[Ffmpeg-devel-irc] ffmpeg.log.20191129

[burek]

[00:19:13 CET] <BeerLover> What is wrong with: ffmpeg -y -i song.mp3 -c:a libfdk_aac -vn -b:a:0 32k -profile:a aac_he tmp/song_32k.mp3 -b:a:0 128k tmp/song_128k.mp3 ?
[00:19:22 CET] <BeerLover>  Invalid audio stream. Exactly one MP3 audio stream is required.
[00:19:50 CET] <BtbN> an .mp3 file can hold exactly mp3, nothing else.
[00:20:28 CET] <Hello71> seems pretty straightforward to me
[00:21:31 CET] <Hello71> hm... I wonder if you can do -c aac x.wav
[00:22:17 CET] <Hello71> yeah, works fine
[00:22:26 CET] <furq> except you can't decode it
[00:23:23 CET] <Hello71> hm.
[00:23:27 CET] <Hello71> that does appear to be an issue
[00:23:48 CET] <Hello71> avi works fine. I thought it was the same as wav
[00:24:15 CET] <furq> they're both based on riff but they're not the same
[00:24:32 CET] <DHE> yeah for AAC maybe you want an m4a instead?
[00:24:34 CET] <Hello71> seems unnecessarily confusing
[00:24:51 CET] <furq> it's a shame you weren't working at microsoft in 1991
[00:24:55 CET] <furq> you could have saved us all a lot of bother
[00:25:08 CET] <furq> that sounds like sarcasm but it is very sincere
[00:27:39 CET] <Hello71> maybe tell them to use UTF-8 while I'm there? heh
[00:29:07 CET] <DHE> did it exist back then?
[00:29:16 CET] <DHE> NT had some unicode but...
[00:30:39 CET] <furq> 92 apparently
[00:43:00 CET] <Hello71> my point is that UTF-16 sucks
[00:53:05 CET] <furq> yes it does
[00:53:52 CET] <furq> almost as badly as UCS-2
[12:11:56 CET] <rocktop> I concatenate two mp4 videos into one  with ffmpeg concat but audio out of sync , how can I resolve this issue ?
[12:19:40 CET] <AiNA_TE> do both files have identical encoding settings
[12:19:47 CET] <AiNA_TE> what is the audio codec?
[12:19:59 CET] <rocktop> AiNA_TE: I don't know how can I chack ?
[12:20:03 CET] <rocktop> check*
[12:20:03 CET] <AiNA_TE> media info
[12:20:40 CET] <pink_mist> ffprobe, surely
[12:21:07 CET] <AiNA_TE> ffprobe works too if you want
[12:21:28 CET] <pink_mist> I mean, considering this is #ffmpeg, ffprobe seems more apropos
[12:21:42 CET] <AiNA_TE> im not a fan :P
[12:21:51 CET] <AiNA_TE> media info lays out the information better
[12:31:52 CET] <rocktop> AiNA_TE: https://bpaste.net/show/4GD6K
[12:33:27 CET] <rocktop> the only difference between both input stream audio is the first one is with 44100 Hz and second is with 48000 Hz
[13:05:26 CET] <oto313> Hi, I am building ffmpeg from source. After build I would like to install libavfilter/rtsp.h to installation folder. Is this possible?
[13:17:08 CET] <AiNA_TE> rocktop_ i would demux both audio tracks to wav
[13:17:32 CET] <AiNA_TE> ffmpeg -v quiet -stats -drc_scale 0 -y -i %input% -ac 2 -y audio1.wav
[13:17:46 CET] <AiNA_TE> then encode both files to aac, then cat them
[13:18:05 CET] <AiNA_TE> also cat the videos on their own, and then mux them back together
[13:20:45 CET] <AiNA_TE> bleh your videos have diffrent pixel formats also
[13:21:16 CET] <AiNA_TE> yuvj420p and yuv420p
[13:22:15 CET] <AiNA_TE> best to just encode both videos adding "-pix_fmt yuv420p" to your encode line
[13:23:04 CET] <AiNA_TE> ffmpeg -i video1.mp4 -vcodec libx264 -pix_fmt yuv420p -acodec aac out1.mkv
[13:23:11 CET] <AiNA_TE> ffmpeg -i video2.mp4 -vcodec libx264 -pix_fmt yuv420p -acodec aac out2.mkv
[13:23:23 CET] <AiNA_TE> then just cat out1.mkv and out2.mkv
[13:34:05 CET] <BtbN> oto313, that's not a public header, it won't get installed and you cannot use it from outside of libavfilter.
[13:36:25 CET] <oto313> I am writing application which consume rtsp stream. Can I somehow send RTSP commands like SET_PARAMETER?
[14:44:50 CET] <funnybunny2> What are planar and non-planar audio sample formats?
[14:45:09 CET] <funnybunny2> I guess non-planar is "packed"
[14:45:31 CET] <funnybunny2> (Trying to understand av_get_sample_fmt)
[14:47:36 CET] <funnybunny2> I guess planar means each audio channel's sample data is contiguous, and packed/non-planar means the sample data for each channel is interleaved
[14:48:14 CET] <funnybunny2> Please confirm!
[15:06:38 CET] <DHE> planar means that 6-channel audio (5.1 and all that) will have 6 discrete arrays in the AVFrame, one for each channel. and similar for other numbers of channel
[15:07:10 CET] <funnybunny2> DHE: Thanks
[15:07:28 CET] <funnybunny2> Is this the right channel for users of the ffmpeg API?
[15:07:33 CET] <funnybunny2> Or should I go to #ffmpeg-devel?
[15:07:58 CET] <DHE> ffmpeg-devel is for patches to ffmpeg itself. this channel is fine for user API use
[15:08:02 CET] <funnybunny2> OK
[15:09:09 CET] <funnybunny2> I think https://www.ffmpeg.org/doxygen/trunk/group__lavc__core.html#ga11f785a188d7d9df71621001465b0f1d should be updated to indicate that avcodec_open2 returns an AVERROR. I looked at the source code and it seems to be always returning AVERRORs. It's ambiguous whether it might not return one, but some arbitrary negative value instead.
[15:09:53 CET] <funnybunny2> Most functions say "returns a negative AVERROR". This one just says "Returns zero on success, a negative value on error"
[15:10:18 CET] <DHE> you can always look up the function av_strerror() to convert an error to a string
[15:12:20 CET] <funnybunny2> DHE: I am using av_make_error_string. I think av_strerror is used to get a description of the AVERROR. It's sort of vague how they're different, but I think av_make_error_string is supposed to be human-readable
[15:12:57 CET] <funnybunny2> My point is that for avcodec_open2, it's not clear from the documentation whether it returns an AVERROR, so it's not clear you use either of these functions with the return value
[15:13:25 CET] <funnybunny2> But from skimming the source of avcodec_open2, it seems to always return an AVERROR
[15:14:59 CET] <DHE> that's just the ffmpeg standard for returning errors. errno codes are sometimes used but they're not always descriptive enough
[15:16:00 CET] <funnybunny2> DHE: I just looked at the source, and av_make_error_string literally just returns av_strerror. How stupid. I guess it's for legacy purposes
[15:17:22 CET] <funnybunny2> Line 109: https://www.ffmpeg.org/doxygen/trunk/error_8h_source.html
[16:48:54 CET] <funnybunny2> Can someone give an example of when avcodec_receive_frame would return AVERROR(EAGAIN)? I'm trying to determine whether I can recover, or if I should just exit the program.
[16:49:58 CET] <funnybunny2> Or in a bigger piece of software, give up on playing the audio file
[17:13:36 CET] <Trieste> Hi, which one of the scaling algorithms would be best if I'm trying to downscale text while keeping it as sharp/legible as possible? :)
[17:27:27 CET] <Blacker47> Trieste, you can cut a small part of a video, like 10 seconds, and then try all of the algos like that: https://paste.debian.net/hidden/1c204aa0/
[17:28:39 CET] <Blacker47> Trieste, i assume bicubic or bilinear would be ok/best.
[17:29:20 CET] <Trieste> hah, did not consider that, thanks Blacker47
[17:30:26 CET] <kepstin> also try 'area' which can give good results retaining sharpness while downscaling
[17:31:35 CET] <kepstin> how much are you downscaling by?
[17:31:46 CET] <kepstin> and how big is the text? :)
[18:05:14 CET] <DHE> funnybunny2: just because you ran avcodec_send_packet() doesn't mean that avcodec_receive_frame will return a frame immediately. out-of-order decoding (B-frames) is a thing
[20:13:19 CET] <igssl> https://pastebin.com/wvVEEsAi < any clue what i did wrong? trying to convert a series of "0000"-named gifs into webms
[20:13:56 CET] <nicolas17> I don't think you can do batch conversion like that
[20:14:16 CET] <nicolas17> it will take each gif as a frame of the single output video
[20:14:47 CET] <igssl> ... there is no way to batch convert gifs in ffmpeg?
[20:15:13 CET] <nicolas17> use a script around it to run ffmpeg multiple times
[20:15:19 CET] <igssl> amazing
[20:15:21 CET] <nicolas17> oh you're on Windows... good luck
[20:40:06 CET] <pink_mist> can't he use ffmpeg to convert the gifs into single-frame images, and then take all those and make them into a webm
[20:40:43 CET] <nicolas17> pink_mist: I think he wants to convert *each* animated gif into a webm
[20:40:50 CET] <nicolas17> n gifs into n webms
[20:41:08 CET] <pink_mist> ohh
[20:41:32 CET] <pink_mist> right, then a small shellscript is the ticket
[00:00:00 CET] --- Sat Nov 30 2019