[Ffmpeg-devel-irc] ffmpeg.log.20161130

burek burek021 at gmail.com
Thu Dec 1 03:05:01 EET 2016


[00:23:23 CET] <kurufu> so what is the A-V: number in the command line when playing something back with ffplay?
[00:40:28 CET] <DHE> audio time minus video time. sync/drift
[00:45:05 CET] <Sashmo> does anyone know how to get around this error "There are not enough buffered video frames. Video may misbehave!" when trying to output video on decklink card?
[00:46:39 CET] <Sashmo> it looks like its a cpu and core issue.... no matter what I use on threads, its maxing out one core on my 16 core setup....  Its a multi threading thing 100%.... any suggestions?
[00:57:50 CET] <kurufu> And what does the 'fd=' part mean DHE ?
[01:22:40 CET] <DHE> kuroro: huh?
[01:23:25 CET] <kurufu> The number right after A-V, given as say "fd= 147"
[01:23:44 CET] <kurufu> Looks like some people think its frames dropped?
[02:16:57 CET] <thebombzen> kurufu: I think it's that. sometimes ffplay will drop frames even if the framerate is lower than the monitor refresh rate
[02:19:14 CET] <kurufu> im just looking to understand what these cryptic letters and numbers mean.
[02:19:23 CET] <kurufu> is it actually frames dropped?
[02:21:05 CET] <kurufu> and if anyone happens to know what the "f=X/Y" portion happens to mean that would be swell.
[02:23:05 CET] <DHE> this is ffplay?
[02:23:16 CET] <DHE> oh you already said so
[02:33:16 CET] <kurufu> Yes i did.
[02:33:17 CET] <kurufu> and it i
[02:33:19 CET] <kurufu> is*
[02:39:53 CET] <jarred> When I try to concat two mov files, I'm getting a strange error "Could not find tag for codec prores in stream #0, codec not currently supported in container". Here's a gist: https://gist.github.com/Jarred-Sumner/a59179c54393d62a129a92d92b2ff5bb. What am I doing wrong?
[02:40:03 CET] <jarred> I need to preserve transparency in the mov
[03:18:54 CET] <c_14> jarred: -c copy is overriding -c:v qtrle
[03:18:58 CET] <c_14> put -c:v qtrle after -c copy
[03:20:27 CET] <jarred> c_14: Could not find tag for codec qtrle in stream #0, codec not currently supported in container
[03:20:27 CET] <jarred> Could not write header for output file #0 (incorrect codec parameters ?): Invalid argument
[03:21:01 CET] <jarred> is what it's saying now
[03:22:00 CET] <c_14> use -f mov or rename mp4 to mov
[03:24:52 CET] <jarred> c_14: [qtrle @ 0x7fa460019a00] Problem: pixel_ptr = -4, pixel_limit = 2741760 now :(
[03:24:55 CET] <jarred> when renaming mp4 to mov
[03:24:57 CET] <jarred> i will try -f mov
[03:25:08 CET] <jarred> (same for -f mov)
[03:26:21 CET] <c_14> hmm, no clue
[03:42:30 CET] <thebombzen> jarred I'm getting a 404 for that gist
[03:43:25 CET] <jarred> thebombzen: https://gist.github.com/Jarred-Sumner/a59179c54393d62a129a92d92b2ff5bb
[03:43:27 CET] <jarred> did that work?
[03:44:40 CET] <thebombzen> jarred if you're only copying one stream why use -c copy?
[03:46:16 CET] <jarred> i want to merge two videos together
[03:46:21 CET] <jarred> or, specifically, i want to append
[03:48:19 CET] <thebombzen> not sure if prores supports stream-level concatenation, but can you use: ffmpeg -f concat -i files.txt -c copy -f mov output.mov
[03:49:28 CET] <thebombzen> cause jarred if ffmpeg doesn't whine on the decoding side then that should work
[03:49:57 CET] <jarred> will that preserve the alpha channel?
[03:51:18 CET] <thebombzen> it should because you're not re-encoding
[03:51:32 CET] <thebombzen> the encoded data will be the same
[03:51:52 CET] <jarred> It worked but created a 7 minute long file https://gist.github.com/Jarred-Sumner/94005cd1e373d94bc1fecd4b88a9be06
[03:52:03 CET] <jarred> (for something that should've been 2 seconds)
[03:53:23 CET] <thebombzen> what happens if you ffprobe the two files? what is the output of that?
[03:53:49 CET] <thebombzen> also note this: [concat @ 0x7f8f25003400] Could not find codec parameters for stream 1 (Unknown: none): unknown codec
[03:54:17 CET] <thebombzen> it's possible prores doesn't support stream-level concatenation. if that's true you'll have to use the concat demuxer.
[03:56:03 CET] <jarred> okay i'll run ffprobe
[03:56:41 CET] <jarred> thebombzen: https://gist.github.com/Jarred-Sumner/e53391805a15d3309fa25780a919fbc0 ffprobe
[03:56:55 CET] <jarred> i tried using the concat demuxer and it didn't work either
[03:57:28 CET] <thebombzen> so there's a data stream in each file
[03:57:40 CET] <thebombzen> how do you expect to concatenate those?
[03:57:52 CET] <thebombzen> there's no "obvious" way to do that.
[04:03:58 CET] <thebombzen> because if you don't care about the data, you can do: ffmpeg -i first.mov -i second.mov -map 0:v -map 1:v -lavfi concat -c prores -f mov output.mov
[04:04:07 CET] <thebombzen> although you might have to set the bitrate with -b:v
[04:04:26 CET] <thebombzen> and yes, that'll preserve the alpha channel if ffmpeg can encode prores in the pixel format
[04:04:52 CET] <thebombzen> if it cannot it'll automatically change the pixel format, and will print a warning, and the output won't have it.
[04:05:32 CET] <thebombzen> the pixel format is yuva444p10le (notice the "a" in "yuva", that's the alpha channel)
[04:21:23 CET] <relaxed> "ffmpeg -h encoder=prores" says it only supports yuv422p10le
[04:24:04 CET] <relaxed> prores_ks supports yuva444p10le
[04:27:29 CET] <jarred> i got it working
[04:27:33 CET] <jarred> what i did was
[04:27:46 CET] <jarred> convert it into an avi file using mencoder
[04:27:57 CET] <jarred> then turn those avi files into webm (this is for playing on a website) with ffmpeg
[06:01:57 CET] <nate> Is there a generic list of what formats ffmpeg will likely always support cross-platform?  ie; MP4, webm, flv, etc
[06:02:00 CET] <nate> *eg;
[06:05:16 CET] <Filarius> i think you need check each build about what formats supported
[06:06:44 CET] <nate> Hm, yeah was afraid of that.  Writing a generic thumbnailer script, was trying to set a generic 'default' list of what ffmpeg would likely always support.  MP4, webm and flv were the only ones that really came to mind though :P
[06:07:02 CET] <nate> (ie; that package managers wouldn't remove from a standard ffmpeg compile)
[06:10:14 CET] <nate> That asked, a more random question as I don't have an ffmpeg installed to test with at the moment (still writing the script based on documentation details), but what exactly does the ffmpeg binary 'return' if a thumbnail process was successful?
[06:10:40 CET] <nate> I'm assuming with `-v fatal` set it'll just return the error string in case of a failure, but otherwise does it just return a 0/1 for false/true?
[06:20:21 CET] <relaxed> nate: 0 = success
[06:21:20 CET] <nate> Cool, thanks much :P
[06:23:36 CET] <relaxed> ffmpeg has native support for all the formats you listed
[06:24:48 CET] <nate> Yeah I figured as much, but I know sometimes package managers 'mess' with stuff, but I figured those three, at minimum, will likely always be safely there
[06:25:06 CET] <nate> I just wanted to define a 'default' list in this script
[08:28:08 CET] <thebombzen> nate: when in doubt, use matroska
[08:28:41 CET] <nate> mkv too? o.O  ffmpeg generally supports that out of the box?
[08:28:50 CET] <thebombzen> why wouldn't it?
[08:28:55 CET] <thebombzen> it's a free format
[08:29:07 CET] <thebombzen> it's one of the more likely ones that would work
[08:29:17 CET] <thebombzen> it's also a web standard now
[08:29:20 CET] <nate> just also I guess uncommon to spot, so didn't think of it :P
[08:29:41 CET] <thebombzen> it's not uncommon, but that depends on your uses
[08:30:05 CET] <thebombzen> it's the most common thing I come across to be honest and I try to use it whenever I can
[08:30:36 CET] <thebombzen> matroska essentially does everything I need it to and doesn't really have any liabilities. the only thing that should work better with FFmpeg is NUT, which is specifically an ffmpeg container.
[08:30:47 CET] <thebombzen> although that won't work with anything not based on libavformat.
[08:31:41 CET] <thebombzen> and ffmpeg supports it out of the box unless you disable it, which you would do if you want a very small build (embedded device?) but there's usually no reason to
[08:32:08 CET] <thebombzen> also, webm IS matroska just so you know. so that also shouldn't surprise you.
[08:32:37 CET] <thebombzen> webm is matroska with a few extra restrictions but all webm files are valid matroska files
[08:55:32 CET] <retard> mkv is uncommon?
[08:55:37 CET] <retard> it's everywhere
[09:05:19 CET] <ritsuka> retard: I guess he meant uncommon in the professional video and distribution world
[09:18:01 CET] <yong> Assuming an uncompressed source (16-bit, 44.1 KHz CD-RIP), how often could you re-encode AAC 320 kbit/s until the quality drops below a 128 kbit/s first-time encode?
[09:20:22 CET] <thebombzen> yong: that depends on the source material
[09:20:55 CET] <thebombzen> certain genres of music are much more compressible than others. metal will lose quality much faster than folkrock for example
[09:21:35 CET] <thebombzen> if it's speech it'll take a lot of re-encodes.
[09:22:10 CET] <thebombzen> it also depends on the encoder. some encoders are relative stronger at higher bitrates and some shine at lower bitrates
[09:23:28 CET] <yong> thebombzen: In that case, assuming metal/electronic type music that isn't easily compressible. Although I'm surprised that it matters considering we're comparing two compressed audio files in the end. Also assuming the best high bit-rate encoder vs the best low bit-rate encode, they don't have to be the same.
[09:24:01 CET] <thebombzen> as for the actual number of times? I have no idea cause I haven't done the test
[09:24:08 CET] <thebombzen> I've had no reason to.
[09:24:15 CET] <thebombzen> there's no reason you can't do this testing though.
[09:24:38 CET] <yong> How'd you set this test up? I'm not sure I wanna trying actually hearing the difference^^
[09:24:46 CET] <yong> -ing
[10:00:57 CET] <yong> thebombzen: Damn, the ffmpeg native aac encode makes it sound absolutely disgusting after 8 reencodes, wtf
[10:01:18 CET] <thebombzen> well I mean, that is sort of what you get for encoding it eight times
[10:01:36 CET] <thebombzen> and yea there's no claim that the native aac encoder is good. it's not "experimental" anymore but it's nowhere near good
[10:02:13 CET] <thebombzen> if you want a libavcodec-based encoder then you could use libfdk_aac, but it's nonfree. you have to build it yourself from source and then compile ffmpeg yourself with --enable-nonfree.
[10:02:25 CET] <thebombzen> alternatively you could use Nero's codec which is good but not that good.
[10:03:01 CET] <thebombzen> and how would I set it up? use a program called "abx" for... surprise! abx testing.
[10:03:23 CET] <thebombzen> it's in the AUR but I'm not sure how to get it on other distros
[10:04:02 CET] <thebombzen> also why do you care about re-encode quality? it's not really a good metric for how good an encoder is.
[10:04:28 CET] <thebombzen> the only reason to care would be if you're constantly re-encoding, but if that's true then you probably shouldn't be doing that.
[10:06:10 CET] <thebombzen> however I'm not sure why you're using AAC so I can't recommend an alternative
[10:06:33 CET] <thebombzen> cause if your goal is quality lossy codec, libopus is usually the best.
[10:07:32 CET] <thebombzen> if you need it to be Quicktime compatible, try ALAC (Apple Lossless). it's supported by Apple and ffmpeg-based projects, and is 100% resilient to re-encodes.
[10:08:24 CET] <thebombzen> for non-apple applications, FLAC is more versatile and slightly better.
[10:08:33 CET] <thebombzen> but then again, why do you want to re-encode it eight times?
[10:09:07 CET] <thebombzen> that's generally "not recommended" i.e. I mean well, "makes it sound absolutely disgusting" sums it up pretty well
[12:33:07 CET] <mpsag> I started a video conversion process and it has the PID 12345. How do I get progress status information from this process?
[12:45:09 CET] <relaxed> mpsag: which OS?
[12:45:19 CET] <mpsag> Debian
[12:46:18 CET] <relaxed> you con
[12:46:30 CET] <relaxed> er, you can't see the console output?
[12:48:06 CET] <mpsag> I could, but there are multiple video conversion processes running and I want to get the progress status of one of them
[12:52:10 CET] <relaxed> You would have to script something to do that.
[15:00:03 CET] <ikevin> can i use ffmpeg to test if a stream is working?
[15:34:18 CET] <fred1807> can I use ffprobe to test is video is VP9 or not?
[16:06:47 CET] <DHE> yes... ffprobe by itself prints out the same stream summary that ffmpeg does. with -show_* parameters it provides more technical and machine-readable output
[17:58:15 CET] <persina> Is it possible to take a video and specify a new width while maintaining the original aspect ratio?
[17:58:28 CET] <persina> ...or specify a new hieght...
[21:02:05 CET] <Mockarutan> Hi ppl, going mad here. Trying to encode h264 and AAC into mp4. 1mb bit rate on video, that works. Audio: 192kb, 48khz, AV_SAMPLE_FMT_FLTP. In VLC, It always just plays ~1 sec audio, like 1 image frame and nothing more. Interlaced with video or not... is this a weird compination or is it just my shit code?
[21:06:36 CET] <relaxed> persina: yes, read about the scale filter
[21:07:01 CET] <relaxed> Mockarutan: pastebin.com your command and console output
[21:11:52 CET] <Mockarutan> No, it's c++
[21:12:20 CET] <Mockarutan> Working on a render plugin for Unity.
[21:12:48 CET] <Mockarutan> I can post the source code if you want
[21:13:11 CET] <relaxed> pastebin your code and maybe someone can help
[21:20:19 CET] <Mockarutan> http://pastebin.com/YZbD8kfd
[00:00:00 CET] --- Thu Dec  1 2016


More information about the Ffmpeg-devel-irc mailing list