20:46] <Daemon404> i thought i was for some reason [20:51] <Daemon404> michaelni, kostya says it is not a typo [20:52] <michaelni> ok, thx [22:48] <wm4> ubitux: is the "rotate" video metadata parameter in clockwise dirextion? [22:48] <ubitux> yes [22:48] <ubitux> at least it seems so [22:49] <wm4> weird [22:49] <ubitux> why? [22:49] <wm4> well, the mathematical standard direction is counterclockwise, isn't it? [22:49] <ubitux> ah i see what you mean [22:50] <wm4> I wonder what direction Libav will choose :) [22:50] <ubitux> yeah well i don't know, consider the value as a value to evaluate to get it correct [22:50] <ubitux> the other way around probably [22:52] <ubitux> wm4: i guess the spirit is that you don't care about how it's rotated but how to rotate to get it correct [22:53] <ubitux> so the metadata just gives the rotation fix [22:53] <Daemon404> fail [22:53] <Daemon404> we should use radians [22:53] <Daemon404> duh. [22:53] <wm4> well, whatever [23:23] <ubitux> Daemon404: what don't you go for av_make_q? [23:25] <ubitux> -#define AV_TIME_BASE_Q (AVRational){1, AV_TIME_BASE} [23:25] <ubitux> +#define AV_TIME_BASE_Q av_make_q(1, AV_TIME_BASE) [23:25] <ubitux> something like this should be enough isn't it? [23:26] <wm4> it's not constant anymore [23:26] <ubitux> av_make_q is static inline [23:26] <wm4> I still don't get why this is a problem [23:26] <ubitux> it should do the trick [23:26] <ubitux> wm4: "c++" probably [23:26] <wm4> ubitux: doesn't change the fact that you can't use it as static initializer [23:27] <Daemon404> ubitux, no [23:27] <Daemon404> because we, and otehr use e.g. &AV_TIME_BASE_Q [23:27] <wm4> AV_TIME_BASE_Q is pretty much just a more convenient way to write 1/AV_TIME_BASE [23:27] <Daemon404> a *lot* [23:27] <wm4> so it's strictly a helper [23:27] <wm4> the timebase might be changed, but this macro won't ever change [23:27] <ubitux> ah indeed static init, forgot that [23:27] <Daemon404> i gave up due to all the bieshedding from both projects [23:27] <Daemon404> and settled on adding a new fun [23:27] <Daemon404> c [23:27] <wm4> good [23:27] <Daemon404> for c++ users [23:28] <Daemon404> if *this* offends someone [23:28] <Daemon404> then fuck it [23:28] <Daemon404> :D [23:28] <wm4> well you could also just ifdef cplusplus it [23:28] <wm4> or is that a taboo [23:28] <ubitux> it should be fine [23:28] <ubitux> (but you can use av_make_q in that function ;)) [23:28] <wm4> I mean, I noticed the headers don't have the typical extern "C" thing [23:28] <Daemon404> ubitux, not that it matters [23:28] <wm4> usually libs do that to be somewhat c++ friendly [23:29] <wm4> but maybe being nice to C++ is a religious taboo around here [23:29] <Daemon404> wm4, everybody already does extern "C" around libav* headers [23:29] <Daemon404> so its non-issue at this point [23:29] <Daemon404> damage done etc [23:29] <ubitux> Daemon404: sure, patch is fine with me as is [23:29] <Daemon404> i figured id send it [23:30] <Daemon404> its the last thing sitting in patchwork [23:30] <Daemon404> befor my backlog is clear [23:31] <wm4> hm, extern variables are evil too? [23:31] <wm4> in the API I mean [23:31] <Daemon404> we worked hard to remove them [23:31] <Daemon404> also theyre epic pain on windows [23:31] <wm4> ok [23:31] <Daemon404> IMO you shouldnt even need av_time_base much [23:31] <Daemon404> you should be using the stream tb [23:32] <Daemon404> a lot of usage of avtimebase in the wild seems to be quite wrong [23:32] <Daemon404> (read: people dont understand timebases) [23:32] <wm4> AFAIK you need it for seeking and for the file duration [23:32] <Daemon404> you can seek relative to a stream [23:32] <Daemon404> so no you dont [23:32] <Daemon404> if you dont want to [23:32] <Daemon404> ;) [23:32] <wm4> that never worked for me [23:33] <Daemon404> works for me at work [23:33] <wm4> and which stream should I seek if there are several? [23:33] <Daemon404> the active one? [23:33] <wm4> what if multiple can be active? [23:33] <wm4> I like my videos with sound [23:33] <Daemon404> i always seek to video stream myself [23:34] <Daemon404> but its not wrong to do it either wa [23:34] <Daemon404> y [23:53] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:f34d3173fcfc: avcodec/fic: fix slice checks [23:53] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:359a719f5fb1: Merge commit 'f34d3173fcfc7f3228095d509a64c4fa4b37b575' [23:54] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:144f7a561d32: avcodec/fic: clear slice_data [23:54] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:e0cdc68bdc0e: Merge commit '144f7a561d32ae9bc657a217e2b4c6dd9a0d7df9' [23:55] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:6f243b17c537: avcodec/fic: Fix return value check [23:55] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:60fdf7dd9ea6: Merge commit '6f243b17c537646b894857d43dfdac65f85ab377' [23:58] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:1c7b71a5bdb8: mpegvideo: Use the current_picture pts [23:58] <cone-977> ffmpeg.git 03Michael Niedermayer 07master:5e5aced585b3: Merge commit '1c7b71a5bdb88ebb69734100405bbb5441b871e8' [00:00] --- Tue Apr 22 2014

1 0

ffmpeg.log.20140421
by burek 22 Apr '14

22 Apr '14

[00:54] <madbytes> I'm generating a video using filter_complex but while overlaying a video. it doesnt include its audio. How can i fix this? [00:57] <madbytes> c_14, http://pastie.org/9096356 In this command, i want "audio.mp3" to be added to the result video. [01:01] <klaxa> pastebin the output too [01:01] <klaxa> do people not read? [01:06] <madbytes> http://pastie.org/9096366 Here. [01:08] <klaxa> try adding -map 3 and scale it to 462x300 or 460x300 [01:08] <klaxa> using either -s or the scale video filter [01:21] <xreal> klaxa: When adding an external MP3 created by lame, ffmpeg doesn't care about the 1105 samples "gap" at the beginning (at 44.1 kHz that's about 25 ms). Is there a way to fix this? It's not in sync with video. [01:22] <klaxa> try using -itsoffset [01:22] <klaxa> i think that's applied before the input [01:22] <klaxa> set it to 0.025 then or -0.025 [01:22] <klaxa> not sure which way you have to shift [01:22] <xreal> klaxa: oh, can't I do it on sample base? [01:23] <klaxa> i don't think so [01:23] <klaxa> you could add samples to the mp3 [01:23] <xreal> klaxa: I'd need to add it to the video then. [01:23] <klaxa> why? [01:23] <xreal> klaxa: the video starts 0.02504558 before the audio. [01:24] <xreal> LAME adds 1105 samples at the beginning of each MP3 [01:24] <klaxa> then either remove those samples (not sure how) or set a negative offset for the mp3 [01:25] <xreal> oh wait wait [01:25] <xreal> line 141: avctx->delay = lame_get_encoder_delay(s->gfp) + 528 + 1; [01:25] <xreal> it takes care of the 529 samples encoder delay! [01:25] <xreal> that are 11,9955 ms at 44.1 kHz [01:28] <xreal> ahhhhhhhhhh [01:28] <xreal> klaxa: Are you here? [01:28] <klaxa> yes? [01:28] <xreal> klaxa: 576 samples encoder delay + 529 samples decoder delay = 1105 samples [01:29] <xreal> klaxa: addition, 1152 samples are added when using VBR [01:29] <klaxa> okay [01:29] <xreal> man, that's pretty deep LAME knowledge :D [01:30] <xreal> lame --decode --decode-mp3delay -529 [01:30] <xreal> 576 samples = forward+backward MDCT transform [01:44] <xreal> Yep, [x] ffmpeg does bugging mp3 decoding. [01:44] <xreal> buggy* [01:44] <xreal> Where can I report bugs? [02:05] <klaxa> http://www.ffmpeg.org/bugreports.html [02:16] <xreal> klaxa: do you know if it's possible to add the LAME header tag to an MP3 created by ffmpeg ? [02:18] <klaxa> no idea [02:18] <klaxa> i have never run into this issue before [02:18] <xreal> klaxa: ffmpeg seems to be in very early development state, isn't it? [02:18] <klaxa> not... really? [02:18] <xreal> klaxa: there are many bugs, I've detected over several versions in the last days. [02:19] <klaxa> well feel free to send patches to the ffmpeg-devel mailing list [02:19] <xreal> klaxa: I'm not a programmer. [02:19] <klaxa> ffmpeg's initial release was in 2000 [02:19] <xreal> I only can describe the problems. [02:20] <klaxa> then write up bugreports [02:20] <klaxa> so a developer can fix them [02:20] <xreal> klaxa: I think, nobody cares about my bugs because <= 25 ms aren't a problem for anyone in here :D [02:20] <klaxa> it's one of the important tasks of software development - writing bugreports [02:20] <klaxa> i doubt that, audio/video sync precision is extremely important [02:24] <xreal> klaxa: Is there a place where to upload a demo wav file? [02:27] <c_14> Did you read the bugreports.html? [02:27] <c_14> >Submitting Sample Media [02:30] <xreal> c_14: LoL, I read the box but not the rest :) [02:30] <xreal> => Email the ffmpeg mailing list and indicate the filename of the sample. [02:30] <xreal> urgs [02:30] <xreal> So I need a bug report on the website and on the mailing list? [02:30] <xreal> I don't want to register at 1000 services. [02:33] <xreal> Can't I write a complete bugreport and you post it? [02:39] <xreal> oh great... avconv and ffmpeg have different results. [02:41] <xreal> is there something like the "latest stable" of ffmpeg for win32? [02:43] <c_14> You can try: http://ffmpeg.zeranoe.com/builds/win32/static/ffmpeg-latest-win32-static.7z [02:44] <xreal> c_14: that's from GIT. Normally, you test the latest stable when doing bug reports. [02:44] <xreal> this seems to be stable: http://ffmpeg.zeranoe.com/builds/win32/static/ffmpeg-2.2.1-win32-static.7z [02:44] <JEEB> xreal, in general releases aren't kept in high regard in either libav or ffmpeg [02:44] <JEEB> you should generally test against current HEAD [02:45] <c_14> It even says so in the 3rd line on the bugreports.html [02:46] <xreal> Can't I write a complete bugreport and you post it? [02:52] <xreal> Hey, come on... I want to help you. [03:13] <xreal> Could anyone please have a quick look over my bugreport before posting it? [03:26] <xreal> There you go: https://trac.ffmpeg.org/ticket/3576 [03:40] <xreal> re [06:57] <metulburr> instead of resaying everything, i will just post a link https://bbs.linuxdistrocommunity.com/showthread.php?tid=2081 ....trying to convert mp4 to mpg [06:57] <metulburr> for pygame.movie [06:59] <metulburr> ffmpeg -i ORIGINAL.mp4 -vcodec mpeg1video -an -ss 61 -t 19 output.mpg [06:59] <metulburr> pridcues what i think might be the error [06:59] <metulburr> [mpeg @ 0x19c9a40] VBV buffer size not set, muxing may fail [06:59] <metulburr> but i am not sure what argument to give it [07:00] <metulburr> ^produces [10:35] <soman> Hi all. I use 'ffmpeg -i input.avi -vcodec copy -acodec copy -ss 00:02:22 -t 00:00:08 2.avi' to cut avi file but output files begins with sound but 1-2sec of blank black screen and sometimes the sound is muted but source videos is OK. what could be a problem? Xubuntu 12.04 ffmpeg version 0.8.10-4:0.8.10-0ubuntu0.12.04.1 [12:14] <anshul_> !log [12:15] <soman1> Hi all. I use ffmpeg as 'ffmpeg -i input.avi -vcodec copy -acodec copy -ss 00:00:00 -t 00:00:15 output.avi' but get about 1-2 sec of empty black screen for every output video (sound is enabled while that screen is shown) and sometimes sound is muted when the source videos is OK. What could be a problem and how to fix that? ffmpeg version 0.8.10-4:0.8.10-0ubuntu0.12.04.1 [12:17] <blippyp_> soman1: First thing I'd do is update your ffmpeg: ffmpeg version 2.2.1 Copyright (c) 2000-2014 [12:21] <soman1> blippyp_: I'll try to find some ppa for ffmpeg. [12:26] <diesel420> ffmpeg static build ftw [12:27] <soman1> I tried it now (static build) - nothing changed [12:37] <soman1> avisplit works well but output video's time is not corresponding specified time period from command line. [12:39] <blippyp_> soman1: I haven't a clue what your problem might be - But I'm guessing it's related to the codec in the original file: run mediainfo on it and find out what codec it's using [12:41] <soman1> blippyp_: here is info http://paste.kde.org/p5ywgmkhu/xsq3bf [12:44] <blippyp_> soman1: All I can say is that I never work with xvid, so it wouldn't surprise me if that was your problem - maybe wait a bit and see if someone else can comment on that who does use xvid, or has had previous experience with xvid - It use to be a very common codec to use... [12:45] <blippyp_> it's only a 13 minute video - maybe just re-encode it use x264, then run your command on it - it shouldn't take very long... [12:46] <blippyp_> ffmpeg -i input.avi -c:v libx264 -crf 0 -preset ultrafast newfile.mkv [12:47] <blippyp_> that will re-encode it quite fast (but will chew up a large chunk of space, so you'll probably want to delete it when you're done) [12:48] <soman1> blippyp_: Unrecognized option 'c:v' + Failed to set value 'libx264' for option 'c:v' [12:49] <blippyp_> you don't have x264 [12:49] <blippyp_> are you usng avcon? [12:49] <soman1> OT works with -vcodec instead of c:v [12:50] <blippyp_> -c:v works, I use it all the time - you got some other kind of build going on.... don't know what to tell you... [12:51] <soman1> Now it works... I'll try my coommand on the result mkv and will write here about results [12:52] <blippyp_> cool [12:53] <blippyp_> oops - forgot to tell you to copy the audio - you might not have any sound - did you add that? [12:55] <soman1> no. the file is about 1gb and it still works on it... It would be good to avoid blank screen at first. [12:55] <blippyp_> if not re-run this: ffmpeg -i input.avi -vcodec libx264 -crf 0 -preset ultrafast -acodec copy -scodec copy output.mkv (this will re-encode the video and copy the audio and subtitle streams as well) [12:56] <blippyp_> although it's using an mp3 codec, you might want to convert that to aac or ac3 or something [12:57] <blippyp_> I usually use ac3 and libx264 (and I split video's all the time with precision) [13:00] <soman1> unfortunately same command at mkv didn't help [13:00] <blippyp_> post your video at pomf.se and then post the link here so I can download it [13:01] <soman1> 185mb. ok? [13:01] <blippyp_> really that big for 13 minutes? [13:01] <blippyp_> that's too big [13:01] <soman1> Yes. [13:01] <blippyp_> it will only accept 50mb [13:01] <blippyp_> do you know where you originally got it still? [13:02] <soman1> I can send you it by p2p if 185mb is OK for you [13:02] <blippyp_> no, I'd rather not [13:02] <blippyp_> did you conver the audio or did you leave it as an mp3? [13:03] <soman1> left as mp3. btw result mkv has sound [13:03] <blippyp_> k, then try this: [13:04] <blippyp_> ffmpeg -i input.avi -vcodec libx264 -acodec ac3 -scodec copy output.mkv [13:05] <blippyp_> oops - don't forget the -crf 0 -preset ultrafast after the libx264 [13:08] <soman1> blippyp_: no success http://paste.kde.org/pqyc0picf/mf6k2y [13:10] <blippyp_> I thought you said you updated your ffmpeg? [13:10] <blippyp_> this is old - it's probably your problem [13:11] <Jeroi> hello [13:11] <blippyp_> it's not liking your format settings (probably for the audio) - you need to specify the bit rate or something [13:11] <blippyp_> hi Jeroi [13:11] <Jeroi> Is there any support how to setup windows minge Qt Creator to compile ffmpeg libs? [13:12] <Jeroi> I get punch of errors [13:12] <blippyp_> not a clue :( [13:12] <Jeroi> Installed latest Qt5 [13:12] <Jeroi> opened one qtwrapper code for ffmpeg [13:12] <Jeroi> and trying to compile the example [13:12] <Jeroi> with mingw as my compiler [13:13] <Jeroi> D:\mingw\include\stdlib.h:362: error: 'size_t' was not declared in this scope [13:13] <Jeroi> _CRTIMP void* __cdecl __MINGW_NOTHROW calloc (size_t, size_t) __MINGW_ATTRIB_MALLOC; [13:13] <Jeroi> ^ [13:13] <Jeroi> similar errors like that [13:14] <blippyp_> it's not understanding what size_t is... [13:15] <blippyp_> maybe look in another source file and find out where it's being declared, then re-declare it again in that file [13:15] <Jeroi> ok I removed my mingw path and now it seems it works but atm it dont find ffmpeg libs so some path names do not work in pro file I guess [13:15] <blippyp_> not a clue - never used mingw :( [13:17] <Jeroi> ok need to donwload ffmpeg libs [13:17] <Jeroi> the folder does not exist any lib files [13:17] <blippyp_> that would probably be a good start... ;) [13:17] <Jeroi> for the wrapper code [13:22] <Jeroi> so I need only the lib folders? [13:22] <Jeroi> libavutil [13:22] <Jeroi> libavformat [13:22] <Jeroi> and so one to include to my project to open video files? [13:44] <galex-713> Hello again [13:45] <galex-713> With -t/-ss options in ffmpeg, man page says it can support either hh:mm:ss[.xxx] or seconds [13:45] <galex-713> Can I use ss[.xxx]? Like a number of seconds greater than 60 [13:46] <galex-713> The time in seconds plus the remaining ms [13:53] <blippyp_> galex-713: no I don't think so - if you want to use ms you have to use the hh:mm:ss[.xxx] format [13:54] <blippyp_> I might be wrong, but I'm pretty sure I've tried this and had to use the longer format (so I just stick with it now) [13:59] <galex-713> argh [17:46] <xreal> What does this answer by cehoyos mean? "Do you believe this is a regression?" [17:46] <xreal> eeh question [17:47] <xreal> I think, nobody understood my bugreport. [17:47] <Bombo> xreal: i think that means a bug in the new version, that was not there in the version before [17:48] <xreal> Bombo: So I need to test older versions, if the bug appears, too? [17:48] <Bombo> xreal: think so [17:48] <xreal> Bombo: ack. I hope, I can find any. [17:49] <Bombo> xreal: if you find out when the bug happened it's easier to find/fix [17:49] <Bombo> s/happened/appeared/ [17:50] <xreal> Bombo: Problem is, it might haven been built against another library version. So I can't test it correctly. [17:51] <xreal> I'll try 1.01, 1.2, 2.02, 2.1, 2.13 [17:51] <JEEB> xreal, don't be put off by cehoyos [17:51] <JEEB> just saying [17:51] <JEEB> I've had my herps and derps with him, too :P [17:52] <JEEB> generally #ffmpeg-devel has then helped if he hasn't gotten anywhere [17:52] <xreal> JEEB: I simply didn't understand has question because I'm no native speaker :) [17:52] <xreal> his* [17:52] <JEEB> there's plenty of static builds so if you can find any release binaries for your platform, those should be good [17:52] <JEEB> I think zeranoe made some? [17:53] <JEEB> did you provide a means to replicate the issue in your ticket btw? [17:53] <xreal> JEEB: My problem is concerning LAME. I think, the LAME lib is changing sometimes. [17:53] <xreal> JEEB: https://trac.ffmpeg.org/ticket/3576 [17:53] <xreal> brb testing [17:53] <JEEB> ok, so samples uploaded [17:53] <JEEB> yeah [17:54] <JEEB> you can reply "I am not sure, and I do not currently have the time to test" [17:54] <JEEB> if you can't test :) [17:54] <JEEB> cehoyos can then do that in case he wishes to [17:56] <xreal> JEEB: I've just completed my tests with 5 stable releases. [17:56] <xreal> 1 sec [17:56] <xreal> interesting [17:57] <xreal> Nope, the problem is always there. [17:57] <xreal> v1.0.1 also had another problem: it created invalid WAV files :) [17:58] <xreal> let me check older versions [18:00] <xreal> Damn, I'd die for a non-GPL version with AAC support on win32 :( [18:00] <xreal> I think, I'll pay someone on rent-a-coder, to write a howto. [18:03] <xreal> Oh, things get worse before 0.8 [18:03] <JEEB> I don't think it's necessary to go that far [18:03] <JEEB> a lot of stuff wasn't even ready at that point yet [18:04] <xreal> JEEB: So the correct answer to "Do you believe this is a regression?" is "No." ? [18:05] <JEEB> "I have tested X,Y,Z , and this does not seem to be a regression" [18:05] <JEEB> x,y,z being the versions [18:06] <xreal> thanks [18:08] <xreal> JEEB: Do you think, this will ever be added? https://trac.ffmpeg.org/ticket/3577#ticket [18:09] <JEEB> no idea, at least the defaults in lavf/lavc tend to try and not leave too much distringuishable information in the file [18:09] <JEEB> but that looks similar to what libx264 writes [18:09] <JEEB> so it could be an option there? [18:09] <JEEB> not sure [18:09] <JEEB> never used LAME API [18:12] <xreal> JEEB: My main problem is: when I use extern LAME.exe, I can't mux audio and video correctly because of the padding LAME does. [18:12] <xreal> So there's a gap at the begging, which makes everything out of sync. [18:12] <xreal> beginning* [18:13] <JEEB> btw, your problem is partially due to how ffmpeg deals with encoder delay. at least when it can read it [18:13] <JEEB> if you take a mp4 file with encoder delay set for the audio track properly and lavf/lavc actually taking it into mention [18:13] <JEEB> lavf will give you negative timestamps [18:13] <JEEB> and if you deal with them correctly, it works as you'd assume [18:13] <JEEB> BUT [18:13] <xreal> JEEB: my problem is: how can I determin the correct delay :) [18:13] <JEEB> if you output that mp4's audio into a wav [18:13] <xreal> using LAME only, it's broken in FFMPEG. [18:13] <JEEB> you will get the whole thing [18:14] <xreal> JEEB: you know what? [18:14] <xreal> Let me create a test file with tones and testing picture. [18:14] <JEEB> because ffmpeg doesn't currently handle audio encoder delay in any way or form if your output format doesn't support some way of noting it (like wav) [18:15] <JEEB> I've talked about it before on the -devel channels but there seems to be no concensus right now [18:15] <xreal> JEEB: it handles it for MP3. But for the beginning only. [18:15] <JEEB> xreal, are you actually reading what I'm saying? [18:15] <xreal> JEEB: Seems like I don't understand itcorreclt.y [18:16] <xreal> let me re-read it [18:16] <JEEB> I'm telling you that this problem partially at least (if not most probably fully) is related to the fact that ffmpeg can't deal with encoder delay [18:16] <JEEB> that is, if the encoder delay is coded somewhere [18:16] <JEEB> like in my example with the mp4 file where it is actually written down [18:16] <JEEB> libavformat and libavcodec will happily read that information, and use it [18:16] <JEEB> and you will get audio packets with negative timestamps [18:17] <JEEB> of which you of course generally cut off the negative timestamped ones [18:17] <JEEB> now what does ffmpeg do? [18:17] <JEEB> (in the case of ffmpeg -i hurrdurr.mp4 out.wav) [18:17] <JEEB> it sees the negative timestamps [18:17] <JEEB> and it sees the fact that wav can't handle negative timestamps [18:18] <JEEB> thus it moves the starting point to the first negative timestamps [18:18] <JEEB> and thus you get everything including the encoder delay in the wav [18:18] <xreal> JEEB: Can't I tell it: "skip negativ timestamps" ? [18:18] <JEEB> not that I know [18:18] <JEEB> also I'm not sure if the timestamps can be read in your case, I'm just saying that if the timestamps will be read, you will still have that problem [18:19] <JEEB> unless that gets fixed at some point [18:19] <xreal> JEEB: From my POV, that's a bug of FFMPEG :) [18:19] <JEEB> of course [18:19] <JEEB> it is [18:19] <xreal> kewl, I've found a real bug [18:19] <JEEB> they just haven't decided how to deal with it yet [18:20] <JEEB> and it's just kind of funny that if you use the libraries, and deal with the values given by it accordingly, you will get the correct result :) [18:20] <JEEB> so the internals actually can handle it [18:20] <JEEB> but not ffmpeg the command line app [18:20] <JEEB> but yeah, in your case there's two layers [18:21] <JEEB> 1) is the encoder delay written somewhere, and if it is - is it read? 2) when we get the encoder delay from the API, we have to now deal with it somehow in ffmpeg [18:22] <xreal> JEEB: For MP3s encoded by LAME, it's easy and well documentated. But other MP3s could make trouble. [18:22] <JEEB> yeah [18:22] <JEEB> it's the same for AAC [18:22] <JEEB> unless you have the encoder delay written somewhere [18:22] <JEEB> you just can't make al ot of assumptions [18:23] <JEEB> I thnk there's a code path for faac [18:23] <xreal> JEEB: Actually, this might explain som things. I've decoded the AAC stream of a MP4 into WAV and there was a delay. [18:23] <JEEB> and then that's it [18:23] <xreal> JEEB: Does FFMPEG output the audio delay somehow? [18:23] <JEEB> (faac leaves a "signature" and its encoder delay is known) [18:23] <JEEB> xreal, as in, you want to check if it gets read? [18:23] <JEEB> from the input [18:23] <JEEB> not sure [18:24] <xreal> JEEB: and the other way, when exporting to WAV. [18:24] <xreal> "warning: 25 ms delay" or something [18:24] <JEEB> well, we already know that ffmpeg will just blurp every decoded packet as-is to the WAV :P [18:24] <JEEB> because wav doesn't support negative timestamps [18:24] <xreal> JEEB: Yes, but with a known delay, I just could cut it. [18:25] <xreal> JEEB: https://trac.ffmpeg.org/ticket/2325 [18:25] <xreal> :) [18:26] <JEEB> yes, there probably is a nice amount of issues regarding encoder delay [18:27] <xreal> JEEB: I don't even know, where to start when encoding a video in FFMPEG and audio in LAME cli ::( [18:29] <JEEB> well, if you were the guy encoding stuff into AVI I think these are not exactly the biggest issues you'd be having [18:30] <xreal> JEEB: I'd use AAC, but I don't trust FAAC. [18:30] <JEEB> I was talking of a container, you brought up another audio format? [18:30] <JEEB> your problems don't exactly change per audio format :D [18:30] <xreal> Argh, let me expand my answer: [18:31] <xreal> I'd use MP4 or MKV, but since my player doesn't support this, I'd need a new one. Since this would support AAC playback then, I wouldn't need MP3 anymore. [18:32] <xreal> JEEB: Actually, I don't need MP3 to WAV using ffmpeg, but I just wondered, why I always get out of sync. [18:32] <JEEB> anyways, just saying that the encoder delay in mp3 files is probably the smallest of delays you might have due to the ways things can work in AVI :P [18:32] <xreal> JEEB: So switching to MP4/MKV would fix anything? [18:32] <JEEB> not everything [18:33] <JEEB> but at least you wouldn't get extra lulziness [18:33] <JEEB> such as players handling b-frame delay and various hacks related to it differently [18:33] <JEEB> and/or a player not liking your AVI output for whatever reason [18:36] <xreal> JEEB: I'm confused now. [18:39] <xreal> Unknown encoder 'libfdk_aac' ... d'oh [18:41] <xreal> What's the GPL correct codec name [18:41] <xreal> ? [18:41] <JEEB> the best one in distributable builds, you mean? [18:41] <JEEB> -c:a aac -strict experimental [18:41] <JEEB> anyways, I'm off for a while [18:41] <xreal> JEEB: Thanks for help. [18:42] <xreal> rror: 2pass curve failed to converge ... narf [18:54] <Jeroi> hello, can ffmpeg libraries open avc file? [18:54] <Jeroi> I found one program that uses qt and ffmpeg to make meadia player, but the media player is unable to open avs files [18:54] <Jeroi> I mean Avisynth files [18:55] <Jeroi> I did read that ffmpeg has some kind of avisynth support directly to encode avs files [18:55] <Jeroi> this should mean that ffmpeg lips are able to read avisynth files directly? [19:54] <xreal> So, I should use CRF when using ffmpegs x264 encoder? [19:55] <klaxa> see https://trac.ffmpeg.org/wiki/x264EncodingGuide [19:55] <klaxa> in most cases yes [19:56] <xreal> klaxa: isn't there a two pass crf? :) [19:57] <klaxa> have you read the page? [19:58] <xreal> klaxa: yep. I thought, Two-Pass doesn't use crf, but some kind of cbr/vbr [19:58] <xreal> Sorry, I stopped ripping DVDs in 2006 and there was AutoGK with XviD :) [19:59] <c_14> https://trac.ffmpeg.org/wiki/x264EncodingGuide#twopass [19:59] <c_14> >This method is generally used if you are targeting a specific output file size and output quality from frame to frame is of less importance. [20:00] <xreal> c_14: Yeah, I understand that. But I thought, there would be a second pass for CRF, too to increase the quality :) [20:01] <c_14> Doing more than one pass doesn't increase the quality. [20:01] <c_14> It just let's you determine the output filesize. [20:02] <c_14> -' [21:37] <brontosaurusrex> xreal, you should use crf (if you are not targeting specific size or average bitrate) and its single pass [21:37] <xreal> brontosaurusrex: thanks [22:43] <JDuke256> hello , whats h263 requires on compression ? yuv444 or yuv422 ? [22:43] <JDuke256> or any other yuv type ? [22:46] <klaxa> >Supported pixel formats: yuv420p [22:46] <klaxa> you can find that out if you run: ffmpeg -help encoder=h263 [22:47] <JDuke256> i m writing code on C++ [22:47] <JDuke256> my problem is i dont know which format i need to use on Directshow to send h263 codec [22:48] <JDuke256> i got MEDIASUBTYPE_YUY2 [22:54] <JEEB> directshow yuv420p is I420 [22:55] <JEEB> YV12 is the same, but with chroma channels swapped, so you'd have to give the pointers in another arrangement [23:56] <xreal> JEEB: Is there an lossless audioformat, which supports delay? FLAC perhaps? [23:56] <xreal> or OGG lossless? [00:00] --- Tue Apr 22 2014

1 0

ffmpeg-devel.log.20140420
by burek 21 Apr '14

21 Apr '14

[00:24] <ubitux> BBB: still the loop filter [00:24] <ubitux> well, i'm trying [00:24] <ubitux> and probably doing it wrong :) [00:55] <ubitux> ok so, now how am i suppose to repack that. [00:56] <ubitux> packuswb(x,x) isn't doing what i want to [01:18] <cone-325> ffmpeg.git 03Peter Ross 07master:f57ac37228b2: avformat/iff: extend IFF demuxer to decode DSDIFF 64-bit chunks [02:18] <michaelni> Daemon404, do you have a testcase for "mjpeg: Do not fail jpeg decoding on bad EXIF data." ? [02:19] <michaelni> iam curious about the bad exif data ... [02:22] <Daemon404> hmm [02:23] <Daemon404> pm'd you a link [02:37] <BBB> ubitux: well the inverse of punpcklbw is packuswb [02:37] <BBB> ubitux: so if you want to use packuswb, use punpcklbw [02:37] <BBB> ubitux: I'm not saying the intermediate packing order makes sense, but it works as intended [04:23] <michaelni> Daemon404, posted a patch that avoids the exif failure [06:09] <cone-986> ffmpeg.git 03Lukasz Marek 07master:5053897b6ab2: lavd/xv: keep aspect ratio [06:09] <cone-986> ffmpeg.git 03Lukasz Marek 07master:9fcdfac894b2: lavd/xv: add window id param [06:09] <cone-986> ffmpeg.git 03Lukasz Marek 07master:de705e52d47f: lavd/xv: implement repaint message [06:09] <cone-986> ffmpeg.git 03Michael Niedermayer 07master:cd4faed89378: Merge remote-tracking branch 'lukaszmluki/master' [11:47] <ubitux> BBB: i'm trying to do packuswb(x, vextracti128(x, 1)) in one step [11:47] <ubitux> i was assuming packuswb(x, x) would do, but it seems not [12:28] <ubitux> ok so it went freaking ugly but i got the lpf working with avx2 [12:40] <ubitux> http://pastie.org/pastes/9095042/text so it's a bit better [12:41] <ubitux> i can probably do much better in various places [13:05] <BBB> lol [13:05] <BBB> ubitux: how does it do speed-wise? [13:05] <BBB> oh those are metrics [13:05] <BBB> n/m [13:05] <BBB> I thought they were a patch [13:06] <BBB> hm 8 cycles faster is a little ... little? [13:06] <BBB> I mean avx2 should have double throughput so you should be able to get double the speed right? [13:06] <ubitux> ignore the first one [13:07] <BBB> you did implement a mix4 function right? [13:07] <BBB> that is a little tricky :) [13:07] <ubitux> no i didn't really [13:07] <ubitux> actually i should remove most of the avx2 func you see there [13:07] <ubitux> it's only relevant with a few cases [13:08] <ubitux> aka where filter6() and filter14() is called [13:08] <ubitux> but i'm still wondering about all the threshold comparison [13:08] <ubitux> i can probably reduce the number of compare call, but with the extract overhead [13:10] <ubitux> BBB: i basically just made filter6() and filter14() exploit ymm regs [13:23] <BBB> oh I see [13:23] <BBB> ok, so here's the plan I had: [13:23] <BBB> forget the dsp functions for a second [13:23] <BBB> go to vp9.c [13:24] <BBB> you know the basic loop filter alto in vp9 (libvpx) is 8-pixel based right? [13:25] <BBB> line 3223 [13:26] <BBB> we're doing edges between blocks horizontally adjacent to each other [13:27] <BBB> e.g. a | b [13:27] <BBB> where | is the edge [13:27] <BBB> if you look carefully, you'll see we do lflvl->mask[][][y] as well as [y+1], i.e. we do 2 lines per iteration [13:28] <BBB> reason we do that is to do 16 pixels per iteration, instead of the basic 8 that vp9 gives us [13:28] <BBB> we can extend this further, obviously, we can do 32 - you need to do some more of the special handling that you see below that to take care of the cases where y and y+1 don't use the same loop filter (e.g. mixing wd=4 and wd=8, or being on an edge) [13:28] <BBB> for 32, that's somewhat harder, because you can mix 16 and 4 and 8, or 16 and empty and 8 [13:29] <BBB> so mix4 is a little bit more complex than mix2, but that gives you 32pixel basic loop filter entry points, that you can then exploit with avx2 [13:31] <BBB> so the cases you want to handle are probably 16|16, 16|n, n|16 and n|n where n is any of 8|8, 8|4, 4|8, 4|4, 4|none, none|4, 8|none, none|8 [13:32] <BBB> (as opposed to mix2, which is just 8|4, 4|8, 4|4 and 8|8) [13:34] <BBB> there's also avx2 mc functions to be created btw, if you're in for more fun [13:35] <BBB> I guess we can start with the filter4/6, it does actually help a fair bit, so that's certainly nice [13:35] <ubitux> yeah i see what you want to do [13:36] <ubitux> not sure i'll make much work on avx2, i really want to continue the arm thing actually :D [13:36] <ubitux> i just wanted to have a little overview of what we can do with avx2 [13:36] <ubitux> and how that works [13:37] <BBB> oh right arm [13:37] <BBB> how far is that? [13:37] <ubitux> not much from previously [13:37] <ubitux> got busy with random things lately [13:37] <ubitux> not much more* from previously [17:21] <cone-68> ffmpeg.git 03Michael Niedermayer 07master:a94de50ba02f: avcodec/exif/exif_add_metadata: add support for SSHORT & SBYTE [17:21] <cone-68> ffmpeg.git 03Michael Niedermayer 07master:e70b9b32d5ba: avcodec/exif: do not follow 0 offsets [18:25] <cone-68> ffmpeg.git 03Michael Niedermayer 07master:69bbe27b45aa: avcodec/huffman: use av_malloc_array() [18:25] <cone-68> ffmpeg.git 03Michael Niedermayer 07master:1fd5c7f1ee2f: avcodec/ratecontrol: use av_malloc_array() [18:25] <cone-68> ffmpeg.git 03Michael Niedermayer 07master:8c88ea76df9b: avcodec/tiff: use av_malloc(z)_array() [18:25] <cone-68> ffmpeg.git 03Michael Niedermayer 07master:de9cd5884822: avcdoec/webp: use av_malloc_array() [20:35] <cone-68> ffmpeg.git 03Christophe Gisquet 07master:319235c67c59: vc1dsp: introduce cases for 8x8 and 16x16 [20:35] <cone-68> ffmpeg.git 03Michael Niedermayer 07master:af89a685c46b: avcodec/arm/vc1dsp_init_neon: fix code so it compiles and passes fate-vc1 [21:07] <cone-68> ffmpeg.git 03Lukasz Marek 07master:4d09bc98974d: lavf/pcm: remove redundant check [21:29] <cone-68> ffmpeg.git 03Clément BSsch 07master:f0d368d75819: avcodec/x86/vp9lpf: merge a few movs with other instructions. [21:34] <cone-68> ffmpeg.git 03Clément BSsch 07master:62d31307c1c5: avcodec/x86/vp9lpf: add a comment above a bunch of SWAP. [00:00] --- Mon Apr 21 2014

1 0

ffmpeg.log.20140420
by burek 21 Apr '14

21 Apr '14

[01:13] <xreal> Is there an easy way to build ffmpeg on Windows with libfdk_aac ? [01:16] <klaxa> depends on how you define easy [01:17] <xreal> klaxa: A nice "how to" without too much knowledge... I'm a linux guy :) [01:18] <klaxa> you will need to compile libfdk-aac for windows and then compile ffmpeg and link it against your libdfk-aac libraries you built [01:19] <xreal> klaxa: I think, I'll install vbox with debian for this... sounds more easy. [01:19] <xreal> And another question: I want to transcode some FLV files to MP4. As I've read, it's best to use this "-vcodec copy -acodec copy" or that "-codec: copy" to create an intermidiate FLV file first and use MP4Box to create the final MP4 (for compatibility reason). Is this still "up to date"? [01:19] <xreal> The FLV contain MP4 and AAC streams. [01:19] <klaxa> don't know about compatibility, try to use the mp4 muxer from ffmpeg first maybe? [01:20] <klaxa> also if you are copying the codecs it's more remuxing and less transcoding [01:20] <xreal> klaxa: exactly, sorry. How can I use ffmpeg's muxer? [01:20] <klaxa> ffmpeg -i input.flv -c copy output.mp4 [01:20] <klaxa> as easy as that [01:21] <xreal> klaxa: 1 sec :) [01:23] <xreal> klaxa: Works, but ffmpeg is used to playback. So this isn't a real test :) [01:23] <xreal> klaxa: MP4Box support things like qt-faststart. [01:24] <klaxa> so does ffmpeg [01:25] <xreal> klaxa: I think, most of the documentations are reports on the web are quite outdated :( [01:26] <xreal> klaxa: can I make ffmpeg show details of my mp4 ? [01:26] <klaxa> try ffprobe for that [01:27] <tachoknight> in case anyone comes across this in a search, this is how I created a 20 second clip from a jpeg and added a in and out fade: [01:27] <tachoknight> ffmpeg -loop 1 -r 23.976 -i test.jpg -t 00:00:20 -vf 'fade=t=in:st=5.5:d=1.5, fade=out:0:400' test.mp4 [01:27] <klaxa> also, see: http://www.ffmpeg.org/ffmpeg-formats.html#Options-3 [01:29] <xreal> klaxa: Here are the results. http://pastie.org/private/3yifgp7lf6wyadjtslmipg [01:29] <xreal> I wonder, why datarate has changed. [01:30] <klaxa> might be muxing overhead [01:30] <klaxa> usually bitrate is calculated by filesize/time [01:31] <klaxa> if the container has less overhead, it will come out as a lower number [01:31] <xreal> klaxa: Ah, ok. [01:32] <xreal> klaxa: What do you think about this one? -codec copy -map 0 -movflags +faststart [01:32] <klaxa> well try it out [01:32] <klaxa> and see if it does what you want it to do [01:33] <xreal> klaxa: works again... I think, I need to get another player. Mine used ffmpeg to plackback. [01:33] <xreal> uses* [01:33] <klaxa> what kind of player is that? [01:33] <klaxa> because most players somehow rely on libav* from ffmpeg [01:34] <klaxa> on linux at least [01:34] <klaxa> there are a lot that don't though [01:34] <xreal> klaxa: it's a codec collection for Windows. [01:34] <klaxa> maybe it's not most players after all [01:34] <xreal> MediaPlayer Home with a common codec collection. [01:35] <xreal> Hm... ffprob doesn't give pretty much details on mp4. [01:35] <klaxa> try mediainfo [01:36] <xreal> k [01:36] <klaxa> also what codec pack? CCCP? KCP? [01:36] <klaxa> what was the other good one? [01:37] <xreal> klaxa: 1 sec [01:37] <xreal> klaxa: K-Lite. [01:37] <xreal> mediainfo also doesn't give info about moov :( [02:12] <xreal> Is it possible to set details when creating MP4 files, like Title, Copyright etc. ? [02:16] <klaxa> xreal: https://ffmpeg.org/ffmpeg.html#Main-options see -metadata [02:16] <xreal> klaxa: Thanks again. [02:23] <xreal> klaxa: Hmm. Is this wrong? -metadata:s:a:1 handler_name="abc" [02:23] <klaxa> i don't know, i never use metadata [05:56] <gustavo_brazil> hey guys, how to record my skype talks with ffmpeg ? [08:00] <tm512> so is x264 lossless actually lossless or not? someone in here was telling me that it isn't, but other places are saying otherwise [08:01] <tm512> x264 -qp 0 -preset ultrafast gives much better compression than huffyuv [08:02] <tm512> 89809 kb/s yuv444p vs 210999 kb/s yuv422p [08:15] <blippyp_> tm512: was just looking into that again (not sure where I first read about it) - I may have informed you improperly - I think it's considered 'not lossless' because converting from rgb to yuv color space generates some lost data. However if you convert rgb to yuv444, then apparently there is no loss (ie - is in fact lossless). So keeping that in mind, you can use libx264 and still get some compression, while still keeping a lossless encodi [08:17] <tm512> there's still loss in color from going from rgb to yuv, it affects things like console text colors [08:18] <tm512> unless mplayer is further downsampling to 420 [08:18] <blippyp_> that would be my guess: here's the reference to what I was explaining: https://trac.ffmpeg.org/wiki/x264EncodingGuide [08:19] <blippyp_> look at the bottom of the page: Why doesn't my lossless output look lossless? [08:22] <tm512> it would be lossless if the source was yuv [08:22] <tm512> but after i have it in yuv, i could infinitely reencode without loss, right? [08:23] <tm512> since all the colors are already in yuv space [08:25] <blippyp_> it's IS lossless when converting from rgb to yuv444 - the colors may be 'off' from the conversion, but there is no drop in quality - do a test - convert from rgv to yuv444, then convert that file back into rgb again - do you colors go back to normal? I'm betting they do and that what you're noticing is a side effect from the conversion. This is what I was trying to explain to you earlier - About the conversion messing up the colors - That [08:29] <tm512> well no after the colors have been converted (rounded, truncated) or whatever to yuv the original rgb data is lost forever right [08:29] <tm512> I mean it's converted to rgb in my media player [08:30] <blippyp_> ignore what you're 'seeing' - your player might be fooling you - do a quick test like I said... [08:30] <blippyp_> if the video looks normal to you have converting back from yuv to rgb again, then you KNOW that the data is still there [08:31] <blippyp_> it's just being displayed differently - to correct it you would likely need to run the video through a hue filter to change the hsv values (to whatever you need) to get it back to what it looked like when it was rgb [08:32] <blippyp_> this would be a very minor change (like lowering saturation by one or brightness by one, maybe both) [08:33] <blippyp_> as an example of what I'm talking about - check out this page: http://www.infognition.com/VirtualDubFilters/color_correction.html [08:33] <blippyp_> I still haven't found one for rgb to yuv444 yet though [08:34] <tm512> anyway using 444 should give me more color information to work with to change around with filters, so that's nice [08:39] <blippyp_> tm512: You'll probably like this page (it seems to have some answers for you) http://ffmpeg-users.933282.n4.nabble.com/RGB-YUV-color-shift-td4654294.html [10:04] <sor_> is there a way to record the full desktop but in a smaller screen insteed of full size? -- i know i scan scale afte just want to know if i can do it during -- thanks [10:18] <blippyp_> sor: -s 640x480 (put this just before or after you declare your codecs for saving and it will resize the video) - not really sure what you mean though or if this answers your question - if not, please be more specific... [10:21] <sor_> thanks will try [10:21] <tm512> doing the colormatrix filter shouldn't mess things up when youtube processes the video right? [10:23] <sor_> blippyp_, it's square and -s 640:-1 doesn't work [10:23] <blippyp_> tm512: yeah, I thought that would be what you'd do - no, I would expect youtube to leave it be - as long as you stay within their limits (like the suggested encoding page I sent you to earlier today) then your video should stay basically the same when tyey're done re-encoding it they way the want/different formats etc.... Although you might see a difference between one format compared to another - not sure, I haven't tested that - but I wo [10:23] <blippyp_> sor: please be more specific, what do you mean it's square? [10:24] <blippyp_> sor: also, not sure if -1 works with the -s switch [10:24] <sor_> blippyp_, i have 1280x800 so 640x480 doesn't show full screen [10:25] <sor_> it should be 640:-1 but that doesn't work unless i recode it after [10:26] <blippyp_> sor: k - I'm still not understanding your problem I think - I assume you've used -vf "scale=640:-1" ? [10:26] <blippyp_> and this isn't doing what you want? [10:27] <sor_> blippyp_, crap forgot th = --- thanks [10:27] <blippyp_> cool [10:27] <sor_> :) [10:31] <tm512> blippyp_: well, interestingly, without the filter colors overall are more accurate [10:31] <blippyp_> maybe you need to 'tweak' the filter more/less? [10:31] <blippyp_> are you just copy/pasting it or did you put some thought into what it's doing? [10:32] <tm512> it's really really close if I don't use the filter [10:32] <tm512> some colors seem to get messed up no matter what [10:32] <blippyp_> to be honest, I would leave it be - smarter minds than us have set these things as a default for reasons unknown to the masses.... ;) [10:33] <tm512> overall I'm happy with the quality of this though [10:33] <blippyp_> I understand it's frustrating that it's not 'identical' to your source, but no one will notice - my opinion [10:34] <blippyp_> i can certainly relate (not about the color variance) but when I first started uploading to youtube it frustrated me to NO END that the resulting video on youtube looked like horrid shit compared to my beautiful encoding I did... Took a lot of research and time to finally accept that they were just destroying it and there was little to nothing I could do about it.... [10:39] <jonascj> Hi all. I run a ffserver with config http://pastebin.com/gyzWkNnL . On a feed machine I run: "ffmpeg -v verbose -r 10 -s 320x240 -f video4linux2 -i /dev/video0 http://host.com:8090/webcam.ffm" but receive a pixel format error: http://pastebin.com/0DjAkAr6 [10:39] <jonascj> I have been asking about this in here previously but I've failed to correct the error. [10:40] <jonascj> the command works perfectly if I substitute "http://host.com:8090/webcam.ffm" with "file.mp4". So I assume my ffmpeg is working. [10:41] <jonascj> Someone in here told me to put PixelFormat in my ffserver.conf file, but ffserver says "incorrect keyword" when I place "PixelFormat ..." in the stream section of the config file. [10:42] <tm512> blippyp_: thanks for the help though. I'm gonna get to sleep [10:42] <tm512> good night [10:43] <blippyp_> no problem tm512 [10:43] <blippyp_> night [10:45] <jonascj> Together with another helpful person in here I thought that maybe video4linux2 library support wasn't there, but compiling with --enable-libv4l2 does not make a difference. With out without that directive the command I've justed posted works with a file output and fails with ffserver feed/stream output. [10:46] <blippyp_> jonascj: what pixel format are you trying to switch it to? [10:47] <jonascj> blippyp_: to tell you the truth I didn't think much about pixel formats when I first started out trying to get this working. So I am not trying anything specific besides getting some images through to my ffserver. I will convert it to what ever that requires :) [10:48] <blippyp_> I haven't used it before (and so far am not finding much documentation) [10:48] <jonascj> last time I discussed this with someone in here (a few days ago) I did not fully understand who (ffserver or ffmpeg) complains about this. [10:48] <jonascj> blippyp_: what haven't you used before? ffserver with ffmpeg or yuvxx pixel formats? [10:49] <blippyp_> try setting it to yuv420p [10:49] <blippyp_> I've used ffserver, but it was ages ago - I never had to switch the pixel format with it... [10:49] <blippyp_> I did very limited testing - got it work and moved onto other things [10:50] <blippyp_> if yuv420p works, but you're looking for better quality, there are other options (give your settings I'm assuming you not caring about that issue too much) [10:51] <blippyp_> god I'm typing like a moron today [10:53] <blippyp_> is there a reason why you're using mjpeg? [10:55] <blippyp_> yuv might not work with that - maybe try a pixel format of rgb8 (256 colors) or rgb24(like 60,000 colors I think) [10:57] <jonascj> blippyp_: I am very novice ot htis. I just saw a collegue of mine monitoring our 3d printer with a webcam hooked up to a windows pc with some windows webcam software. To view the webcam feed he connected to the computer with the webcam using Teamviewer. That just made me want to look into streaming. [10:58] <jonascj> and everyone seemed to be using ffmpeg and ffserver to get webcam feeds of off raspberrypi's and similar. [10:58] <blippyp_> you're on the right track, ffserver will do that for you [10:58] <blippyp_> yeah, it's common - ffmpeg/utils are really powerful and under-appreciated by so many people [10:58] <jonascj> mjpeg was just what a lot of people used because it has native support in many browsers. So it can be directly embedded as an image in a webpage etc. [10:59] <blippyp_> ah... [10:59] <jonascj> blippyp_: if I wanted the feed in a mediaplayer I suppose I could use something else. [10:59] <blippyp_> it looks like yuv420p should work with mjpeg - I'd go with that one [11:01] <jonascj> no matter what I specify using "-pix_fmt x" with my ffmpeg command (on the computer running the webcam) I get the same error back. That yuyv422 is not supported. [11:01] <blippyp_> you mean -pix_fmts? [11:01] <blippyp_> with an s [11:02] <jonascj> blippyp_: but I am confused as to whether it should be an ffmpeg or ffserver setting. The manual kind of suggest that the settings of the ffserver is "pushed" to the ffmpeg when ffmpeg connects to ffserver. [11:03] <jonascj> https://ffmpeg.org/ffmpeg.html this say thta -pix_fmts is to list available pixel formats and that -pix_fmt is to specify pixel format... [11:03] <blippyp_> good enough [11:03] <blippyp_> couldn't remember [11:03] <jonascj> :) [11:03] <blippyp_> did you change it to yuv420p? [11:03] <blippyp_> it's complaining about yuv422? [11:04] <blippyp_> if that's the case change it to the one it's complaing about instead - see what that does [11:04] <jonascj> "ffmpeg -pix_fmt yuv420p -f video4linux2 -i /dev/video0 http://wuhtzu.dk:8090/webcam.ffm" this still says "[mjpeg @ 0x2c57260] Specified pixel format yuyv422 is invalid or not supported". so maybe I fail to specify it properly [11:05] <blippyp_> use ffmpeg -pix_fmt yuyv422 -f ..... [11:06] <blippyp_> are you using windows/linux/osx? [11:06] <jonascj> ffserver and ffmpeg on linux [11:06] <blippyp_> duh - v4l2 - should have known [11:07] <jonascj> :) [11:07] <blippyp_> did that fix it? [11:07] <jonascj> what bothers me is that "ffmpeg -pix_fmt yuv420p -f video4linux2 -i /dev/video0 /tmp/lol.mp4" works just fine (looks like crap, but I get some video recorded) [11:08] <jonascj> nope, I cannot get it to reply anything else than "[mjpeg @ 0x2c57260] Specified pixel format yuyv422 is invalid or not supported" [11:08] <blippyp_> run ffmpeg -pix_fmts|grep yuyv422 [11:08] <blippyp_> did it display? [11:09] <blippyp_> should say IO... yuyv422 3 16 [11:09] <jonascj> yup, I get "IO... yuyv422 3 16" [11:09] <jonascj> :P [11:09] <blippyp_> hmmm [11:09] <blippyp_> what is your pixelformat setting in the ffm file? [11:10] <blippyp_> I never used the ffm files [11:10] <blippyp_> so I'm searching in the dark with you on that issue [11:10] <jonascj> blippyp_: http://pastebin.com/gyzWkNnL is my config. If I put "PixelFormat" into the stream seciton (where the manual says it should be) I get "incorrect keyword PixelFormat" from my ffserver [11:10] <blippyp_> I looked up PixelFormat, it is a setting [11:11] <blippyp_> you must be using it wrong - I haven't been able to find a good example of that yet.... [11:11] <jonascj> the <Feed> and <Stream> setting is just taken from the sample config file [11:11] <jonascj> but yeah, maybe I should scrap my ffserver.conf file and start over following another example [11:12] <jonascj> or not mixing the sample config file with suggestions from random guides [11:12] <blippyp_> I think it has something to do with mjpeg [11:12] <blippyp_> I've never worked with that format before [11:12] <c_14> According to ffmpeg -h encoder=mjpeg: Supported pixel formats: yuvj420p yuvj422p yuvj444p [11:12] <blippyp_> from your previous posts - I'd assume that it's using yuyv422, but clearly that isn't the case - not sure why it's mentioning that at all - unless I'm missing something [11:13] <blippyp_> k - use one of those with your ffmpeg command instead [11:13] <blippyp_> your'e not specifically altering the pixelformat in your ffm file, so it should stick with something standard [11:14] <blippyp_> with only three options, it should be easy to determine which one it's defaulting to [11:14] <jonascj> ffmpeg -pix_fmt yuvj420p -f video4linux2 -i /dev/video0 http://wuhtzu.dk:8090/webcam.ffm [11:15] <jonascj> that gives the same error about yuyv442 [11:15] <jonascj> *yuyv422 [11:16] <blippyp_> maybe the issue with with v4l2 [11:16] <jonascj> but is it a setting missing from my ffserver.conf file then or is it on the ffmpeg side? ffmpeg says the input stream (from /dev/video0) is "Stream #0:0: Video: rawvideo (YUY2 / 0x32595559), yuyv422, 320x240," [11:16] <blippyp_> maybe it's converting it [11:16] <c_14> What I'd also try is instead of using -pix_fmt as an input option, use it as an output option. [11:16] <blippyp_> what happens if you remove the -pix_fmt option altogether from your ffmpeg command? same error? [11:17] <jonascj> c_14: that is some basic stuff I do not know about. Does -pix_fmt mean different things depending on where you give the argument? [11:17] <jonascj> I would have assumed there was an "-input_pix_fmt" and "-output_pix_fmt", but that of couse is not the case. So then I assumed that "-pix_fmt" always meant one thing (output for example) [11:18] <blippyp_> I think he's saying to try this: ffmpeg -f v4l2 -i /dev/video- http://wh......... -pix_fmt yuyv422 [11:18] <jonascj> blippyp_: yeah, so the position (in the argument list) of arguments give them different meanings? [11:18] <c_14> Yes. [11:18] <blippyp_> yes - very much so with ffmpeg [11:18] <jonascj> blippyp_: without pix_fmt altogether gives the same error [11:19] <blippyp_> I think he might be right, I'm thinking it will work for you... [11:20] <c_14> From what I can see from the ffmpeg output you posted recently, ffmpeg is detecting that your webcam uses yuyv422 and then uses that as an output pix_fmt, but mjpeg doesn't like yuyv422. So I'd try using -pix_fmt yujv422p as an output option. [11:22] <blippyp_> yeah, if it doesn't work - do what c_14 just said... yujv422p should be compatible (I don't know where it's getting the yuyv422 - probably from v4l2, but I don't have a clue) [11:22] <jonascj> "ffmpeg -f video4linux2 -i /dev/video0 http://wuhtzu.dk:8090/webcam.ffm -pix_fmt yuvj422p" still gives this "yuyv422 not supported" error. [11:23] <jonascj> c_14: I think you are right. the ffmpeg output clealy states that the webcam input is yuyv422 and if nothing is done that it probably also the output format [11:23] <c_14> Can you give us the output, maybe with -loglevel debug? I want to know why ffmpeg is choosing yuyv422 instead of what -pix_fmt is specifying. [11:23] <blippyp_> yeah, I'm curious myself [11:23] <blippyp_> glad you knew how to get that [11:25] <jonascj> with debug level:http://pastebin.com/Q2R4QstR [11:25] <blippyp_> I can see it in his original output [11:25] <blippyp_> it is coming from v4l2 I think [11:27] <jonascj> this worries me a bit: "Incompatible pixel format '(null)' for codec 'mjpeg', auto-selecting format '(null)'" [11:28] <blippyp_> maybe try ffmpeg -f v4l2 -pix_fmt yuvj422p -i ...... [11:28] <jonascj> do you guys know why it works when I specify another output than my ffserver. E.g. if I just output to "file.mp4"? [11:28] <jonascj> is that becuase the fileoutput is happy with whatever pixelformat ffmpeg outputs whereas mjpeg (on the ffserver) is not happy about it? [11:29] <c_14> I'd think so. Have you tried writing to a local file forcing -c:v mjpeg -f mjpeg output ? [11:29] <blippyp_> because the mp4 format doesn't have a problem with the yuv420p format [11:29] <blippyp_> (I think) [11:31] <jonascj> this works fine: "ffmpeg -loglevel debug -f v4l2 -pix_fmt yuvj422p -i /dev/video0 /tmp/lol.mjpg -c:v mjpeg -f mjpeg" [11:31] <blippyp_> yeah - then converting the v4l2 feed is what needed to be done [11:31] <blippyp_> v4l2 was screwing it all up for you [11:32] <blippyp_> oh wait - you saved locally [11:32] <jonascj> but that local command still had "-f v4l2" (an input option?) [11:32] <blippyp_> now try the same command, but the way you were doing it before [11:32] <blippyp_> yes, but by putting the pix_fmt option directly after the v4l2 format you were converting the pixel format for v4l2 [11:33] <jonascj> "ffmpeg -loglevel debug -f v4l2 -pix_fmt yuvj422p -i /dev/video0 http://wuhtzu.dk:8090/webcam.ffm -c:v mjpeg -f mjpeg" this still fails with "yuyv422 not supported" [11:33] <jonascj> so maybe it is the ffserver after all. [11:34] <blippyp_> hmmm [11:34] <blippyp_> I think it is [11:35] <blippyp_> you've tried converting the pixel format in every way that I can think of locally - it must be on the server side [11:35] <jonascj> http://sirlagz.net/2013/01/07/how-to-stream-a-webcam-from-the-raspberry-pi-… what I'm trying to do seems to work out of the box for this guy [11:36] <jonascj> without having the word "pixel", "format", "pix" or "fmt" in either ffmpeg command or ffserver config [11:36] <jonascj> blippyp_: I'll try another ffserver. maybe try to install it locally and see if that helps [11:37] <blippyp_> the last two commands you tried pretty much ensures that there's something wrong with your ffm file from what I can tell... [11:37] <blippyp_> but like I said, I've never used them, so I can't really offer what's wrong with it specifically [11:38] <jonascj> ARGH, it works on my local ffserver [11:38] <jonascj> so maybe it is just the version of ffsserver on my remote machine [11:38] <blippyp_> this may be a stupid question - but is it firewalled without the port open? [11:39] <blippyp_> give me a second, I'm gonna try setting this up myself now (I'm getting as frustrated as you) ;) [11:40] <jonascj> blippyp_: It is iptable'd. I muts admit I haven't checked logs or iptables config because the ffserver (debug log output) says "New connection" [11:41] <jonascj> blippyp_: so I assumed that the connection was successful since the ffserver log showed "new connection". The debug output from ffmpeg also showed "webcam.ffm downloaded successfully". So I do not think it is a firewall problem [11:41] <blippyp_> I agree, that sounds like it's connecting [11:41] <c_14> What version of ffserver do you have on the remote? [11:43] <jonascj> "FFserver version SVN-r0.5.10-4:0.5.10-1" is what ffmpeg says when I give it no arguments. It was just what came with "apt-get install ffmpeg" on the remote ubuntu 12.04 machine [11:43] <Mavrik> wow. [11:44] <Mavrik> which dusty corner of a museum did you get that from [11:44] <c_14> You might want to download a newer version/compile from source& [11:44] <blippyp_> yeah - I agree [11:45] <jonascj> I will try that right away. Debian or Ubuntu also have this weird ffmpeg vs avconv thing going on [11:46] <jonascj> Also, sadly, both firefox and chrome just try to download my mjpeg stream/image. Maybe I should be looking for another good way to embed it [11:48] <blippyp_> jonascj: sorry, didn't notice the time - I gotta go hide easter-eggs for my daugher before she wakes up - I wish I could stick around more - hope you figure it out... [11:49] <jonascj> blippyp_: I think it was the version of ffmpeg on the remote machine. I feel quite confident that updating it will resolve my issue (since I've got it working, apart from some browser issues, locally) [11:49] <jonascj> Thank you for all your help! [11:50] <c_14> np [11:52] <jonascj> an .swf stream on my local machine confirmed that the principle setup is working [11:52] <jonascj> I just need to update my remote machine and figure out what the most friendly stream format is (mjpeg, swf etc.) [11:54] <jonascj> c_14: any recommendations there? I am looking to embed the streamoutput in something like a webpage (local system for monitoring some 3d printers). [13:10] <blippyp_> jonascj: you still around? [13:18] <c_14> He timed out at 12:12 [13:21] <blippyp_> c_14: yeah I noticed... should have been paying more attention... Just wanted to let him know it was working for me using mjpeg [13:21] <blippyp_> but I have the latest version [13:29] <makara> hi. Any idea how I can demux WMV on Linux? [13:30] <blippyp_> makara: be more specific on what you mean? [13:32] <makara> blippyp_, WMV seek is terrible, and the videos I. I want to convert to MKV without re-encoding [13:32] <makara> the videos I have are mono [13:32] <blippyp_> don't think you can do that [13:32] <blippyp_> you can do it with mp4, but I don't think you can do that with wmv [13:36] <makara> why not? [13:36] <blippyp_> different codecs [13:36] <blippyp_> I just tried it [13:36] <blippyp_> it pulled the audio, but no video [13:36] <blippyp_> hold on [13:37] <blippyp_> I might be wrong [13:37] <blippyp_> are you on linux? [13:37] <blippyp_> do you have mediainfo [13:38] <blippyp_> I just ran mediainfo on my wmv and it says it's in mpeg4 format, so that can be re-muxed with a bitstream [13:38] <blippyp_> I'm testing it now [13:39] <makara> linux yes [13:39] <blippyp_> get mediainfo and run that on one of your wmv files and see what it says for the codec [13:41] <blippyp_> no, I don't think you can, it doesn't seem to be working [13:42] <blippyp_> even if I try going to mp4 it doesn't work - maybe I'm doing something wrong [13:42] <makara> blippyp_, this works: avconv -i dom.wmv -acodec copy -vcodec copy dom.mkv [13:42] <blippyp_> really? [13:43] <blippyp_> that wasn't working for me... [13:43] <makara> yeah, I'm surprised [13:43] <blippyp_> good enough [13:43] <blippyp_> glad it's working for you :) [13:44] <makara> blippyp_, http://pastebin.com/PgGWWFUN [13:44] <makara> the output of `avprobe dom.wmv` [13:45] <blippyp_> maybe because you're using avcon? [13:45] <makara> avconv version 0.8.10-6:0.8.10-0ubuntu0.13.10.1, Copyright (c) 2000-2013 the Libav developers [13:45] <makara> and a fairly recent version [13:45] <blippyp_> yeah I'm using the latest ffmpeg [13:46] <makara> blippyp_, OS ? [13:46] <blippyp_> people think they're alike, but they aren't (not saying that's the issue in this case though - and I've never used avcon) [13:46] <blippyp_> arch [13:46] <blippyp_> I pulled my ffmpeg from github [13:46] <makara> well, doesn't get better than that [13:47] <blippyp_> yeah it does... :( [13:47] <blippyp_> I'm trying to re-compile it with frei0r enabled, but haven't had any luck so far... :( [13:47] <blippyp_> not sure why [13:48] <blippyp_> it's probably something obvious and I'm too stupid to figure it out... ;) [13:51] <makara> i swear by packet managers [13:51] <blippyp_> well, that's what I'm using - the aur [13:51] <makara> debian is the best thing since chocolate [13:51] <blippyp_> but i never find an ffmpeg with frei0r enabled... [13:52] <blippyp_> so I downloaded the source and compiled it all myself [13:52] <blippyp_> didn't get any errors during the compiling, but running it gives me segment faults [13:52] <blippyp_> I think I'm just having some weird conflic of some kind [13:53] <blippyp_> This is my output when I run ffempg: http://pastebin.com/3nDCiq3k [13:53] <blippyp_> My problem is where it says WARNING: library configuration mismatch... [13:54] <blippyp_> Basically, I have conflicting libraries in my system or something - Just not sure how to resolve it [13:55] <makara> blippyp_, why you on Arch? [13:55] <blippyp_> because I like it? [13:56] <blippyp_> everything is always up to date, my computers basically suck and can't be bothered with compiling everything myself [13:56] <blippyp_> arch keeps everthing nice and new for me [13:57] <blippyp_> this is the first and only time I've ever run into something that I couldn't get [13:57] <blippyp_> I was always running into problems with other installations because nothing was updated enough [13:57] <makara> oh [13:57] <blippyp_> I think the problem with this instance though is that my system is TOO updated??? [13:58] <makara> that's why they call it `unstable` [13:58] <blippyp_> not really sure, maybe I did something wrong - clearly I don't know... :( [13:58] <blippyp_> nah [13:58] <blippyp_> it's not unstable [13:58] <blippyp_> trust me - been using it for 4 years now [13:58] <blippyp_> it never breaks [13:58] <blippyp_> I just updated my system(s) for the first time in over a year - nothing wrong happened [13:59] <blippyp_> the key to using arch is to check the archlinux.org site before updating - ALWAYS [13:59] <blippyp_> if there are any known issues that can occur - they will be spelled out in black and white and how to fix them. Usually very easy, requiring a couple of steps. A monkey could do it [14:00] <blippyp_> And it's rare to need to do this [14:02] <makara> now I'm trying to turn this single channel into stereo without re-encoding [14:02] <blippyp_> I doubt you'll get away with that... ;) [14:02] <blippyp_> audio encodes fast though [14:02] <makara> i tried with the switch "-ac 2" but it re-encoded with libvorbis [14:02] <blippyp_> you got the video - that's easily the worst part done [14:03] <makara> http://trac.ffmpeg.org/wiki/AudioChannelManipulation [14:03] <makara> there's this line, but it's only for an audio file `ffmpeg -i input.mp3 -filter_complex "[0:a][0:a]amerge[aud]" -map "[aud]" output.m4a` [14:03] <makara> using the amerge filter [14:04] <blippyp_> if you have to use a filter, I'm guessing it would insist on re-encoding it [14:08] <blippyp_> gotta go makaa - my daughter will be up soon for easter and I need to take a shower (need to clean up for pictures) - she's only two, so it's kind of a big deal for her today... ;) [14:08] <blippyp_> good luck - but I'm guessing you'll need to re-encode the audio... [14:12] <makara> blippyp_, cheers b [14:12] <makara> happy easter [14:50] <xreal> Why can't I set tracknames seperatly for video and audio with ffmpeg? I tried anything :( [15:04] <makara> any way to create joint stereo aac? I'm converting from mono [15:04] <Mavrik> what you're asking makes no sense :) [15:04] <Mavrik> there's no joint stereo AAC [15:06] <makara> oh ok [15:07] <makara> Mavrik, its just that my converted video files are twice the size [15:07] <makara> when I convert audio to 64k the sound is bad [15:07] <Mavrik> well [15:07] <makara> so I do aac 128k [15:07] <Mavrik> are you setting quality properly :) [15:07] <Mavrik> also, quality depends on which AAC encoder you're using [15:07] <Mavrik> HE-AACv2 gives usually better results at 64k [15:08] <makara> avconv -i dom.wmv -c:a aac -strict experimental -b:a 128k -vcodec copy dom.mkv [15:09] <makara> Mavrik, i would have to compile avconv to get HE-AACv2 [15:09] <Mavrik> well [15:09] <Mavrik> avconv isn't ffmpeg. [15:09] <Mavrik> so *shrug* [15:12] Action: xreal goes crazy about metadata :( [15:13] <JEEB> makara, the binary distributable encoders can't do any sort of HE-AAC (IIRC), so you'd have to compile either libav or ffmpeg yourself with fraunhofer's fdk-aac yourself, yes [15:14] <xreal> Is there a good tutorial to convert DVD streams (with permission) to XviD using ffmpeg? I haven't done it for 6 years and used AutoGK for that before :D [15:15] <makara> JEEB, should I uninstall the debian version first? [15:15] <JEEB> makara, not necessary, as long as you have your own binary in front of the installed packaged one in PATH, you should be fine [15:15] <xreal> This doesn't work: "ffmpeg -i INPUT -metadata:s:a:1 title="test" OUTPUT" [15:42] <xreal> Anyone with an idea, why this doesn't work: "ffmpeg -i INPUT -metadata:s:a:1 title="test" OUTPUT" [15:44] <makara> JEEB, I cloned libav, configured it with `./configure --enable-libfdk_aac --enable-nonfree --enable-gpl`, make and make install [15:45] <makara> ah, its working [15:46] <makara> that wasn't that hard [15:47] <JEEB> -profile:a <the profile's name> should be the way to set the profile wanted, although I think fdk-aac by default sets some kind of HE profile in case of low enough bit rate [15:47] <JEEB> also, #libav for libav, #ffmpeg is ffmpeg's channel :) Some people here are rather touchy about the subject (I personally help wherever my knowledge can be of help). [15:53] <xreal> JEEB: But there's not #avconv :) [15:54] <JEEB> the channels are for the projects, not the apps :) [15:59] <makara> JEEB, i'm following http://trac.ffmpeg.org/wiki/AACEncodingGuide [16:00] <JEEB> yes, that generally matches [16:00] <makara> they suggest a bitrate switch as well as / as part of the profile switch [16:00] <JEEB> some encoders auto-switch depending on the bit rate given [16:00] <JEEB> but you can also set a profile manually [16:01] <makara> aac_he works, but I get an error with aac_he_v2 [16:01] <JEEB> a "profile not found" kind of thing, or something else? [16:01] <xreal> Anyone with an idea, why this doesn't work: "ffmpeg -i INPUT -metadata:s:a:1 title="test" OUTPUT" [16:02] <makara> oh, i see v2 doesn't handle mono [16:04] <makara> xreal, what about it doesn't work? [16:07] <xreal> makara: It doesn't name give name to audio/video track. [16:07] <xreal> I tried -metadata:s:a:1 | -metadata:s:a:0 | -metadata:s:v:1 |-metadata:s:v:0 [16:07] <xreal> makara: I need mp4box to fix it. [16:14] <makara> xreal, try s:a.1 [16:16] <xreal> dot? nice, 1 sec [16:18] <xreal> makara: Nope, doesn't work. Player still only gives the handler_name "SoundHandler". [16:18] <makara> xreal, then I think there's something wrong with your player [16:18] <makara> both avprobe and vlc see the track name [16:19] <xreal> makara: which avprobe settings do you use? [16:19] <makara> avprobe name-of-file [16:20] <xreal> I wonder, why it works with MP4Box. Need to compare the MP4s. [16:23] <xreal> makara: latest avprobe with "name-of-file" doesn't even return the handler. Let me post it. [16:25] <xreal> makara: http://pastie.org/9095452 [16:25] <xreal> better to read: http://pastie.org/pastes/9095452/text [16:40] <xreal> Ah, I think this is the problem: I'm using it with -c:v copy -c:a copy. Perhaps ffmpeg/avconv changes the metainfo to the source?! [17:02] <xreal> seems to be a bug [17:19] <xreal> I've used a debugger. ffmpeg *always* writes SoundHandler... you can't overwrite this. it's hardcoded! [18:22] <xreal> when using FFMPEG to encode XVID, is XVID lib really used or is it just a fourcc stored ? [18:22] <sacarasc> Depends on which codec you use. [18:22] <sacarasc> libxvid is xvid. [18:23] <sacarasc> Or was it xvidcore? [18:23] <sacarasc> There is an actual one, though. [18:24] <xreal> sacarasc: oh, don't know. "-c:v mpeg4 -vtag xvid" :) [18:24] <sacarasc> That is just setting the fourcc. [18:25] <sacarasc> The -vtag option is what does that. [18:25] <xreal> Uh, that's the example in ffmpeg.org's wiki to do Xvids :D [18:25] <xreal> sacarasc: Could you give me a hint, how to use the codec? [18:25] <sacarasc> XviD is just a MPEG4 ASP encoder. [18:25] <xreal> sacarasc: Ah, here: c:v libxvid [18:26] <xreal> sacarasc: I know, they are compatible. But doesn't XviD has some tweakings from the standard? [18:28] <xreal> my god, libxvid seems to be single core :D [18:28] <JEEB> welcome to early 2000s :P [18:28] <JEEB> and no, it doesn't have "tweakings from the standard" [18:29] <xreal> JEEB: why does it exist then? [18:29] <JEEB> anyways, I'm surprised you even chose to encode with xvid [18:29] <xreal> JEEB: old DVD player [18:29] <JEEB> xreal, because it was a fork of divx that was open source for a while [18:29] <JEEB> and it was the only mpeg-4 part 2 encoder for a while [18:29] <xreal> JEEB: I know, but it's like MP3: many encoders with different quality [18:29] <JEEB> they're generally on a similar'ish base [18:30] <JEEB> at least the lavc and xvid [18:30] <JEEB> it's just that xvid is regarded as simpler to use [18:32] <xreal> xvid has been updated some days ago to 1.3.3 ah multi-core has been added in 1.3.1 [20:24] <kittonian> hi all. i'm needing to convert a bunch of avi files to m4v and was wondering if there was any way to speed things up. i'm only getting about 30-40 fps because it's doing a full a/v re-encode for each file (taking forever). I tried just changing the container but of course that didn't work because the avi audio is mp3. [20:25] <kittonian> i'm using -c:v libx264 -preset slow -crf 20 -c:a libvo_aacenc -b:a 128k [20:33] <c_14> You could try copying the video instead of reencoding it. [20:33] <c_14> That would be your main bottleneck. [20:43] <kittonian> cool. i'll give it a shot [20:44] <xreal> I really can't understand, why I can't set audio/video stream title when doing stream copy. [20:48] <kittonian> can't copy the video stream. i get the pts has no value error and the video is all jumpy [20:49] <kittonian> the only way it comes out perfect is with the command i'm already using. guess i'll just have to wait for the encode to happen in realtime. [20:49] <kittonian> thank anyway [20:49] <kittonian> *thanks [21:07] <xreal> Uh, what's that? Stream #0:3(und): Data: none (rtp / 0x20707472), 11 kb/s (default) [21:21] <kittonian> someone around that can give me a hand writing a batch processing script for all of these conversions? sitting and waiting for each one to be done at this pace is taking forever [21:21] <kittonian> i've looked at a bunch of examples but haven't found one that will work for this situation [21:22] <kittonian> on mac os x btw [21:23] <c_14> for i in *.avi; do ffmpeg input_options -i "$i" output_options "${i%avi}avi"; done [21:23] <c_14> Should work, but no guarantees. [21:23] <c_14> That second avi should be an m4v [21:23] <xreal> Anyone with an idea, how to change metadata for video & audio streams which using "copy stream" ? [21:24] <kittonian> c_14: ha. that's the example i found [21:24] <kittonian> i need to be able to specify an input and output directory and then have it do one conversion at a time until the entire directory has been converted [21:25] <kittonian> the same filenames can be used for both the input and output files. just the extension change from avi to m4v [21:25] <c_14> for i in /path/to/input/dir/*.avi; do ffmpeg input_options -i "$i" output_options "/path/to/output/dir/${i%avi}m4v"; done [21:27] <kittonian> does this look right? [21:27] <kittonian> http://pastebin.com/vg8xSbUu [21:28] <c_14> $OUTDIR has to be ${OUTDIR} [21:28] <c_14> Other than that, it should be fine. [21:29] <kittonian> not the same for INDIR? [21:29] <c_14> You might want to get rid of the escapes in the pathnames though and just quote the variable declaration. [21:29] <c_14> You only need it for the output because you're appending a string after it. [21:29] <c_14> You might need it for INDIR, it wouldn't hurt though. [21:29] <kittonian> you lost me on the variable declaration [21:30] <c_14> Just make it ${INDIR} as well, it shouldn't hurt and may prevent errors. [21:30] <kittonian> and what about a . before m4v? [21:30] <c_14> Don't need it because you aren't deleting the . before the avi that you're deleting. [21:30] <kittonian> ahh [21:31] <kittonian> and this will do one conversion at a time, not try to do all of them at once right? [21:31] <c_14> yep [21:31] <kittonian> cool, let me give it a shot [21:34] <kittonian> got a bunch of "no such file or directory" errors due to the long path names in INDIR and OUTDIR [21:36] <c_14> Could you replace the do line with do echo "$i" "${OUTDIR}/${i%avi}m4v" and see if the output it gives you is the actual location of the files? [21:38] <kittonian> just tried that. it's not seeing it all as a single path, it's seeing it as a bunch of different paths [21:42] <c_14> for i in ${INDIR}/*.avi; do echo "$i" "${OUTDIR}/${i%avi}m4v"; done [21:42] <c_14> That works for me. [21:45] <kittonian> do you have " around the path variables? [21:46] <kittonian> and are you still escaping all the spaces? [21:46] <c_14> Yes. [21:46] <c_14> No. [21:46] <c_14> INDIR="/media/path/places" [21:46] <c_14> OUTDIR="/media/otherpath/out" [21:46] <kittonian> INDIR="/something's wrong/with/the/path" [21:46] <kittonian> that's what i have (there are special characters in some of the folder names) [21:47] <kittonian> every space and special character is being treated as a separate path [21:48] <clever> for i in "${INDIR}/*.avi"; [21:49] <clever> i think this will solve it [21:49] <c_14> That won't glob though. [21:50] <clever> hmmm, yeah, it treats it as a single argument [21:50] <c_14> Put the quotes only around indir like so: "${INDIR}"/*.avi [21:50] <c_14> That works. [21:50] <clever> yep [21:50] <clever> c2d ~ # for x in "/media/videos/4tb/dcc/"*;do echo $x;done [21:50] <clever> lists every file&dir in there [21:51] <kittonian> hmm, closer [21:51] <kittonian> it's trying to do all the files at one time [21:51] <kittonian> and it's still not finding the correct output file to use [21:51] <kittonian> should the quotes on the OUTDIR be different as well? [21:52] <c_14> "${OUTDIR}"/"${i%avi}m4v" [21:52] <clever> i see what your trying to do, let me try applying it to my video heap [21:52] <c_14> I dislike escaping things. :( [21:53] <clever> c2d ~ # for x in "/media/videos/4tb/dcc/"*mkv;do echo "$x" ${x%mkv}avi;done [21:53] <clever> prints out every mkv file, and its avi pair [21:53] <clever> one pair per line [21:53] <clever> /media/videos/4tb/dcc/Rise.of.the.Continents.S01E04.Australia.720p.HDTV.x264-DHD.mkv /media/videos/4tb/dcc/Rise.of.the.Continents.S01E04.Australia.720p.HDTV.x264-DHD.avi [21:54] <c_14> You should probably quote the ${x%mkv} part in case your filenames have spaces. [21:54] <clever> yeah, if i toss it at ls, all hell break loose [21:54] <clever> c2d ~ # for x in "/media/videos/4tb/dcc/"*mkv;do ls -lh "$x" "${x%mkv}avi";done [21:55] <clever> now its only one error per pair, since the avi versions dont exist [22:13] <kittonian> sorry about that. didn't mean to disappear. had to show the house we're selling :) [22:21] <kittonian> you guys still with me? [22:21] <c_14> ye, have you tried it with the new quoting yet? [22:23] <kittonian> i have not. which quoting would you like me to try? [22:23] <c_14> Put the quotes only around indir like so: "${INDIR}"/*.avi [22:23] <c_14> "${OUTDIR}"/"${i%avi}m4v" [22:23] <c_14> And outdir like so^ [22:23] <kittonian> ok, one sec [22:25] <kittonian> nope [22:25] <kittonian> it's treating both INDIR and OUTDIR as one long path [22:26] <kittonian> and still trying to do all files at once [22:27] <kittonian> http://pastebin.com/P7udTFbg [22:34] <c_14> http://sprunge.us/FgXO [22:34] <c_14> That works for me. [22:38] <kittonian> so far so good. it's converting the first file. this will take a little bit until we can see if it moves onto the next, but awesome thus far! [22:39] <kittonian> thank you [22:39] <c_14> np [22:39] <kittonian> it'll be about 30 minutes or so before we can see if all is well. i'll post back in a bit [22:39] <c_14> kk [22:42] <xreal> Anyone with an idea, how to change metadata for video & audio streams which using "copy stream" ? Neither -metadata:s:a:1 title="test" nor -metadata:s:a.1 title="test" works :/ [22:44] <c_14> Have you tried just -metadata title="test" ? [22:45] <xreal> c_14: I don't want it for the file, I want to name the streams. Like "Camera 1" and "Audio Cam 1" etc. [22:45] <xreal> brb 5 minutes (ofen in kitchen) [22:57] <xreal> re [22:57] <xreal> c_14: mp4box can do it, but it creates crazy MP4s sometimes. [22:58] <c_14> Have you tried it without the trailing ":1" ? I'd like to know if it's failing completely or just failing on that stream specifically. [22:58] <xreal> c_14: 1 sec [23:00] <xreal> c_14: Nope. Doesn't work. [23:00] <xreal> c_14: check the output of probe http://pastie.org/pastes/9095452/text [23:05] <c_14> Try it with a :0 instead of a :1 [23:06] <c_14> Hmm, weird. [23:06] <c_14> For me it works with mkv, but not with mp4. [23:08] <blippyp_> not sure where I read this: but ffmpeg apparently is broken for metadata - it also works differently depending on the container [23:08] <xreal> blippyp_: oh damn [23:09] <blippyp_> what container are you using xreal? it worked for mkv's for c_14 - try that one [23:09] <xreal> blippyp_: I'm using mp4 [23:09] <blippyp_> try mkv [23:09] <xreal> blippyp_: but I need mp4 :) [23:09] <xreal> but let me try [23:09] <blippyp_> just try it [23:10] <blippyp_> if it works then you know that's probably the case (if you believe me...) ;) [23:10] <xreal> yep. it works :( [23:10] <blippyp_> sorry dude [23:10] <blippyp_> that's probably your problem... :( [23:11] <blippyp_> bu I could be wrong - I think it happened ONCE before... ;) [23:11] <xreal> Is mp4 such outdated that nobody cates about it? [23:11] <xreal> cares* [23:11] <blippyp_> no - I think the metadata stuff isn't 'standardized' - google around - you'll get more reliable information about it than I could give ya [23:12] <xreal> blippyp_: with mp4box it works as expected, but it sometimes creates crazy files with multiple audio tracks [23:13] <xreal> blippyp_: But I understand what you mean :D [23:13] <blippyp_> like I said - it's not standardized - and when I read about it people were complaining about ffmpeg being broken for metadata - I'm guessing it's a shot in the dark - an 'unreliable' issue.... [23:14] <blippyp_> for lack of better words - I'm going off of hazy memories of an article I read once - so like I said you should google around, you'll get better and much informative information about it... [23:14] <xreal> blippyp_: wikipedia says: "MP4 files can contain metadata as defined by the format standard, and in addition, can contain Extensible Metadata Platform (XMP) metadata" [23:15] <xreal> so it can contain metadata, but it's not standardized? [23:15] <blippyp_> you forget that I stated that people were complaining about ffmpeg being 'broken' about the metadata.... [23:15] <blippyp_> just because mp4's are suppose to have the ability for that metadata doesn't mean that ffmpeg has been designed for it yet [23:16] <blippyp_> I can't prove any of this - but you just saw it with your own eyes... you be the judge [23:18] <xreal> blippyp_: I think, most of the guys want details, like "episode" etc. My needs are too special :( [23:18] <xreal> Nobody cares about the video stream's name :) [23:18] <blippyp_> ffmpeg might not be what you want then... [23:19] <xreal> blippyp_: <ducks himself> what's the alternative? [23:19] <c_14> It's open source, just patch it in ;) [23:19] <blippyp_> you can probably do the encoding and stuff - 'the hard part' with ffpeg and then use a different program to attach the metadata you want afterwards though [23:19] <blippyp_> there's always that too... ;) [23:19] <xreal> blippyp_: that's what I am doing right now [23:19] <blippyp_> have fun with that btw... haha [23:19] <xreal> ffmpeg: I am encoding the files, splitting them [23:19] <xreal> then I push them in mp4box [23:20] <xreal> But MP4box adds unknown streams sometimes: Stream #0:3(und): Data: none (rtp / 0x20707472), 11 kb/s (default) [23:20] <blippyp_> see if ffmpeg will rip them out [23:20] <xreal> What's in the data stream at 11 kb/s ? [23:20] <xreal> blippyp_: using raw ? [23:21] <blippyp_> ffmpeg is suppose to 'copy' the metadata from the source - maybe it will keep the changes you made [23:21] <blippyp_> round and round you go... kinda a shitty situation [23:21] <xreal> blippyp_: Is there also a way to remove the metadata from the source? [23:21] <blippyp_> not sure [23:22] <blippyp_> when you copy the video - it apparently takes the metadata with it - but to extract just the metadata I don't know [23:22] <xreal> What I don't like about mp4box: it adds dates. [23:22] <xreal> Since I'm on Windows, I can't recompile modded sources that easy as on Linux. [23:22] <blippyp_> maybe that's part of the metadata 'standard' for mp4? [23:23] <xreal> blippyp_: but 11 kb/s is pretty much. [23:24] <blippyp_> sorry xreal - I gotta go - only came downstairs for a smoke - got a easter party to go to now with my daughter - good luck [23:24] <xreal> ok thanks. happy eastern [23:24] <blippyp_> no problem and happy easter to you guys too [23:30] <kittonian> c_14: it finished the first file and has successfully moved on to the next one. this is great. thanks again for all your efforts. [23:30] <c_14> np [23:41] <xreal> damn, libmp3lame doesn't write MP3 tag... [00:00] --- Mon Apr 21 2014

1 0