[Ffmpeg-devel-irc] ffmpeg.log.20150527

[burek]

[00:01:19 CEST] <moontails> llogan: thank you! I will take a look at those
[00:02:03 CEST] <llogan> moontails: maybe you could implement something with ffprobe -show_frames
[00:09:03 CEST] <moontails> thank you!
[00:09:18 CEST] <moontails> you have given me two starting pointers for me to research
[00:09:24 CEST] <moontails> i have never used ffmpeg before
[00:09:30 CEST] <moontails> so will need to read up on those
[00:12:46 CEST] <llogan> i don't know of a current method with ffmpeg/ffprobe to give you volume info per frame, excluding something like "ffmpeg -i input -af aselect=<whatever option>,astats -f null -" but that will only give info for the particular selected frame/whatever
[01:18:46 CEST] <jeffshanab> When I specify --prefix on configure, the very first message I see is "Configured with: --prefix=/Applications/Xcode.app/Contents/Developer/usr --with-gxx-include-dir=/usr/include/c++/4.2.1"   Did configure ignore the --prefix option?
[02:23:19 CEST] <incal> avconv -i concat:file_1.mp3\|...\|file_n.mp3 -c copy out.mp3 says
[02:23:19 CEST] <incal> 'could not find codec parameters'?
[02:25:27 CEST] <incal> I'm on Debian
[02:26:06 CEST] <relaxed> incal: ask in #libav
[02:56:09 CEST] <Hello71> but i'll bet you $5 that your libav doesn't support mp3
[10:57:05 CEST] <user890104> hello, is it possible to capture video using x11grab from a secondary X11 server that is currently inactive?
[10:59:24 CEST] <user890104> i can see the TCP traffic (using 127.0.0.1:6001) in Wireshark, but it looks like the image is not rendered so it can be captured
[11:58:01 CEST] <chama> Hi I have an issue with amerge functionality. I need it to append each audio to the end of final output audio. but insted of appending it overlaps the audio files....
[11:58:17 CEST] <chama> anyone have any idea?
[11:58:20 CEST] <c_14> If you want to append, concat.
[12:00:16 CEST] <chama> no i can not concat  because those audio files are crossfaded and then after only i need to join each...
[12:00:59 CEST] <c_14> concat n seconds of silence to the beginning of the audio files representing their offset to each other and then amerge
[12:01:32 CEST] <chama> here is the complete issue
[12:01:33 CEST] <chama> http://stackoverflow.com/questions/30395469/ffmpeg-audio-crossfade
[12:05:12 CEST] <c_14> concat n seconds of silence to the beginning of the audio files representing their offset to each other and then amix them all together
[12:08:00 CEST] <chama> I'm sorry i didn't get it. can you please provide a simple example for these steps...
[12:08:09 CEST] <chama> I'm new to ffmpeg
[12:11:23 CEST] <c_14> assuming you have 2 audio files and the first is 3 minutes long stereo and you want to crossfade 5 seconds, then `-filter_complex "[0:a]afade=t=out:st=175:d=5[a0];[1:a]afade=t=in:st=0:d=5[t0];aevalsrc=0:d=175[s];[s][t0]concat=n=2:v=0:a=1[a1];[a0][a1]amix[a]"
[12:11:31 CEST] <c_14> ' should do it
[12:11:59 CEST] <c_14> eh, aevalsrc=0:d=15:c=2 of course
[12:13:13 CEST] <Anoia> nice and discoverable :p
[12:13:40 CEST] <chama> ok thanks a lot.... I'll try that :)
[12:20:22 CEST] <intfloat32> Hello everyone,
[12:20:22 CEST] <intfloat32> I'm creating a mix of 3 files via amix. The problem is that the outputfile is significantly quieter than expected. I guess it's because it uses normalization to avoid clipping. This is my CLI http://pastebin.com/srVLTQHU . If it's quieter because of normalization, I would like to know if there is a way to mix the files using linear/logarithmic rang
[12:20:22 CEST] <intfloat32> e compression?
[12:21:04 CEST] <chama> thanks c_14 it worked like a charm... thanks a lot. you saved my time :)
[12:25:15 CEST] <user890104> hello again, reposting my question: is it possible to capture video using x11grab from a secondary X11 server that is currently inactive? i can see the TCP traffic (using 127.0.0.1:6001) in Wireshark, but it looks like the image is not rendered so it can be captured
[13:49:31 CEST] <top_4> Hello everyone,
[13:49:32 CEST] <top_4> I'm creating a mix of 3 files via amix. The problem is that the outputfile is significantly quieter than expected. I guess it's because it uses normalization to avoid clipping. This is my CLI http://pastebin.com/srVLTQHU . If it's quieter because of normalization, I would like to know if there is a way to mix the files using dynamic range compressi
[13:49:32 CEST] <top_4> on?
[15:10:51 CEST] <top_4> I think i'll try it with the compand filter
[15:26:31 CEST] <xreal> Is there a function in ffmpeg to download the best HLS stream in a playlist?
[15:33:38 CEST] <mmm_> hello. it is possible to set TAG_SETBACKGROUNDCOLOR on swf output?
[15:48:49 CEST] <xreal> mmm_: I think, we have to wait some hours to get an answer ;)
[16:29:54 CEST] <animax> hi all. I'm still concerned with encoding png sequences. can anyone tell me what's the difference between -framerate, -r and -fps? I'm confused about it. is it sufficient to use -r when it comes to set different input and output framerates? is the -framerate option obsolete? I don't find it in the ffmpeg help. and when do I have to take '-f image2'? it seems to me that 'ffmpeg -i img%04d.png' alwa
[16:29:55 CEST] <animax> ys is enough to point to the input files.
[16:31:49 CEST] <animax> working on windows
[18:14:23 CEST] <mmm_> xreal: oh, boy :( for me no problem. but after searching and reading source code might be not possible
[18:14:43 CEST] <xreal> Is there a function in ffmpeg to download the "best" HLS stream in a playlist?
[19:11:39 CEST] <moontails> Hello everyone
[19:12:13 CEST] <moontails> I am trying to extract the volume level frame by frame from a video file. I am new to ffmpeg and so wanted to know if anyone has done something like this
[19:12:31 CEST] <BtbN> audio doesn't realy have frames
[19:13:03 CEST] <xreal> Is there a function in ffmpeg to select a stream in a playlist?
[19:13:27 CEST] <moontails> I mean can i get the corresponding audio level for a frame in the video?
[19:25:23 CEST] <Mavrik-> moontails, ah... digital video doesn't really work like that
[19:27:12 CEST] <Mavrik> and I think ffmpeg doesn't even have a filter to show you that information on a time-window basis :/
[19:32:27 CEST] <lordkron1or> I think if you want to measure audio levels, such as loudness r128gain might be worth looking into... never tested it myself
[19:33:56 CEST] <lordkron1or> also, not part of ffmpeg
[19:46:18 CEST] <ChocolateArmpits> https://www.ffmpeg.org/ffmpeg-all.html#ebur128 and https://www.ffmpeg.org/ffmpeg-all.html#replaygain
[20:28:48 CEST] <xreal> I'm streaming an HLS stream to file. Why does ffmpeg write it own tags, when using -map and -c copy
[20:28:59 CEST] <xreal> Is there a way to stop this?
[20:35:42 CEST] <ChocolateArmpits> Because it's rewrapping it
[20:36:24 CEST] <ChocolateArmpits> And just so happens the library that's being used to rewrap has a name that self-insert
[20:36:29 CEST] <ChocolateArmpits> s
[20:38:21 CEST] <xreal> ChocolateArmpits: Can I turn of re-tagging?
[20:39:27 CEST] <BtbN> What re-tagging? You mean the version it puts in there, because it muxed the file?
[20:39:35 CEST] <xreal> BtbN: yep.
[20:39:45 CEST] <xreal> Right now, I need to use "livestreamer" because it doesn't write anything new to the tags.
[20:39:46 CEST] <BtbN> That's not re-tagging. It created that file.
[20:40:03 CEST] <BtbN> So writing itself as muxer in there is correct
[20:40:08 CEST] <xreal> BtbN: is there a way to make ffmpeg not write creation tags?
[20:40:19 CEST] <BtbN> Why is that a problem?
[20:40:46 CEST] <xreal> BtbN: I don't want it in the final files.
[20:43:23 CEST] <ChocolateArmpits> xreal: That's not possible with a default build
[20:43:35 CEST] <ChocolateArmpits> Or at least the documentation doesn't disclose
[20:43:35 CEST] <xreal> ChocolateArmpits: okay, then I'll compile it on my own.
[20:44:18 CEST] <ChocolateArmpits> You could try editing the file in a hex editor if you find the line that corresponds to the text used
[20:44:31 CEST] <ChocolateArmpits> But I'm only guessing
[20:45:08 CEST] <xreal> ChocolateArmpits: perhaps some checksums might go bad then? :D
[20:45:19 CEST] <xreal> ChocolateArmpits: it's easier to remove it from sourcecode and recompile
[20:45:36 CEST] <ChocolateArmpits> That depends on use case
[20:45:50 CEST] <ChocolateArmpits> But do as you see fit
[20:45:52 CEST] <c_14> xreal: I think you can add -flags +bitexact (or was it -fflags +bitexact), that should get rid of most of those
[20:46:23 CEST] <xreal> c_14: I'll try soon.
[20:46:53 CEST] <xreal> is it possible to select a specific stream from a playlist? I've got to use -map right now to "mix" the correct audio and video.
[20:47:12 CEST] <xreal> The playlist has several streams in different resolutions and qualities.
[20:47:30 CEST] <ChocolateArmpits> You mean switching dynamically ?
[20:48:11 CEST] <ChocolateArmpits> oh
[20:48:24 CEST] <xreal> ChocolateArmpits: No, I'd like to say "stream with the highest quality" for example.
[20:48:33 CEST] <xreal> ChocolateArmpits: or "the 3rd stream from the playlist".
[20:49:02 CEST] <ChocolateArmpits> xreal: try livestreamer and pipe it's output to ffmpeg
[20:49:22 CEST] <xreal> ChocolateArmpits: what do I need ffmpeg for then? :)
[20:49:31 CEST] <xreal> ChocolateArmpits: livestreams does what I want, but I like ffmpeg.
[20:49:45 CEST] <xreal> livestreamer*
[20:50:20 CEST] <ChocolateArmpits> ok I forgot it can output to a file
[20:50:43 CEST] <xreal> ChocolateArmpits: Actually, it works in ffmpeg, but not as elagant as with livestreamer.
[20:50:50 CEST] <xreal> elegant*
[20:51:13 CEST] <ChocolateArmpits> What do you mean elegant? doesn't get more barebones than a single command
[20:51:18 CEST] <ChocolateArmpits> parameter *
[20:51:55 CEST] <xreal> ChocolateArmpits: "livestreamer <url> best" vs. "ffmpeg -i <url> -map 0:6 -map 0:4 -c copy"
[20:52:21 CEST] <ChocolateArmpits> I don't see your point
[20:52:22 CEST] <ChocolateArmpits> sorry
[20:52:44 CEST] <xreal> ChocolateArmpits: with livestreamer you can directly access a stream, which is in the playlist.
[20:53:04 CEST] <xreal> ChocolateArmpits: with ffmpeg, you need to look through the seperate streams first and map them manually.
[20:53:07 CEST] <c_14> xreal: ffmpeg automatically tries to choose the "best" audio/video streams from the source; if you don't like the default choice you have to map the ones you want.
[20:53:27 CEST] <xreal> c_14: where is the docs is that written? I've look for this the whole day! :D
[20:54:03 CEST] <c_14> https://ffmpeg.org/ffmpeg.html#Stream-selection
[20:55:06 CEST] <xreal> c_14: go away :)
[23:28:27 CEST] <DelphiWorld> sup dudes
[23:28:40 CEST] <DelphiWorld> could someone suggest a h264 transcoding for a tv show?
[23:36:15 CEST] <chungy> I've been encoding Star Trek TNG with "-c:v h264 -crf 21 -vf yadif -tune film" (yadif is the deinterlace filter; necessary for coming from DVD)
[23:36:38 CEST] <chungy> seems decent in my case. it'll probably change depending on yours
[23:37:41 CEST] <DelphiWorld> chungy: this is allready h264 main profile, comming from hd tv
[23:38:41 CEST] <chungy> You probably don't need the yadif filter then
[23:38:52 CEST] <DelphiWorld> its a hd 1080P
[23:38:58 CEST] <DelphiWorld> i want to reduce to 720p or less
[23:39:21 CEST] <chungy> Just play with it. use "-ss <start time> -t 60" and "-preset ultrafast" combined with various -crf settings and see what quality you like best
[23:39:39 CEST] <chungy> Best if you have both high-motion and relatively still scenes
[23:40:01 CEST] <DelphiWorld> chungy: give a acceptable example please?
[23:41:42 CEST] <chungy> "ffmpeg -ss 3:00 -i input.mkv -c:v h264 -crf 21 -vf scale=-1:720 -tune film -t 60 output.mkv"
[23:42:40 CEST] <chungy> oh yeah, and a "-preset ultrafast" somewhere after the -c:v.  It'll basically encode from times 3:00 to 4:00, optimizing for encoding speed rather than file size, but you can see the quality. Adjust -crf accordingly (lower is higher quality)
[23:43:31 CEST] <llogan> -c:a copy
[23:44:07 CEST] <chungy> heh that too
[23:54:08 CEST] <DelphiWorld> no no
[23:54:11 CEST] <DelphiWorld> chungy: see here
[23:54:22 CEST] <DelphiWorld> ffmpeg -ss 3:00 -i '20150527 2000 - Al Jazeera HD - JSC.ts' -c:a libfdk_aac -profile:a aac_he_v2 -b:a 96k -ar 44100 -c:v h264 -crf 21 -vf scale=-1:720 -tune film -t 60 jsc1.mp4
[23:55:59 CEST] <chungy> should be good?
[23:57:05 CEST] <DelphiWorld> i think
[23:57:09 CEST] <DelphiWorld> audio is he-aac
[23:57:48 CEST] <DelphiWorld> ffmpeg -re -i '20150527 2000 - Al Jazeera HD - JSC.ts' -c:a libfdk_aac -profile:a aac_he_v2 -b:a 96k -ar 44100 -c:v h264 -crf 21 -vf scale=-1:720 -tune film jsc1.mp4
[23:59:18 CEST] <chungy> you don't need the "-re"
[23:59:27 CEST] <DelphiWorld> lol
[23:59:30 CEST] <DelphiWorld> i always do it;)
[00:00:00 CEST] <DelphiWorld> hey chungy
[00:00:00 CEST] --- Thu May 28 2015