[Ffmpeg-devel-irc] ffmpeg.log.20170314

[burek]

[00:00:06 CET] <feliwir> (guessing because YUV 420 allocates 3 planes)
[00:00:08 CET] <JEEB> that's the P in YUV420P
[00:00:23 CET] <JEEB> and that's why I used + and the 2*
[00:00:36 CET] <JEEB> so you have one that is without subsampling (Y)
[00:00:47 CET] <JEEB> and then you have two that are width and height subsampled (4:2:0 chroma)
[00:01:36 CET] <feliwir> the calculation in libvp62 is like this: totalSize = pVP62->ySize + (pVP62->uvSize * 2); .I guess that makes sense
[00:05:05 CET] <feliwir> is there a way to disable inlining during configure stage? It makes it hard for me to debug some av_always_inline functions
[00:06:44 CET] <feliwir> currently i am using those settings: "./configure --enable-debug --disable-optimizations --disable-stripping
[00:06:44 CET] <feliwir> "
[00:36:30 CET] <faLUCE> "More than one AAC RDB per ADTS frame is not implemented"  <--- what can cause this problem while encoding?
[00:37:00 CET] <linuxuser9000> howdy-do ffmpeg-erinos
[03:17:45 CET] <ZeroWalker> anyone know how to use swr_convert_frame?
[06:33:05 CET] <acresearch> hey JEEB are you online?
[06:39:13 CET] <acresearch> hello people, anyone can explain to me these flags? http://paste.ubuntu.com/24175062/
[09:13:18 CET] Action: abxy hot girls http://sexygirls2000.xyz
[09:17:51 CET] <acresearch> hello people, anyone can explain to me these flags? http://paste.ubuntu.com/24175062/
[09:25:34 CET] <sile> Hi! i'm playing with filter_complex, stitching some videos together side by side. All of them last 5 seconds, I want to make some of the loop twice ( so the total duration of the video is 10 seconds ). So far I got this http://pastebin.com/WrdvuY5M, but for some reason the video stops at around the 7 seconds mark. Any clues?
[09:25:38 CET] <sile> thanks! \o/
[09:30:05 CET] <sile> woo, i fixed it by using "fifo"
[10:23:42 CET] <Drugbird> Heya, I have some trouble cross compiling ffmpeg with cuda support for ARM with the latest ffmpeg version from the git. Is this the right place to look for help for that?
[10:49:00 CET] Last message repeated 1 time(s).
[10:58:15 CET] <spaam> Drugbird: tell us about your issue. then maybe someone can help you.
[11:08:58 CET] <Drugbird> I have problems getting configure to run properly when cross compiling for arm and enabling cuda support with the error message: "ERROR: cuda requested but not found". I can configure properly for my own system (x86) properly, where cuda is properly found and the corresponding hardware decoders are enabled.
[11:11:12 CET] <fried> Hello there. In my software I want to get the metadata of each frame from a video stream (as "ffprobe ... -show_entries frame_tags=lavfi.signalstats.XYZ" it does). Unfortunately there are no meaningfull examples in this point. Anyone an idea or an example Link, how to setup a stream and iterate its frame metadata?
[11:13:12 CET] <Drugbird> Minimal example that fails: ./configure --cc="aarch64-linux-gnu-gcc" --extra-cflags="-march=armv8-a -mcpu=cortex-a57" --enable-cuda --cross-prefix=aarch64-linux-gnu- --target-os=linux --arch=aarch64 --cpu=armv8-a
[11:14:13 CET] <fried> to track ffprobe (together with cmdutil) in this point to find out is much to heavy; too many options and global dependencies/variables.
[11:16:49 CET] <fried> to track ffprobe.c (together with cmdutils.c) in this point to find out is much to heavy; too many options and global dependencies/variables.
[11:44:55 CET] <acresearch> hey JEEB are you online?
[11:45:12 CET] <acresearch> JEEB: can you explain to me these flags? http://paste.ubuntu.com/24175062/
[11:49:27 CET] <DHE> 6 and 7 are parameter and value
[13:02:35 CET] <acresearch> DHE: can you help with this? i do not understand the flags, so i can change them
[13:13:02 CET] <trapchat> whats so many people doing in here?
[13:13:13 CET] <acresearch> asking questions
[13:13:44 CET] <trapchat> sony pictures employees?
[13:13:54 CET] <acresearch> what?
[13:14:10 CET] <trapchat> what question do you have?
[13:15:02 CET] <acresearch> to explain to me these flags? http://paste.ubuntu.com/24175062/
[13:15:12 CET] <trapchat> what question does ilove11ven has?
[13:15:24 CET] <trapchat> fried: what is your question?
[13:15:41 CET] <trapchat> dont see anyone with questions
[13:16:06 CET] <fried> hello, I asked it already, but I cant see the history
[13:16:53 CET] <DHE> acresearch: https://ffmpeg.org/ffmpeg-all.html  it's a mountain, but these are the docs for the ffmpeg tool.
[13:17:18 CET] <fried> In my Software I want to access frames metadata such as "ffprobe ... -show_entries "lavfi.signalstats.<XYZ>" does
[13:17:49 CET] <trapchat> what question does ilove11ven1 has?
[13:18:25 CET] <fried> I need to know how to setup/open the stream, so that I get the metadata Dictionary for each frame
[13:19:13 CET] <fried> there is no documentation to use the lib in this pont (only ffprobe examples)
[13:19:37 CET] <trapchat> at this point
[13:20:14 CET] <trapchat> at this point in time
[13:20:28 CET] <trapchat> not in this point at time
[13:23:13 CET] <fried> what do you mean? It's a ffmpeg channel, right?!
[13:24:37 CET] <acresearch> DHE: this is a big document
[13:25:22 CET] <BtbN> fried, I'd guess it's just in the AVFrames metadata field.
[13:25:36 CET] <fried> right
[13:25:50 CET] <fried> but empty
[13:26:05 CET] <BtbN> Well, did you run it through the signalstats filter?
[13:26:17 CET] <fried> probably not set up right
[13:27:19 CET] <fried> I tried to setup via av_dict_set() before open the stream
[13:27:49 CET] <acresearch> what does -sn (output) do?
[13:28:03 CET] <BtbN> -sn disables subtitles on that output.
[13:28:10 CET] <acresearch> ohhh ok
[13:28:26 CET] <BtbN> fried, if you want signalstats, you have to actually use the signalstats filter.
[13:28:30 CET] <BtbN> They don't appear on their own
[13:29:59 CET] <fried> right, du you have an example link for me?
[13:30:06 CET] <acresearch> what does -ac 2 do?
[13:30:35 CET] <BtbN> https://ffmpeg.org/ffmpeg-filters.html#signalstats-1
[13:31:06 CET] <fried> thanks, but its for the ffprobe command line tool.
[13:31:15 CET] <BtbN> It's for the signalstats filter
[13:31:21 CET] <BtbN> It doesn't matter who invoked it
[13:31:33 CET] <BtbN> It's just a regular video filter
[13:31:39 CET] <fried> .. so I have to trace ist in the code
[13:31:47 CET] <BtbN> no?
[13:31:58 CET] <BtbN> Just run the frame through that filter, and it propagates the metadata.
[13:32:02 CET] <fried> ok, so I have to learn about the filter usage
[13:32:31 CET] <fried> ok, Ill see, thank you
[13:36:29 CET] <acresearch> what does -c:v libx264 mean?
[13:40:22 CET] <furq> codec, video, libx264
[13:44:16 CET] <acresearch> furq: oh so ffmpeg will code the output with libx264?
[13:44:54 CET] <furq> yes
[13:45:30 CET] <acresearch> so is libx264 a compression algorithm?
[13:45:55 CET] <furq> https://en.wikipedia.org/wiki/X264
[13:52:41 CET] <DHE> it's an H264 encoder
[13:53:16 CET] <acresearch> i am sorry guys, i am new to this and having a hard time understanding all the terms, H264 is HD? is there something higher?
[13:54:15 CET] <DHE> h264 is a codec. image resolution is largely irrelevant beyond the decoder's capabilities
[13:54:35 CET] <acresearch> ok
[13:54:42 CET] <acresearch> what is -preset veryfast?
[13:55:11 CET] <BtbN> Have you tried reading the documentation?
[13:55:28 CET] <acresearch> BtbN: trying to understand it, it is very LARGE
[13:55:31 CET] <acresearch> and complicated
[14:00:30 CET] <lelduayen> Hi there, I'm new to IRC so any directions on how to behave are welcome
[14:01:03 CET] <lelduayen> I'm having problems installing ffmpeg using https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu guide
[14:01:23 CET] <lelduayen> the error is "ERROR: libass not found using pkg-config"
[14:02:52 CET] <acresearch> lelduayen: are you on linux?
[14:06:30 CET] <acresearch> BtbN: what does 2 audio channels mean? (-ac 2)
[14:07:57 CET] <Kadigan_KSB> Hi. I'm seeing some weird issues with ffmpeg on Windows (nonfree) build by Kyle Schwarz (git-e18d9d9) when converting from MKV to MKV with overlay applied. The audio plays properly to the end, but the video (and the progress slider) both stop a few seconds short.
[14:08:03 CET] <Kadigan_KSB> Is this a known issue, perhaps?
[14:09:09 CET] <lelduayen> Hi acresearch, yes, ubuntu 14.04
[14:15:16 CET] <lelduayen> config.log suggest to add the directory containing `libass.pc' to the PKG_CONFIG_PATH environment variable, but no success in there
[14:27:51 CET] <lelduayen> I've pasted entire log in http://pastebin.com/wikkinHG
[14:29:32 CET] <nyuszika7h> how do I properly crop a 1.85:1 PAL DVD and scale it to 16:9 (so Display Aspect Ratio is not needed)?
[14:29:47 CET] <nyuszika7h> I'd also like to burn in the vobsubs
[14:30:11 CET] <nyuszika7h> (the MPEG-2/AC3/VobSub streams are already in an .mkv)
[14:30:51 CET] <nyuszika7h> cropdetect is giving me negative values
[14:33:10 CET] <nyuszika7h> here's a 1 minute sample if it helps: https://u.nya.is/jwmbqe.mkv
[14:34:13 CET] <nyuszika7h> oh cropdetect gave me proper values now on the sample
[14:34:34 CET] <nyuszika7h> crop=720:432:0:72
[14:38:02 CET] <acresearch> lelduayen: did you try sudo apt-get install ffmpeg?
[14:44:21 CET] <nyuszika7h> oh so it's more like 2.35:1
[14:48:49 CET] <nyuszika7h> how do I crop/scale in filter_complex?
[14:49:00 CET] <nyuszika7h> I can't do -filter_complex "[0:v:0][0:s:0]overlay[vid]" -filter:v:0 "crop=720:432:0:72,scale=720:304"
[14:51:25 CET] <nyuszika7h> okay I think I got it
[14:52:07 CET] <IntruderSRB> hey everyone, can someone share good tutorial or good documentation with use cases
[14:52:09 CET] <IntruderSRB> for AVCodec/AVParser
[14:52:24 CET] <IntruderSRB> I'm sending from my Java app to my JNIBridge C app byte arrays containing VideoNAL units
[14:52:32 CET] <IntruderSRB> and I want to use libavcodec to read information from those slices
[14:52:40 CET] <acresearch> IntruderSRB: https://ffmpeg.org/documentation.html
[14:52:43 CET] <nyuszika7h> okay, now just one more problem
[14:52:56 CET] <nyuszika7h> how do I reposition the vobsub so that it's *inside* the cropped video?
[14:53:00 CET] <nyuszika7h> hmm maybe if I overlay it after crop/scale
[14:53:35 CET] <IntruderSRB> acresearch: yes I was there but it's related to the usage of ffmpeg command line tools, failed to find library documentation on how to use it inside my source code
[14:53:48 CET] <acresearch> oh
[14:53:51 CET] <feliwir> JEEB, you see what i am doing wrong at my conversion to RGB?: https://gist.github.com/feliwir/e5da8d64f47fdde67d4656a6264f1a2a
[14:53:59 CET] <IntruderSRB> or I'm reading it wrong :/
[14:54:06 CET] <feliwir> i am getting IndexOutofBounds (so my loop is going for too long)
[14:54:57 CET] <IntruderSRB> found https://ffmpeg.org/libavcodec.html but it's not very usefull :)
[14:55:03 CET] <nyuszika7h> now the subtitles didn't get overlayed at all
[14:56:09 CET] <nyuszika7h> what's wrong with this? http://dpaste.com/06G5W5Q
[14:57:50 CET] <nyuszika7h> (I realized I should also probably downmix the audio to 2.0, but that's irrelevant now, the problem is the subtitles not showing up)
[14:58:59 CET] <nyuszika7h> http://dpaste.com/0CYV8TR here, this may be better, removed the hard newlines from the command prompt window
[14:59:26 CET] <c_14> IntruderSRB: https://ffmpeg.org/doxygen/trunk/index.html
[15:03:30 CET] <nyuszika7h> my sample file gets smaller with every encode (of course encoding from the original every time), lol
[15:05:30 CET] <IntruderSRB> c_14: thanks! there's examples ... just what I was looking for :)
[15:06:27 CET] <nyuszika7h> hmm so overlay takes x and y parameters, I'll trythat
[15:10:08 CET] <lelduayen> acresearch: i haven't tried with apt-get because of trac msg "The static builds do not support non-free encoders."
[15:11:05 CET] <lelduayen> I couldn't found which are non-free encoders, so I decided to compile instead
[15:11:18 CET] <furq> fdk-aac is the only nonfree encoder
[15:11:56 CET] <nyuszika7h> and now this happened http://t.6697.eu/g/file_22368.jpg
[15:11:59 CET] <nyuszika7h> with overlay=0:-50
[15:12:03 CET] <lelduayen> thanks furq, I going to check if I need fdk-acc
[15:15:18 CET] <luc4> Hello! I'm studying this doc http://ffmpeg.org/sample.html and I see it is possible to stream content via RTSP multicast. I used the same configuration and I can make it work. However I see the sdp can only be provided with http. Isn't it somehow possible to get it using rtsp protocol?
[15:17:47 CET] <lelduayen> I've checked and ffmpeg is not available for 14.04
[15:19:07 CET] <lelduayen> Someone can point me where is the problem?
[15:29:31 CET] <c_14> ffmpeg is only available from 16.04 up
[15:29:34 CET] <c_14> use a static build
[15:29:38 CET] <c_14> http://johnvansickle.com/ffmpeg/
[15:32:04 CET] <nyuszika7h> okay I got it
[15:34:02 CET] <nyuszika7h> but now I have another problem, it's dropping frames
[15:39:29 CET] <nyuszika7h> http://dpaste.com/1VV9ASG
[15:44:58 CET] <nyuszika7h> okay, it seems it doesn't handle the cover attachment right, it shouldn't even be a video track, I'll just tell it to ignore that
[15:45:42 CET] <nyuszika7h> perfect
[15:54:53 CET] <feliwir> any idea what this could be related to: http://imgur.com/a/HMAaG .My custom vp6 decoder decodes the first video frame as green, eventhough it should be black :(
[15:55:12 CET] <feliwir> might this be a YUV to RGB issue?
[16:05:23 CET] <atomnuker> { 0, 0, 0 } in yuv is green, not black
[17:29:13 CET] <feliwir> atomnuker, can you maybe take a look if my conversion code makes any sense: https://gist.github.com/feliwir/1f83f65e60b4ec41230a5a49c40a97c4 ?
[17:30:45 CET] <atomnuker> feliwir: does it only happen on the first frame?
[17:45:49 CET] <feliwir> hm, what exactly does this do on a short: "!!s->block_coeff[b][0]" ? I think my C# code is behaving differently there
[17:45:58 CET] <Kadigan_KSB> So... Converting from .mkv to .mkv, same a/v codec settings, applying a complex_filter overlay as 1:1 (had the image prepared for the final resolution beforehand) -- result video file has a weird issue where upon playback video progress bar finishes, the video freezes on the frame that's seen last, but audio continues for another 3-4 seconds. I am 100% sure the frame seen as "last" is in fact not the last.
[17:46:19 CET] <Kadigan_KSB> Happens on a Windows build from 2014, and on a Mac build from late 2016.
[17:46:53 CET] <Kadigan_KSB> I convert to interim file (say, prores_ks + pcm_s16le mov) and then back to .mkv w/ overlay - issue doesn't occur.
[17:47:52 CET] <Kadigan_KSB> I -may- have had "-c:a copy" in the first method, but this behavior I wouldn't have expected -- should I have?
[17:51:28 CET] <Kadigan_KSB> Hm... seems specifying audio settings fixes this. I am obviously not understanding the "copy" option.
[18:12:31 CET] <DHE> feliwir: '!!' is used to normalize boolean values. any non-zero value of is logically true such as '7', but '!!' will convert it to a standard 0 or 1. At least that's what normal C does, not sure what C# does different
[18:21:25 CET] <feliwir> thanks DHE , so values below lower than 0 are converted to 1 aswell?
[18:27:42 CET] <shincodex> #define LIBAVCODEC_VERSION_MAJOR 56 #define LIBAVCODEC_VERSION_MINOR  12 #define LIBAVCODEC_VERSION_MICRO 101
[18:27:51 CET] <shincodex> h264 pixelation big time
[18:27:54 CET] <shincodex> Yet vlc dont do this
[18:28:02 CET] <shincodex> Default h264 codec on ffmpeg decoding
[18:28:14 CET] <shincodex> vlc i think default libx264 anyone confirm?
[18:28:23 CET] <BtbN> x264 is an encoder.
[18:28:25 CET] <shincodex> Pixelation scanline mirroing
[18:28:33 CET] <BtbN> vlc uses lavc to decode h264.
[18:28:41 CET] <shincodex> when ytou say LAVc
[18:28:53 CET] <bencoh> LibAVCodec
[18:28:58 CET] <shincodex> you mean -lavcodec
[18:29:07 CET] <shincodex> ok
[18:29:11 CET] <shincodex> I think you said this before
[18:29:15 CET] <shincodex> and I guessed right
[18:30:01 CET] <shincodex> So your saying I dont encode with same file h264.c
[18:30:10 CET] <shincodex> but rather libx264 is the encoder
[18:30:29 CET] <shincodex> so if i try to encode without configure of libx264.. then i wont encode squat
[18:30:54 CET] <BtbN> It will probably default to some poor default quality
[18:32:20 CET] <shincodex> I have tested
[18:32:23 CET] <shincodex> options all over
[18:32:29 CET] <shincodex> from profile main baseline High
[18:32:31 CET] <shincodex> High 10
[18:32:56 CET] <shincodex> FF_PROFILE_H264_BASELINE
[18:33:00 CET] <shincodex> FF_PROFILE_H264_HIGH_10
[18:33:13 CET] <shincodex> No matter what i set
[18:33:14 CET] <shincodex> int ff_h264_get_profile(SPS *sps)
[18:33:19 CET] <shincodex> always returns 77 i think
[18:33:20 CET] <shincodex> which stands for
[18:33:29 CET] <shincodex> #define FF_PROFILE_H264_MAIN                 77
[18:33:56 CET] <shincodex> if (avcodec_open2(codecContext, codec, 0) > -1)
[18:34:02 CET] <shincodex> i set options passed it where you see the zero there
[18:34:05 CET] <shincodex> I set it in priv_data
[18:34:15 CET] <shincodex> and part of int err = avformat_open_input(&formatContext, openPath.c_str(), inputFormat, &options);
[18:34:16 CET] <shincodex> what gives
[18:38:18 CET] <DHE> feliwir: 0 is false. all other values are true. but for normalization '1' is preferred as true.
[18:39:19 CET] <brabo> hello, i am trying to make qsv work, but ffmpeg keeps segfaulting when i try to use either h264_qsv/hevc_qsv/mpeg2_qsv and the debug output is not very helpful, are there known issues with qsv in ffmpeg?
[18:40:04 CET] <brabo> note, i am atm using intels mediasdk, not https://github.com/lu-zero/mfx_dispatch
[18:50:05 CET] <feliwir> thanks DHE
[19:02:37 CET] Action: brabo oggles https://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=1f26a231bb065276cd80ce02957c759f3197edfa
[19:14:46 CET] <jkqxz> brabo:  Segfaulting early generally indicates that you missed something in the installation (libmfx is not very robust to that).
[19:16:03 CET] <jkqxz> You need the modified libva and libdrm (and to not allow your distribution to override them with newer vanilla versions), and you need to have built the specific old kernel it requires with all 9001 patches applied.
[19:17:21 CET] <brabo> jkqxz: going to try out that commit now, the ffmpeg repo we were using was missing that one, we have the libs in place and stuff so that should all be okay
[19:17:42 CET] <brabo> the vlc qsv plugin succesfully detects the hardware, and the sample tests of the media sdk all go perfect
[19:18:17 CET] <brabo> but vlc either also segfaults (maybe same commit will help) or the fifo buffer gets filled and cleared on vlc side
[19:18:21 CET] <shincodex> [NULL @ 05962E40] non-existing SPS 1 referenced in buffering period [h264 @ 0A30AB20] non-existing SPS 1 referenced in buffering period [h264 @ 0A30AB20] Invalid POC 2<8 [NULL @ 05962E40] RTP: missed 70 packets [h264 @ 0A38EB40] left block unavailable for requested intra4x4 mode -1 at 0 41 [h264 @ 0A38EB40] error while decoding MB 0 41, bytestream 26979
[19:18:54 CET] <brabo> since the vlc qsv plugin is not to be considered working, but ffmpeg is, i'm first trying to make ffmpeg work
[19:21:26 CET] <brabo> then ofc using the intel sdk vs the lu-zero mfx_dispatch may cause problems as well, will have to test that too if it keeps failing
[20:41:41 CET] <kn0w> codecs fscc for catancia debian ?
[20:41:56 CET] <kn0w> tscc (camtasia) -> h264 (libx264))  << for Linux ? codecs ?
[20:42:18 CET] <kn0w> formt Video.dm2 << ?
[21:00:11 CET] <FFmpegFTL> Hey guys ... I am currently trying to get a RTSP server working with ffmpeg command as mentioned here (https://lists.ffmpeg.org/pipermail/ffmpeg-user/2013-April/014726.html). The problem is that when a client connects, the server stops with no error message
[21:01:09 CET] <FFmpegFTL> When a client connects to the server, ffmpeg has the following output :
[21:02:15 CET] <FFmpegFTL> [rtsp @ 0000000001be26a0] Host myHostname differs from expected 0.0.0.0     Last message repeated 1 times [rtsp @ 0000000001be26a0] Updating control URI to rtsp://myHostname:8080/test0 [udp @ 000000000048ce60] 'circular_buffer_size' option was set but it is not supported on this build (pthread support is required) [udp @ 0000000001bf41a0] 'circular_buffer_size' option was set but it is not supported on this build (pthread support 
[21:02:44 CET] <FFmpegFTL> [rtsp @ 0000000001be26a0] Could not find codec parameters for stream 0 (Video: h264, none): unspecified size Consider increasing the value for the 'analyzeduration' and 'probesize' options
[21:23:50 CET] <FFmpegFTL> Here is the pastebin link for the error log
[21:23:52 CET] <FFmpegFTL> http://pastebin.com/jgsapit1
[21:27:07 CET] <alexpigment> FFmpegFTL: have you seen this thread? https://github.com/ZoneMinder/ZoneMinder/issues/811
[21:28:31 CET] <alexpigment> actually, disregard that. it came up with a google search but that error message is not explicitly mentioned - it just has "circular buffer" a few times in the article
[21:40:28 CET] <FFmpegFTL> The question is, is that message ('circular_buffer_size' option was set but it is not supported on this build (pthread support is required)) an error ?
[21:40:35 CET] <FFmpegFTL> Or is it a warning ?
[21:40:46 CET] <FFmpegFTL> It has a yellow color
[21:41:13 CET] <alexpigment> i believe it's just a warning
[21:41:16 CET] <alexpigment> here's more info here; http://ffmpeg-devel.ffmpeg.narkive.com/cxOIY8TZ/patch-lavf-udp-warn-in-case-the-circular-buffer-size-or-overrun-nonfatal-cannot-be-supported
[21:41:46 CET] <alexpigment> now for the other issue, i'm not entirely sure, it's related, but this thread has the exact same error: https://github.com/datarhei/restreamer/issues/3
[21:42:24 CET] <alexpigment> see the second post from "yangpu" in that thread
[21:44:14 CET] <c_14> yellow is warning
[21:53:56 CET] <shincodex> vlc say haha fool we no use lavc it sucks we use gpu
[21:54:03 CET] <shincodex> but that dont help solve anything
[21:54:13 CET] <shincodex> he might even just be troll idk
[21:54:23 CET] <shincodex> Maybe my configure script to anal
[21:54:28 CET] <shincodex> I disable post proc
[21:54:42 CET] <shincodex> so now Im just going to bernie sanders the configure and hope
[22:00:06 CET] <FFmpegFTL> Okay. I dont see how this should help alexpigment
[22:00:25 CET] <FFmpegFTL> This guy tries to grab a stream from some source
[22:01:09 CET] <alexpigment> honestly, i don't know. you asked a question and it was crickets in here. I was googling around to help spark some information that might relate to your problem. I presume you've already googled around as well?
[22:02:17 CET] <faLUCE> I still believe that the decision of separating the ADTS header from the AAC encoder, in ffmpeg 3.2,  is a wrong decision. ADTS works only with AAC, then there's no reason for decoupling them. And in order to write a playable AAC file you have to mux it, which increases complexity of the code. Why don't  you do a regression in the source code of libav?
[22:02:42 CET] <shincodex> googled around is what sluts do
[22:02:45 CET] <shincodex> or do they?
[22:03:06 CET] <JEEB> faLUCE: if you care that much about enabling the ADTS muxer, then please report that as an issue on the trac
[22:03:09 CET] <JEEB> kthx
[22:03:16 CET] <shincodex> its a jeeb
[22:03:19 CET] <JEEB> will get more caring eyes than on this IRC channel
[22:03:42 CET] <FFmpegFTL> yes I did. I dont find any solution
[22:03:54 CET] <shincodex> so since my version to know i noticed that they changed source code names.. h264.c h264dec.c
[22:04:09 CET] <faLUCE> JEEB: yes, I'll do that
[22:04:19 CET] <SpeakerToMeat> Ok not here
[22:04:31 CET] <faLUCE> but I'm asking for IRC people's opinion
[22:06:07 CET] <JEEB> in theory I do actually agree with things becoming more modular instead of automagically enabling things, if you are doing disable-everything builds. but that's just me so go on and post an issue
[22:06:39 CET] <faLUCE> JEEB: yes, but then MP2 should be changed in the same way, for choerency
[22:07:13 CET] <JEEB> sure
[22:07:32 CET] <faLUCE> JEEB: currently, there's not a choerency
[22:07:48 CET] <JEEB> but thankfully I'm one of those people who doesn't do disable-everything builds so fortunately idgaf
[22:08:02 CET] <FFmpegFTL> What about the "bind failed: Error number -10048 occurred" error alexpigment
[22:09:44 CET] <shincodex> hue hue hu ^
[22:09:50 CET] <furq> https://github.com/FFmpeg/FFmpeg/blob/master/libavformat/adtsenc.c#L102-L103
[22:09:51 CET] <furq> oh man
[22:09:53 CET] <shincodex> i had a disable everything build
[22:09:54 CET] <furq> don't tell me this is a thing
[22:10:02 CET] <shincodex> but enable what i need
[22:10:06 CET] <JEEB> furq: unfortunately it fucking is
[22:10:10 CET] <shincodex> now im going other way to see what the shit is going on
[22:10:11 CET] <furq> ;_;
[22:10:20 CET] <furq> as if id3v2 hasn't done enough damage
[22:10:24 CET] <JEEB> because awful people somehow think it's a good idea to mux without container
[22:10:31 CET] <JEEB> and still want metadata
[22:11:08 CET] <furq> the only argument against using m4a as an audio container is not one that's resolved by using id3
[22:11:33 CET] <shincodex> --disable-libopus there needs to be something for this
[22:11:41 CET] <shincodex> so i dont have to disable-everything
[22:12:14 CET] <JEEB> exceot by default most libs aren't enabled... (except some). in which case I agree, there should be coherency with it :P
[22:12:24 CET] <shincodex> you know whats funny
[22:12:24 CET] <JEEB> either you autodetect things, or you don't
[22:12:44 CET] <shincodex> --enable-libopus         enable Opus de/encoding via libopus [no]
[22:12:47 CET] <shincodex> i assume no means
[22:12:48 CET] <shincodex> no
[22:13:11 CET] <shincodex> also if it helps
[22:13:11 CET] <JEEB> which means the library isn't enabled even if found, unless that doc is fucked
[22:13:11 CET] <shincodex> --toolchain=msvc
[22:13:26 CET] <shincodex> Snapshot zip file i downloaded
[22:13:28 CET] <JEEB> there's now an internal opus decoder and encoder, just fyi :P
[22:13:30 CET] <shincodex> hoping it worked i did
[22:13:31 CET] <furq> are external decoders/encoders ever enabled by default
[22:13:42 CET] <JEEB> some random libs are
[22:13:47 CET] <JEEB> zlib being one
[22:13:48 CET] <furq> i thought that was only ever for devices
[22:13:55 CET] <furq> and miscellaneous libs like zlib
[22:14:04 CET] <JEEB> anyways, it's still lulzy
[22:14:05 CET] <shincodex> libavcodec.a(opusdec.o) : error LNK2019
[22:14:11 CET] <shincodex> _swr_alloc referenced in function _opus_decode_init
[22:14:12 CET] <shincodex> so
[22:14:16 CET] <shincodex> Its ignoring me
[22:14:21 CET] <JEEB> shincodex: no, that's the internal one
[22:14:22 CET] <furq> that's the internal one
[22:14:51 CET] <shincodex> oh oh
[22:14:53 CET] <JEEB> having a second decoder was a requirement by European Broadcast Union
[22:14:55 CET] <shincodex> Wait a minute your saying...
[22:15:04 CET] <JEEB> so we got opus standardized
[22:15:07 CET] <JEEB> in le broadcast
[22:15:16 CET] <shincodex> im not linking swsresample
[22:15:35 CET] <furq> well you probably should
[22:15:36 CET] <JEEB> yea, then that thing might be broken in your random snapshot
[22:15:44 CET] <shincodex> Well
[22:15:51 CET] <JEEB> since it should have brought swresample in or disabled
[22:16:02 CET] <shincodex> normally only link in my compact build codec util scale format
[22:16:14 CET] <shincodex> i forgot that when i turn everything on it want everything
[22:16:49 CET] <JEEB> unless you fucked around with the configure script and all, it should have not enabled the opus decoder in case swresample wasn't being built
[22:16:58 CET] <shincodex> Negative
[22:16:59 CET] <JEEB> or that was a bug
[22:17:07 CET] <shincodex> this configure script is pure from my fingers
[22:17:10 CET] <shincodex> so far
[22:17:26 CET] <JEEB> anyways, if it still happens with current HEAD report it, that sounds like a booger
[22:17:36 CET] <JEEB> if you have disabled swresample and it still tries to enable shit requiring it
[22:18:11 CET] <shincodex> what is a latest safe build
[22:18:13 CET] <shincodex> :P
[22:18:41 CET] <JEEB> if there's any similar configuration to yours on FATE, then the last passing one there, but I have a feeling we don't have --disable-swresample build :P
[22:18:46 CET] <shincodex> https://github.com/FFmpeg/FFmpeg.git
[22:18:49 CET] <shincodex> is this safe?
[22:19:08 CET] <FFmpegFTL> yes it is
[22:19:16 CET] <JEEB> https://git.videolan.org/git/ffmpeg.git
[22:19:23 CET] <FFmpegFTL> its just a clone of the regular ffmpeg repo
[22:19:24 CET] <JEEB> the github if ever gets sync'd manually
[22:19:31 CET] <JEEB> so always use the videolan one
[22:19:55 CET] <shincodex> http://pastebin.com/kd16t2bv
[22:20:01 CET] <shincodex> that my configure for debooger
[22:20:19 CET] <shincodex> ignore opus disable
[22:20:21 CET] <JEEB> oh, we actually did have disable-swresample FATE instance
[22:20:32 CET] <shincodex> #--disable-everything --disable-sdl --disable-avdevice --disable-swresample --disable-postproc --disable-avfilter \
[22:20:34 CET] <JEEB> last run Feb 25th
[22:20:35 CET] <JEEB> http://fate.ffmpeg.org/report.cgi?time=20170225125446&slot=x86_64-archlinux-gcc-disableswresample
[22:20:39 CET] <shincodex> i comment that out part of my line
[22:21:05 CET] <shincodex> that was part of my configure just testing with bare minimum to try and solve h264 problems
[22:21:46 CET] <JEEB> wait you don't even do disable-swresample?!
[22:21:47 CET] <JEEB> wtf
[22:21:55 CET] <shincodex> I did
[22:21:57 CET] <JEEB> oh, it's in your IRC only :P
[22:22:00 CET] <shincodex> should i put it back
[22:22:00 CET] <JEEB> not in the pastebin
[22:22:01 CET] <shincodex> lol
[22:22:11 CET] <shincodex> you see where that fake disable-opus is
[22:22:16 CET] <shincodex> thats where the # i posted just now is
[22:22:19 CET] <shincodex> was
[22:22:33 CET] <shincodex> I dont need device
[22:22:40 CET] <shincodex> i dont do anything with web cams scanners or those things
[22:25:15 CET] <FFmpegFTL> So hopefully someone could help me. I am trying to stream with OBS to an RTSP server, created by ffmpeg. The problem is that as far as ffmpeg picks the stream from the incoming connection its stops almost immediately
[22:25:39 CET] <FFmpegFTL> My console in and output is this
[22:25:40 CET] <FFmpegFTL> http://pastebin.com/eeVUPDn5
[22:26:18 CET] <JEEB> shincodex: at least the configure script's output had the opus decoder disappear after disable-swresample
[22:26:32 CET] <JEEB> lessee if the libs build
[22:27:28 CET] <BtbN> ffmpeg can act as rtmp server?
[22:27:48 CET] <FFmpegFTL> yes
[22:27:51 CET] <furq> in a sense
[22:27:54 CET] <furq> in a more real sense, no
[22:28:25 CET] <FFmpegFTL> but u have to build it without --enable-librtmp
[22:28:39 CET] <BtbN> Well... don't do that I'd say
[22:28:54 CET] <BtbN> Use a proper rtmp server
[22:29:22 CET] <furq> what he said
[22:30:32 CET] <shincodex> does ur name mean
[22:30:36 CET] <shincodex> ffmpegforthelose
[22:30:42 CET] <shincodex> or for the loose?
[22:49:10 CET] <SpeakerToMeat> Hello all.
[22:51:21 CET] <SpeakerToMeat> Couple questions, I need to convert three sets of image sequences to video and change colorspace. a) the concat demuxer file list in -i mylist.txt accepts the "directory/file%06D.png" notation in it? (do I need to explicitly set -f image2 ? will this conflict with -f concat ?)
[22:52:08 CET] <furq> probably, no, and no
[22:52:27 CET] <SpeakerToMeat> 2) Hmmm maybe I should split this in 2 parts, first convert each set of images and 5.1 audio channels into a video, and then concat all the videos.
[22:52:35 CET] <furq> although it'd be simpler to just use the concat filter
[22:53:32 CET] <SpeakerToMeat> Yes, I have 3 sets of images, and 3 sets of 6 single channels wav files, I need to concat, color space change and map all of those :/ it might be easier to do it in blocks, convert, color and map each set, and then concat the three results
[22:53:49 CET] <furq> you can do all that in one command
[22:54:09 CET] <SpeakerToMeat> Yes but it'll turn into a hairy (if fun) command... let's see I'll see the different parts and see if I can manage that
[22:54:16 CET] <c_14> 3 image2 demuxers, 6 wav demuxers, a concat filter and an amerge
[22:54:19 CET] <c_14> shouldn't be that hard
[22:54:25 CET] <furq> you might want to combine the wavs first just to make the command less horrible
[22:54:39 CET] <furq> but after that, it's easy enough
[22:55:13 CET] <SpeakerToMeat> Ok for the filter, I'd need to use a filter_complex right?
[22:55:25 CET] <c_14> yes
[22:55:47 CET] <SpeakerToMeat> furq: Yes I think I'll join the wavs into multichanel single files, that'll shorten the map notably wouldn't it?
[22:55:58 CET] <furq> -i a%d.png -i b%d.png -i c%d.png -i a.wav -i b.wav -i c.wav -filter_complex "concat=n=3:v=1:a=1[out]" -pix_fmt yuv420p -map "[out]"
[22:56:19 CET] <SpeakerToMeat> Thanks I'll use that as a guide
[22:56:28 CET] <SpeakerToMeat> let me sox these wavs first
[22:56:29 CET] <furq> you might need to add the inputs to the start of filter_complex
[22:57:00 CET] <c_14> you will have to add the inputs (or sort the -i differently)
[22:57:16 CET] <c_14> since concat needs video,audio,video,audio and that'll pass video,video,video,audio,audio,audio
[22:57:37 CET] <SpeakerToMeat> the inputs in this style: [0:v:0] right? for the audio, if it's 6 channel, how would I do, [0:a] ?
[22:57:50 CET] <c_14> Just 1:a
[22:57:53 CET] <furq> sounds easier to just sort the inputs like c_14 said
[22:58:01 CET] <SpeakerToMeat> yes
[22:58:25 CET] <SpeakerToMeat> Now let me clear some HDD space, throw XYZ->RGB on that mix and try it
[22:58:32 CET] <SpeakerToMeat> Thanks furq and c_14
[23:02:39 CET] <SpeakerToMeat> Sorry how was it I could show the pixel formats, and options for a specific encoder?
[23:02:53 CET] <c_14> ffmpeg -h encoder=
[23:03:14 CET] <SpeakerToMeat> Thanks
[23:10:52 CET] <SpeakerToMeat> Aaaand, last thing, sorry how was it to make a "center cut" transform? material is in 2048x1080 with black bands up and down... I... I did this the other day and lost the command and forgot.... let me google.
[23:11:19 CET] <furq> !filter crop
[23:11:19 CET] <nfobot> furq: http://ffmpeg.org/ffmpeg-filters.html#crop
[23:12:14 CET] <SpeakerToMeat> I can add this as an extra filter after the concat, right? not to touch the complex filter... or should I glue it onto the complex filter
[23:12:30 CET] <furq> you can't specify more than one filterchain
[23:12:44 CET] <furq> just add ,crop=... before [out]
[23:12:54 CET] <SpeakerToMeat> Ok, so I need to add it to the complex filter.
[23:13:10 CET] <SpeakerToMeat> I will, thanks
[23:36:57 CET] <feliwir> any idea where such a pattern might come from: http://imgur.com/RDJa8VA
[23:37:06 CET] <feliwir> this image is supposed to be completly black
[23:37:52 CET] <SpeakerToMeat> The x and y for crop or similar filters are indexed from 0 or 1?
[23:38:04 CET] <c_14> SpeakerToMeat: 0, always
[23:38:15 CET] <feliwir> hehe it looks like a optical illusion :P
[23:40:17 CET] <SpeakerToMeat> c_14: Sorry to ask directly, do you have any experience with prores (in quicktime muxing)?
[23:40:37 CET] <c_14> not really, why?
[23:41:12 CET] <SpeakerToMeat> c_14: I know PCM audio to be used in mov with prores must be 16 bits, but I'm unsude if little endian or big endian, signed or unsigned
[23:43:06 CET] <c_14> signed, but I don't think endianess matters
[23:43:13 CET] <SpeakerToMeat> Ok, thanks
[23:45:40 CET] <SpeakerToMeat> When using -i file%08d.png is there any way to overcome an error if the numbering doesn't start at 00000000 ?
[23:45:51 CET] <c_14> -start_number
[23:46:17 CET] <SpeakerToMeat> c_14: Right after/before the corresponding -i ?
[23:46:21 CET] <c_14> before
[23:46:42 CET] <c_14> options always affect the "file" they're in front of
[23:46:48 CET] <SpeakerToMeat> And it wont affect other previous -i
[23:46:51 CET] <SpeakerToMeat> ok
[23:48:09 CET] <SpeakerToMeat> Hmmm
[23:48:11 CET] <SpeakerToMeat> Too many inputs specified for the "crop" filter
[23:48:32 CET] <SpeakerToMeat> -filter_complex "concat=n=2:v=1:a=1,crop=2048:858:0:111[out]"
[23:48:41 CET] <c_14> you can't put the crop filter right after the concat
[23:48:46 CET] <c_14> because crop doesn't accept audio
[23:49:06 CET] <c_14> do concat=n=2:v=1:a=1[vout][aout];[vout]crop[out] -map [out] -map [aout]
[23:49:06 CET] <SpeakerToMeat> hmmm ok... so I need the crop... before the concat... will it affect all inputs anyhow?
[23:49:17 CET] <SpeakerToMeat> sure
[23:50:15 CET] <SpeakerToMeat> And, working, thanks for all the toil c_14
[23:50:28 CET] <SpeakerToMeat> Filter language always sacres me, someday I ought to learn it more deeply
[23:52:01 CET] <c_14> you get used to it
[23:52:20 CET] <c_14> especially once you've started debugging filtergraphs spanning >10 lines
[23:52:27 CET] <SpeakerToMeat> holy
[23:52:50 CET] <SpeakerToMeat> Do you debug them the old fashioned way, or do you use tools like graphviz to see the structure?
[23:53:00 CET] <JEEB> at that point you might want to think about running your own libav* API user
[23:53:11 CET] <JEEB> instead of doing it with filter_complex :D
[23:53:20 CET] <SpeakerToMeat> Yeah
[23:54:05 CET] <SpeakerToMeat> It reminds me of going from 10 component SQL calls to 20 line PL/SQL monstrosities
[23:57:27 CET] <SpeakerToMeat> Hmmm prores_ks profile hq should be using yuv422 but it's using yuv444
[23:57:36 CET] <SpeakerToMeat> Maybe I'll specify the pixel format by hand
[23:57:55 CET] <JEEB> default picked usually depends on your source format
[23:58:13 CET] <JEEB> prores has both 4:2:2 and 4:4:4 modes
[23:58:25 CET] <SpeakerToMeat> Yeah, thing is -h encoder=prores_ks didn't list valid pixel formats (or I missed it) so I was wary at hand picking one
[00:00:00 CET] --- Wed Mar 15 2017