[Ffmpeg-devel-irc] ffmpeg.log.20161202
burek
burek021 at gmail.com
Sat Dec 3 03:05:01 EET 2016
[08:34:08 CET] <thebombzen> welp, I just found the framestep filter
[08:34:15 CET] <thebombzen> which is what I've been trying to do this whole time
[08:35:08 CET] <thebombzen> framestep=2 does the trick
[11:13:22 CET] <superware> can someone please have a look at https://trac.ffmpeg.org/ticket/5615 ?
[14:38:25 CET] <Lope> anyone here familiar with video encoding or avconv? To reduce a video's size I reencoded a video as follows: `avconv -i in.mp4 -vcodec libxvid -b 500k -acodec libmp3lame -ac 1 -ab 64k out.mp4` It plays perfectly in VLC, but Whatsapp plays it without sound. Chrome won't play it with no error given. FF complains about some libav vulnerability and won't play it.
[14:38:43 CET] <Lope> I mean what kind of a shitty video player can't handle lame mp3 audio?
[14:39:57 CET] <BtbN> avconv is not an ffmpeg tool.
[14:40:27 CET] <BtbN> And you want h264 + aac for wide support.
[14:40:49 CET] <Lope> I mean what kind of a shitty video player can't handle lame mp3 audio?
[14:41:05 CET] <Lope> BtbN: Oh cool :) thanks.
[14:41:31 CET] <Lope> Yeah I just luckily picked AAC looking at `avconv -codecs` AAC seemed familiar common, I've seen it on a lot of rips.
[14:42:10 CET] <Lope> Sorry I didn't mean to repeat my 3rd last remark. I meant to say: I tried encoding the audio with AAC now, and it plays from /tmp/ in chrome, with sound. *shrugs*
[14:42:45 CET] <furq> yeah you really shouldn't use xvid
[14:42:54 CET] <furq> i'm surprised any browsers play it
[14:42:56 CET] <Lope> I used: avconv -i in.mp4 -vcodec libx264 -b 500k -acodec libvo_aacenc -ac 1 -ab 64k out.mp4
[14:43:08 CET] <furq> yeah that ought to work
[14:43:09 CET] <Lope> Well the quality is a hell of a lot better than mpeg4
[14:43:16 CET] <furq> you can probably get rid of -b 500k though
[14:43:17 CET] <Lope> mpeg4 makes me suicidal.
[14:43:33 CET] <Lope> well, if I don't include -b 500k the filesize gets too high.
[14:43:33 CET] <furq> there's no reason to use mpeg4 in this day and age
[14:43:45 CET] <Lope> furq: what format do you recommend?
[14:43:48 CET] <furq> with x264?
[14:43:48 CET] <jkqxz> "what kind of a shitty video player can't handle x" -> most video players, probably? Wide codec support is just not a thing outside the lavc-based players (i.e. vlc, mpv, etc.).
[14:44:19 CET] <Lope> "<furq> yeah you really shouldn't use xvid" do you mean I shouldn't use libx264?
[14:44:24 CET] <furq> no i mean xvid
[14:44:28 CET] <furq> you should use x264
[14:44:36 CET] <Lope> oh, good :)
[14:45:05 CET] <Lope> Is there some recommended list of codecs like favour and avoid?
[14:45:17 CET] <furq> it depends what you're targeting but x264 is generally a good choice
[14:46:29 CET] <m3nTe> hi guys
[14:47:51 CET] <Lope> I tried to use the "aac" codec but libav complained that it's not stable or whatever. So I used libvo_aacenc and it didn't complain, and the video worked.
[14:47:56 CET] <Lope> (in chrome)
[14:48:13 CET] <furq> you should probably upgrade to a newer version of ffmpeg
[14:48:39 CET] <furq> https://www.johnvansickle.com/ffmpeg/
[14:48:48 CET] <Lope> I find it amazing avconv doesn't support encofing ogg by default. I'm thinking "WTF" it's a free format. Last time I wanted to encode ogg with avconv I had to compile it from source.
[14:49:02 CET] <furq> just use those static builds if your distro ffmpeg/libav sucks
[14:49:26 CET] <BtbN> avconv is still not ffmpeg.
[14:50:19 CET] <Lope> Why do the builds in ubuntu etc suck ass? I mean wtf? It's like they build it and cripple it. I could understand if it was some ancient really arbitrary crappy encoder that nobody uses but fucking OGG. I mean seriously...
[14:50:32 CET] <furq> you're on an old ubuntu
[14:50:42 CET] <m3nTe> question: I'm trying to using the "read native frame rate" using the `-re` option, and works amazingly. I need to stream a video file to RTMP. I'm doing this because I need it to loop infinitely.. /usr/local/bin/ffmpeg -re -fflags +genpts -i video.mp4 -c copy -f flv "rtmp://...."
[14:50:43 CET] <Lope> Yeah 14.04
[14:50:48 CET] <furq> debian has had proper ffmpeg with sensible defaults for a few years now
[14:50:54 CET] <furq> ubuntu 16+ has it as well afaik
[14:51:01 CET] <m3nTe> the issue is that it doesn't loop. any help?
[14:51:19 CET] <furq> why would that loop
[14:51:20 CET] <Lope> BtbN: Yeah ffmpeg didn't ship with ubuntu last time I checked, it's a whole dumb political thing. Why not support both, they're different projects now, right?
[14:51:30 CET] <furq> Lope: because you're on an old ubuntu
[14:51:30 CET] <m3nTe> question: I'm trying to using the "read native frame rate" using the `-re` option, and works amazingly. I need to stream a video file to RTMP. I'm doing this because I need it to loop infinitely.. /usr/local/bin/ffmpeg -re -fflags +genpts
[14:51:35 CET] <BtbN> ffmpeg ships with recent versions of ubuntu.
[14:51:39 CET] <m3nTe> -i -stram_loop - 1 video.mp4 -c copy -f flv "rtmp://...."
[14:52:13 CET] <Lope> BtbN: furq: alrighty
[14:52:14 CET] <furq> m3nTe: add `-stream_loop -1` before -i
[14:52:58 CET] <furq> oh nvm it's there
[14:53:16 CET] <m3nTe> furq: sorry, I pasted it wrongly.. yeah, I'm already doing that. but doesn't loop at all. when the video reaches the end I receive this error: av_interleaved_write_frame(): Broken pipe
[14:53:19 CET] <furq> oh right
[14:53:20 CET] <Lope> Is ogg supported in crappy video players like whatsapp/chrome/ff?
[14:53:25 CET] <furq> you've got it after -i
[14:53:34 CET] <furq> Lope: it's supported in most browsers
[14:53:40 CET] <furq> not safari iirc
[14:53:47 CET] <furq> and not old IE, i forget if edge supports it
[14:53:50 CET] <Lope> what's better: ogg or aac?
[14:53:56 CET] <furq> well you're using mp4 so you should use aac
[14:54:02 CET] <Lope> Ok cool.
[14:54:02 CET] <furq> i don't think vorbis is supported in mp4
[14:55:30 CET] <Lope> Regarding: avconv complaining "aac" is not stable. and libvo_aacenc worked. Should I be concerned?
[14:55:49 CET] <Lope> You said I should use a newer version of ffmpeg, do you mean "aac" is stable in new versions?
[14:55:55 CET] <Lope> What's the diff between the encoders?
[14:56:18 CET] <AssPirate> Yeah. aac is stable now
[14:57:17 CET] <BtbN> libvo aac is horrible. Don't use it
[14:57:38 CET] <m3nTe> question: ffmpeg -re -fflags +genpts -stream_loop -1 -i video.mp4 -c copy -f flv "rtmp://&.. when the video reaches the end I receive this error: av_interleaved_write_frame(): Broken pipe and it doesnt loop at all. any help?
[15:01:31 CET] <Lope> suckage
[15:01:43 CET] <Lope> i hate my old distro.
[15:02:26 CET] <Lope> So the x264 with libvo aac doesn't play in chrome when it's online. Only from /tmp :p
[15:02:28 CET] <Lope> fuck
[15:03:03 CET] <Lope> Does libvo aac get labelled as libvowhatever or aac?
[15:03:17 CET] <Lope> once it's encoded in the file?
[15:04:05 CET] <AssPirate> m3nTe: I used to get that same error when I was doing stuff like that out to an rtmp stream. But I was concatenating multiple files. Never failed when the output was a file. And it was inconsistent. Sometimes it worked. Sometimes it didn't. Never really found out why though.
[15:04:59 CET] <m3nTe> AssPirate: damn. so do you think the issue is because is streaming to an rtmp server..?
[15:07:47 CET] <AssPirate> Not necessarily the cause, but I'm fairly certain it's related to it. Can you try to output to a file? See if it loops properly then.
[15:08:07 CET] <m3nTe> AssPirate: I'll give it a try..
[15:10:07 CET] <m3nTe> AssPirate: yeah, works. hmm, weird.
[15:11:32 CET] <AssPirate> Right? ¯\(°_o)/¯
[15:32:34 CET] <Lope> Okay, well my video works in whatsapp, so alls well :/
[15:32:36 CET] <m3nTe> AssPirate: the issue looks like ffmpeg takes a while to restart the loop of the video....
[15:32:39 CET] <m3nTe> mmmm
[15:32:40 CET] <m3nTe> why?! wtf
[15:32:53 CET] <Lope> thanks for the help BtbN furq
[15:47:41 CET] <AssPirate> m3nTe: Could be. I was reencoding my videos on the fly and concatenating them. I've switched to reencoding them beforehand. Haven't had that issue since.
[16:12:45 CET] <Jaex> I'm trying to decide which one should be default nvenc preset for screen recording purposes: https://dl.dropboxusercontent.com/u/14076298/ShareX/2016/12/ShareX.vshost_NiuGu1FvPO.png anyone have suggestion?
[19:06:04 CET] <sardok> Hi, i have an audio file which has 5 seconds. I want to extend it to 15 seconds while padding silence. i used this command 'ffmpeg -i 4927.mp3 -filter_complex 'apad' -t 15 -y /tmp/padded-15.aac' but resulting video is still 5 seconds for some reason. Any idea?
[19:07:34 CET] <kbarry> not sure
[19:07:47 CET] <kbarry> But you might want to produce a 5 second silent video
[19:07:53 CET] <sardok> if i dont give .mp3 as an extension to the resulting file, it is ok.
[19:07:53 CET] <kbarry> then use ffmpeg to append them together
[19:08:01 CET] <kbarry> silent then video then silent again
[19:13:48 CET] <sardok> thanks kbarry
[19:14:47 CET] <kbarry> sardok: Yeah, sorry i can't be of more help. Just trying to do my part (i'm a newb myself)
[19:14:57 CET] <kbarry> I can say that the "-t" might not do what you think it does
[19:15:09 CET] <kbarry> ie, it might only work for limiting things, not for lengthening things
[19:15:26 CET] <kbarry> IE, play a 500 second video with -t 15, and you get 15 seconds,
[19:15:28 CET] <furq> that should work based on the description of apad
[19:15:51 CET] <furq> If neither the pad_len nor the whole_len option is set, the filter will add silence to the end of the input stream indefinitely.
[19:21:14 CET] <sardok> if i remove encoding, duration is as expected, however adding encoding to some other format (aac in this example) does not give duration as expected.
[19:22:30 CET] <MrMonkey31> hey um, guys
[19:23:10 CET] <durandal_170> yes?
[19:23:33 CET] <MrMonkey31> does anyone use vlc here? I keep looking at their codec information. the "display resolution" keeps turning out different in my encodes from "resolution" and it's bugging me out
[19:24:21 CET] <durandal_170> that depends on sample aspect ratio
[19:24:25 CET] <MrMonkey31> this wouldn't have anything to do with "pixel format" would it?...
[19:24:49 CET] <MrMonkey31> @durandal_170, or are they completely separate?
[19:26:24 CET] <durandal_170> what's different and by how much?
[19:28:23 CET] <MrMonkey31> thx for asking, man. the two are identical in my source vid, 480x352
[19:29:02 CET] <MrMonkey31> then I use a scale and a crop, like so: crop=w=428:h=348:x=27:y=4, scale=434:348
[19:30:04 CET] <MrMonkey31> and after that the reported values are res: 448x352, display res 434x348
[19:31:13 CET] <MrMonkey31> coupled with the resolution restrictions of hevc, it becomes a confusing jumble in my head. /rant I guess I should be taking in the nominal resolution, ignoring display resolution for my purposes - //but// it does not add up :(
[19:31:36 CET] <durandal_170> scale filter does that to keep dar
[19:33:18 CET] <MrMonkey31> alright is there some missing paramater to prevent such automation, or...?
[19:35:13 CET] <durandal_170> dunno, I can't remember, there are setsar and setdar filters
[19:36:50 CET] <kerio> which resolution restrictions?
[19:37:22 CET] <MrMonkey31> oh, to be sure! so far it still does not make sense to me, why scale would try anything automatically
[19:37:47 CET] <MrMonkey31> kerio, I'm referring to messages complaining about your chosen resolution when encoding with hevc
[19:38:59 CET] <MrMonkey31> I mean hypothetically I could just keep 1/2 the video on the vertical axis. then the aspect ratio would be doubled. would it double up my number of pixels again?
[19:47:43 CET] <MrMonkey31> obviously not - or we hope so anywya. well I suppose I could simply stop caring, unless I see like, 150 pixels appear. all I was interested in was counting up to make sure my encode has the expected number of pixels
[19:48:33 CET] <kerio> does hevc really require a multiple of 32
[19:48:51 CET] <MrMonkey31> but it doesn't appear to be possible. unless there's yet a third value, one that's different from what vlc reports
[19:49:51 CET] <MrMonkey31> for e.g. I just did the same encode using crop but without scale, still got separate values for resolution and display resolution. so crop must do something as well, since the only other filter I've got is eq
[19:51:23 CET] <MrMonkey31> my other, more general anxiety is that this level of complexity will eventually relegate encoders to some sort of state of hermitude if they are to continue pursuing a specific desired result
[19:56:35 CET] <grublet> MrMonkey31: that is why i stopped encoding
[19:57:00 CET] <grublet> i had to futz with it forever to figure out how to make it stop doing that, and i cant remember now how i did it
[19:58:02 CET] <grublet> anamorphosis is perhaps the worst thing to happen to video other than interlacing and reduced chroma resolution but gonna check out now before i get too ranty
[19:58:25 CET] <MrMonkey31> grublet, we lament together: NOOOOOOOOOO!
[19:58:54 CET] <grublet> yeah i feel your pain bro
[19:59:10 CET] <grublet> im pretty sure scale has some parameter to prevent this autoscaling
[20:03:03 CET] <grublet> trying to figure out what i did before cause now its bothering me lol
[20:04:08 CET] <grublet> best solution for now is probably just using setsar and setdar
[20:04:24 CET] <grublet> personally, i hate when programs make these kinds of decisions for me, especailly with stuff like pixel formats or colorspaces
[20:04:59 CET] <grublet> kinda miss the days of having to write long ass command lines for x264 before the preset system
[20:06:16 CET] <grublet> kerio: imo hevc is a ridiculous format, it just looks like im watching divx/xvid videos all over but with more smearing
[20:06:26 CET] <kerio> lmao
[20:06:28 CET] <kerio> why use it then
[20:06:31 CET] <grublet> i dont
[20:06:42 CET] <grublet> if i encode anything nowadays its usually vp8 in webm
[20:06:58 CET] <grublet> before that i would do h264/aac in mp4
[20:07:07 CET] <kerio> >not ao1/opus
[20:07:15 CET] <kerio> *av1
[20:07:28 CET] <grublet> not a fan of audio codecs that internally resample audio, and i never used av1 or heard of it til now
[20:07:39 CET] <kerio> grublet: what's the fascination with vp8?
[20:08:03 CET] <grublet> kerio: its basically guaranteed to work for anyone at this point, due to the surge of webm popularity recently
[20:08:04 CET] <kerio> to be fair opus only resamples if your audio wasn't 48kHz already
[20:08:12 CET] <kerio> ...as it should've been
[20:08:13 CET] <kerio> :<
[20:08:14 CET] <grublet> and most audio i have isnt 48khz
[20:08:23 CET] <kerio> grublet: not even movie audio?
[20:08:23 CET] <furq> x264 is also guaranteed to work for anyone
[20:08:28 CET] <furq> and is generally much better than vp8
[20:08:35 CET] <kerio> yeah, no vp8 on apple-y stuff
[20:08:35 CET] <grublet> kerio: movies are a very samll portion of the audio i encode
[20:08:43 CET] <grublet> most stuff out there barring movies is 44.1
[20:08:50 CET] <kerio> yeah but i didn't suggest opus in general
[20:08:53 CET] <kerio> i suggested av1/opus
[20:08:56 CET] <grublet> vp8 woprks in most browsers
[20:09:08 CET] <furq> h264 works in more browsers though
[20:09:08 CET] <kerio> doesn't work on mine :^)
[20:09:16 CET] <grublet> furq: that is not my experience
[20:09:17 CET] <markvandenborre> grublet: I would understand if you would say you do vp9 for political purposesm; that would be a valid reason
[20:09:20 CET] <kerio> whereas h264 works in every browser i guess
[20:09:37 CET] <furq> http://caniuse.com/#feat=mpeg4
[20:09:39 CET] <grublet> markvandenborre: what do you mean by poltiical purposes?
[20:09:40 CET] <markvandenborre> but for practical reasons, h264 is really superior unfortunately
[20:09:48 CET] <kerio> is av1 finalized?
[20:09:52 CET] <furq> http://caniuse.com/#search=webm
[20:10:04 CET] <grublet> there are too many open video/audio standards at this point
[20:10:13 CET] <kerio> not really
[20:10:16 CET] <furq> the only reason to use vp[89] over h264 is because one is open source and license-free
[20:10:29 CET] <furq> or vp8 rather
[20:10:35 CET] <furq> vp9 should compress better at least
[20:10:44 CET] <kerio> furq: is vp9 actually supported tho
[20:10:52 CET] <furq> yes
[20:11:04 CET] <kerio> i thought it was similar to hevc
[20:11:08 CET] <furq> oh
[20:11:12 CET] <furq> well it's supported in browsers
[20:11:25 CET] <furq> youtube use it for desktop firefox/chrome and probably edge
[20:11:28 CET] <markvandenborre> (but it will take more cpu to decode)
[20:11:39 CET] <furq> maybe not for mobile because of the lack of hw decode
[20:11:46 CET] <kerio> meanwhile quicktime still doesn't support h264 with yuv444
[20:12:08 CET] <kerio> does anything actually accelerate one of those fancy open formats tho
[20:12:15 CET] <furq> does quicktime support anything
[20:12:16 CET] <kerio> like, do android phones have vp8 acceleration?
[20:12:20 CET] <furq> i doubt it
[20:12:21 CET] <kerio> furq: h264 yuv420p
[20:12:35 CET] <markvandenborre> kerio: there's a lot of hardware that does vp9 decoding recently
[20:12:44 CET] <furq> yeah it's very recent though
[20:12:49 CET] <furq> especially on desktop
[20:12:50 CET] <markvandenborre> encoding not so much yet
[20:13:00 CET] <kerio> but why, when av1 is better? :^)
[20:13:05 CET] <furq> well exactly
[20:13:36 CET] <furq> from what i've heard av1 is mostly vp10, so vp9 development seems to be pretty much dead
[20:21:50 CET] <__raven__> hi
[20:24:08 CET] <__raven__> i need to temporarily revive an old optiplex analogue video pci io card for capturing with ffmpeg. unfortunately very rare information is displayed and video0 does not show anything yet. could you take a look at it?
[20:26:30 CET] <tdr> __raven__, probably means you're missing the v4l driver for it. look at what lspci -k calls it to make sure you have a driver there and then google for the name (from that output) + linux driver
[20:27:43 CET] <tdr> a binary distro prob has the driver and just needs it modprobed, but it may be split into its own package depending on how your distro packages kernel stuff
[20:28:33 CET] <__raven__> tdr this is what i got so far: http://pastebin.com/1r9yTUK6
[20:29:13 CET] <__raven__> lspci -k shows 05:00.1 Multimedia controller: Optibase Ltd Device 5001 (rev ff) without a line below - how to figure out the right module?
[21:56:22 CET] <livingBEEF> Is it possible to "remap" colors? As in make it to turn some specified colors into other specified colors.
[21:58:16 CET] <kiroma> Hello
[21:59:56 CET] <DHE> livingBEEF: depends on what you need. hue and saturation can be played with. eg: red->green, green->blue, blue->red. but that's a global thing, you force that rotation on the whole image on every pixel.
[22:02:24 CET] <livingBEEF> Yeah, that's not what I want. I want it like if I swapped color palette in a gif image without touching other data
[22:03:27 CET] <kiroma> "recompile with -fPIC" "Requested extra CFLAGS ' -fPIX' not supported by compiler"... Oh.
[22:07:02 CET] <kiroma> Ok, I am blind.
[22:11:48 CET] <kerio> rip
[22:29:19 CET] <kiroma> Well it turns out that ffmpeg doesn't really like libvpx1.6
[23:35:30 CET] <Sashmo> does anyone know why I keep getting this error?? [h264 @ 0x4430bb40]mmco: unref short failure nothing wrong with video, but it throws the error
[00:00:00 CET] --- Sat Dec 3 2016
More information about the Ffmpeg-devel-irc
mailing list