[Ffmpeg-devel-irc] ffmpeg.log.20190430

Wed May 1 03:05:01 EEST 2019

[00:02:58 CEST] <w1d3m0d3> how can I drop a single stream?
[00:03:11 CEST] <w1d3m0d3> or well no, that's easy, I can map with a negative, but two streams?
[00:10:33 CEST] <c_14> -map 0 -map -0:x -map -0:x
[00:10:35 CEST] <c_14> afaik
[00:11:49 CEST] <another> or you can drop a class of streams with -an -vn -dn
[00:12:07 CEST] <another> -sn
[00:15:36 CEST] <TikityTik> does libvpx vp8 use I, P, and B frames too?
[00:15:57 CEST] <TikityTik> like h264?
[00:18:03 CEST] <another> iirc, b frames were already in mpeg4 part 2
[00:18:15 CEST] <another> p frames are even older
[00:18:48 CEST] <another> so the answer is yes
[00:20:31 CEST] <TikityTik> what about b frames?
[00:20:42 CEST] <another> what about them?
[00:20:51 CEST] <TikityTik> does libvpx vp8 have b frames?
[00:20:55 CEST] <another> of course
[00:21:13 CEST] <TikityTik> i'm trying to learn more of the advanced encoding for vp8 to squeeze out more quality
[00:21:38 CEST] <TikityTik> so just learning about these frames right now. Though it's sad that i can't find any advanced guides for specifically libvpx vp8
[00:23:48 CEST] <another> have you tried this one? https://www.webmproject.org/docs/encoder-parameters/
[00:24:06 CEST] <TikityTik> yeah i read it, but the options are not 1:1
[00:24:28 CEST] <TikityTik> also i don't get the difference between abr and constrained quality mode
[00:24:44 CEST] <TikityTik> why i would want to use one over the other
[00:25:03 CEST] <de-facto> Guys, do you know if there are variables for refering to: <full video playback time> in "setpts=..."    and <video width> <videoheight> in "overlay=..."  filters?
[00:25:46 CEST] <de-facto> I need to make my script more dynamic in regards to possible inputs....
[00:25:49 CEST] <furq> no and yes
[00:25:50 CEST] <c_14> TikityTik: why are you using vp8?
[00:25:57 CEST] <TikityTik> c_14: yes
[00:25:58 CEST] <furq> https://ffmpeg.org/ffmpeg-filters.html#overlay-1
[00:28:08 CEST] <TikityTik> de-facto: maybe PTS for the full video playback time
[00:28:21 CEST] <another> TikityTik: the manual has the equivalent vpx options in brackets https://ffmpeg.org/ffmpeg-all.html#libvpx
[00:28:56 CEST] <TikityTik> i read in in my man pages before but this is nicer looking
[00:29:12 CEST] <TikityTik> 'best' : Use best quality deadline. Poorly named and quite slow, this option should be avoided as it may give worse quality output than good.
[00:29:19 CEST] <TikityTik> why would it give worse quality?
[00:29:34 CEST] <de-facto> thanks furq, i see it may be main_w, W, main_h, H in overlay=...
[00:30:03 CEST] <de-facto> TikityTik, how would i access PTS for the full video playback time?
[00:30:18 CEST] <de-facto> ...in setpts=...
[00:30:38 CEST] <TikityTik> i mean PTS might be the full playback time, just a baseless guess
[00:31:01 CEST] <TikityTik> in setpts=...
[00:31:27 CEST] <TikityTik> de-facto: why do you want the full video playback time?
[00:31:57 CEST] <de-facto> ah i see, i am gonna try to play with that, right now i am using something like [s0]setpts=PTS-(3*${DURATION})/(4*TB)[t0];
[00:32:28 CEST] <de-facto> but i dont really always know DURATION in advance, so i would like to make it dynamic, eg DURATION=<full playback time>
[00:42:17 CEST] <Kaedenn> I wish to place a black box over a video for a specific range of timestamps, censoring the content of the video. The video contains sensitive information that was accidentally leaked.
[00:43:13 CEST] <Kaedenn> What's the easiest way to do this, other than splitting the video into before-censor, audio-from-censor, after-censor, and making a sufficiently long black video, and merging them all together?
[00:45:15 CEST] <TikityTik> -filter_complex and using a filter to trim the same input and output each segment as different streams then concat everything
[00:45:20 CEST] <TikityTik> Kaedenn:
[00:45:37 CEST] <Kaedenn> Looks like I have some reading to do
[00:46:23 CEST] <TikityTik> Kaedenn: here's a similar example. http://ix.io/1HBc
[00:46:44 CEST] <Kaedenn> What does this command do?
[00:47:15 CEST] <TikityTik> you can do it with one command
[00:47:29 CEST] <TikityTik> i was combining 2 MKVs that hardcoded subtitles ontop of eachother
[00:47:38 CEST] <TikityTik> and hardcoded*
[00:49:51 CEST] <Kaedenn> Can you explain the content of -filter_complex? I see three expressions: "[0:v]subtitles=out.mkv[v0]", "[1:v]subtitles=out2.mkv[v1]", and "[v0][0:a][v1][1:a]concat=n=2:v=1:a=1 [v] [a]". I'm especially lost at the third expression.
[00:50:01 CEST] <Kaedenn> It seems the first two just set up "variables"
[00:50:30 CEST] <TikityTik> qtakes first file input stream, uses the subtitles filter on it and sets the stream labeled as v0, and son on with the 2nd one
[00:50:37 CEST] <TikityTik> so on*
[00:51:03 CEST] <Kaedenn> As well as -c:v:"[v]" and -c:v:"[a]", do those just define other variables?
[00:51:05 CEST] <pzich> then it takes the last one takes the v0 and v1 input, and the audio from the first fwo files, and concatenates them into one.
[00:51:12 CEST] <Kaedenn> er -c:a:"[a]"
[00:51:14 CEST] <TikityTik> the concat concats everything and outputs 2 streams [v] and [a]
[00:51:16 CEST] <pzich> Kaedenn: those are the outputs of concat
[00:51:27 CEST] <Kaedenn> ohhh
[00:51:30 CEST] <pzich> the first subtitles takes in [0:v] and comes out as [v0]
[00:51:37 CEST] <pzich> goes into concat as [v0] and comes out as [v]
[00:51:40 CEST] <pzich> same with audio.
[00:51:52 CEST] <Kaedenn> aha, -c:v: is video concat, -c:a: is audio concat
[00:52:04 CEST] <another> no
[00:52:05 CEST] <pzich> I think it's video codec
[00:52:10 CEST] <Kaedenn> (it's been so many years since I've done literally anything with ffmpeg) oh?
[00:52:16 CEST] <TikityTik> i probably made it look messier than it needed to be
[00:52:36 CEST] <TikityTik> Kaedenn: -c:v:"[v]" sets the video codec for the [v] stream
[00:52:40 CEST] <Kaedenn> I just want to overlay a black box over the video from aa:bb:cc to aa:bb:cc+7s
[00:52:57 CEST] <pzich> no subtitles or concatenating?
[00:52:58 CEST] <Kaedenn> Making the entire video black is fine
[00:53:04 CEST] <Kaedenn> No subtitles or concatenating. Just censoring.
[00:53:18 CEST] <Kaedenn> "Make these seven seconds black, preserving audio"
[00:53:31 CEST] <pzich> maybe something like this? https://stackoverflow.com/q/49983536
[00:53:42 CEST] <de-facto> TikityTik, afaik setpts= ...PTS ... sets a new presentation time stamp for each frame from current PTS, so as far as i understand the encoder does never know at any point how long the video is going to play in the future
[00:54:19 CEST] <TikityTik> de-facto: you would have to use a tool i think to get the length of the video then i think, i'm no expert on ffmpeg
[00:54:31 CEST] <de-facto> but maybe there is a trick to get it from metadata? i just didnt find a variable in setpts= refering to full playback time
[00:54:46 CEST] <Kaedenn> If it helps, my timestamps are 02:14:50 to 02:14:57
[00:55:02 CEST] <Kaedenn> I'll look into -filter_complex
[00:55:04 CEST] <TikityTik> de-facto: if you want the hackish way, you could awk with output from ffprobe that is not as verbose
[00:55:56 CEST] <TikityTik> or ffprobe -f format
[00:56:02 CEST] <de-facto> yeah i thought about that, but my script is going to be ported to windows batch (yuck), so i cant use such tools
[00:56:12 CEST] <de-facto> was quite happy to see ffmpeg can do arithmetics itself
[00:56:28 CEST] <TikityTik> windows can run ffprobe executable
[00:56:43 CEST] <TikityTik> if you get a windows build
[00:57:46 CEST] <TikityTik> errr not -f i think
[00:58:39 CEST] <Kaedenn> Now, if I wanted to extract the segment in question, is my command ffmpeg -i recording.mp4 -ss 8090 -t 8097 segment.mp4?
[00:58:46 CEST] <Kaedenn> Assuming recording.mp4 is the entire video
[00:59:55 CEST] <TikityTik> you can also use timestamps like 3:50
[01:00:32 CEST] <TikityTik> but the trim filter can work too
[01:00:36 CEST] <TikityTik> https://ffmpeg.org/ffmpeg-filters.html#trim
[01:01:29 CEST] <pzich> de-facto: I've sadly had to use use ffmpeg/ffprobe + awk/sed many times in the past.
[01:01:49 CEST] <pzich> I generally go with ffmpeg + sed since I usually already need ffmpeg to be there and sed makes it easy.
[01:02:30 CEST] <de-facto> yeah that would be ok-ish on linux i guess, is there really no way to refer to full playback time inside ffmpeg filters?
[01:05:49 CEST] <pzich> de-facto: I can't seem to find a list of all the filter variables available, but I'm not aware of one offhand
[01:06:53 CEST] <de-facto> i can imagine its avoided because the future PTS's arent known prior to really decoding them, yet something from metadata would be fine for me already
[01:07:56 CEST] <de-facto> but i only can use variables in setpts= anyways, so i guess its not possible
[01:10:49 CEST] <Kaedenn> "ffmpeg -i recording-full.mp4 -ss 8090 -t 7 output1.mov" seems to spend a while at 100% CPU and has yet to do anything. Input video encoder is "AVC Coding" and output "Lavc56.46.101 libx264", is it transcoding the entire file?
[01:11:23 CEST] <TikityTik> Kaedenn: you can use threads. -threads 8 if you got 8 cores
[01:11:35 CEST] <TikityTik> and there's different types of seeking
[01:11:46 CEST] <Kaedenn> it's using all 8 threads :)
[01:11:55 CEST] <Kaedenn> oh? I just want to extract those 7 seconds
[01:11:55 CEST] <TikityTik> Kaedenn: you're using accurate seek which goes through each frame specifically instead of fast seed where you put -ss and -t before the input file
[01:12:04 CEST] <Kaedenn> aha
[01:12:14 CEST] <TikityTik> Kaedenn: https://trac.ffmpeg.org/wiki/Seeking
[01:12:29 CEST] <pzich> definitely one of my favorite nice pro tips!
[01:12:33 CEST] <pzich> niche*
[01:12:43 CEST] <TikityTik> s/fast seed/fast seek/
[01:12:57 CEST] <another> also, it's encoding with x264
[01:13:09 CEST] <another> since you didn't specify otherwise
[01:13:10 CEST] <TikityTik> Kaedenn: yeah i would use -c copy so it doesn't encode
[01:13:18 CEST] <Kaedenn> oh wow, it says the output file is 1:19:56 long ,even though it's 7s long
[01:13:27 CEST] <TikityTik> it will practically copy paste the data vs transcoding it again
[01:13:28 CEST] <pzich> you might not be able to get a precise seek then though, right?
[01:14:25 CEST] <pzich> bbl, you're in far more capable hands already!
[01:14:37 CEST] <Kaedenn> okay, that worked much faster, but it still says the video is an hour and 20 minutes long
[01:16:29 CEST] <TikityTik> Kaedenn: here's an example. ffmpeg -ss 2:14 -t 2:15 -i "[Erai-raws] Dororo - 16 [720p][Multiple Subtitle].mkv" -c copy out.mkv
[01:16:55 CEST] <TikityTik> Kaedenn: also your formats aren't matching. rcording-full.mp4 vs output1.mov
[01:17:00 CEST] <TikityTik> make sure they have the same extensions
[01:17:19 CEST] <Kaedenn> oh sorry, typo. Input and output are both mp4
[01:17:30 CEST] <TikityTik> copy paste what you wrote
[01:17:32 CEST] <de-facto> ok i think i am going to go with DURATION="$(ffprobe -v error -show_entries format=duration -of default=noprint_wrappers=1:nokey=1 "${SOURCE}")" for now...
[01:17:58 CEST] <Kaedenn> ffmpeg -ss 8090 -t 7 -i GMT20190425-161600_University_3840x2160.mp4 part-8090-7s.mp4 -c copy
[01:18:19 CEST] <TikityTik> Kaedenn: why do you have 2 inputs?
[01:18:25 CEST] <Kaedenn> ...I have two?
[01:18:30 CEST] <TikityTik> oh wait...
[01:18:39 CEST] <TikityTik> Kaedenn: put your output at the very last part
[01:18:41 CEST] <Kaedenn> I thought part8090-7s.mp4 is an output, because no preceeding argument
[01:18:43 CEST] <Kaedenn> ok
[01:19:16 CEST] <TikityTik> Kaedenn: does it work now?
[01:21:32 CEST] <Kaedenn> ok, now it's copying everything from 8090s onward to the end of the video; I just want the first seven seconds
[01:24:45 CEST] <TikityTik> Kaedenn: copy paste your command
[01:24:56 CEST] <Kaedenn> ffmpeg -ss 2:14:50 -i GMT20190425-161600_University_3840x2160.mp4 -to 2:14:57 -c copy part-8090-7.mp4
[01:25:04 CEST] <Kaedenn> I've tried -t and -to
[01:25:15 CEST] <TikityTik> Kaedenn: try -t after -ss
[01:25:36 CEST] <TikityTik> try to keep them group them together before or after the input
[01:25:38 CEST] <TikityTik> grouped*
[01:25:49 CEST] <TikityTik> ah can't write
[01:25:51 CEST] <Kaedenn> Still has the  same problem: the output video is an hour and 20 minutes long
[01:26:01 CEST] <Kaedenn> ffmpeg -ss 2:14:50 -t 2:14:50 -i GMT20190425-161600_University_3840x2160.mp4 -c copy part-8090-7.mp4
[01:26:13 CEST] <TikityTik> Kaedenn: ffmpeg -ss 2:14:50 -t 7 -i GMT20190425-161600_University_3840x2160.mp4 -c copy part-8090-7.mp4
[01:26:51 CEST] <TikityTik> Kaedenn: -to is for actual timestamps, -t is the duration you want
[01:27:05 CEST] <TikityTik> not really sure how -to works as i remember i had troubles with it before
[01:27:09 CEST] <Kaedenn> How do I get the length of the video?
[01:27:17 CEST] <Kaedenn> Is there an ffmpeg option for displaying that?
[01:27:23 CEST] <TikityTik> Kaedenn: ffprobe
[01:27:42 CEST] <Kaedenn> ffprobe doesn't show track length (or I haven't found it)
[01:27:49 CEST] <Kaedenn> oh nvm
[01:28:16 CEST] <Kaedenn> ffprobe part-8090-7.mp4 gives "  Duration: 01:19:56.64, start: 0.464000, bitrate: 424 kb/s"
[01:28:56 CEST] <Kaedenn> Let me paste the output of ffprobe, I think it might shed light on things
[01:29:25 CEST] <Kaedenn> https://gist.github.com/Kaedenn/e6484c3a1e4e97c4e1ac0e79bfc3c614
[01:29:29 CEST] <Kaedenn> It has chapters
[01:29:46 CEST] <TikityTik> Kaedenn: what does ffmpeg -version give?
[01:30:01 CEST] <Kaedenn> ffmpeg version N-47898-g5233f25-static
[01:30:07 CEST] <Kaedenn> gcc 4.9.3
[01:30:18 CEST] <de-facto> Kaedenn, http://trac.ffmpeg.org/wiki/FFprobeTips
[01:30:46 CEST] <TikityTik> Kaedenn: are you on windows, linux?
[01:30:50 CEST] <Kaedenn> linux
[01:30:56 CEST] <TikityTik> and ffmpeg is updated?
[01:31:01 CEST] <Kaedenn> ubuntu 14.04 (I know)
[01:31:05 CEST] <Kaedenn> only to 14.04 :)
[01:31:14 CEST] <TikityTik> that's old
[01:31:30 CEST] <Kaedenn> Should I download and build ffmpeg from source?
[01:31:46 CEST] <TikityTik> i think ffmpeg should still work, but just saying
[01:32:00 CEST] <TikityTik> Kaedenn: copy paste the last encode you done
[01:32:02 CEST] <de-facto> newest static builds for lnx https://johnvansickle.com/ffmpeg/
[01:32:16 CEST] <Kaedenn> ffmpeg -ss 2:14:50 -t 7 -i GMT20190425-161600_University_3840x2160.mp4 -c copy part-8090-7.mp4
[01:32:43 CEST] <TikityTik> try putting -ss 2:14:50 -t 7 after the input
[01:33:00 CEST] <TikityTik> otherwise, might be a bug with your old ffmpeg
[01:33:05 CEST] <TikityTik> or something like that
[01:33:34 CEST] <Kaedenn> that did it!
[01:34:09 CEST] <TikityTik> maybe your keyframes for your video is messed and couldn't find 7 seconds ahead, don't know
[01:34:17 CEST] <TikityTik> or it's an old ffmpeg issue
[01:37:39 CEST] <GuiToris> hi, which profile should I choose?
[01:37:51 CEST] <TikityTik> profile for what? what's your usecase?
[01:37:58 CEST] <GuiToris> I'm encoding a x265 video
[01:38:45 CEST] <GuiToris> TikityTik, does it affect the quality?
[01:39:13 CEST] <TikityTik> i don't encode x264 or x265, but i know it would probably affect for sure compression
[01:39:39 CEST] <GuiToris> do you happen to know which profile is the most supported?
[01:39:47 CEST] <TikityTik> GuiToris: what is your use case? are you streaming or encoding a video file?
[01:40:18 CEST] <GuiToris> I'm just encoding a video file, I'll store it on my disk and if needed I'll burn it on a CD and distribute it
[01:40:32 CEST] <TikityTik> do you care about filesize?
[01:40:53 CEST] <TikityTik> do you care about quality?
[01:40:59 CEST] <GuiToris> both yes
[01:41:05 CEST] <GuiToris> that's why I chose x265
[01:42:09 CEST] <TikityTik> i'm gonna guess that preset is considered to be profile, i would use veryslow. https://x265.readthedocs.io/en/default/presets.html
[01:42:24 CEST] <TikityTik> and also add in the lossless option when encoding
[01:42:51 CEST] <TikityTik> which i think is -x265-params lossless=1
[01:43:02 CEST] <TikityTik> depending if the filesize is that important
[01:43:19 CEST] <GuiToris> I've already set these :)
[01:43:31 CEST] <GuiToris> I just don't know which profile I should choose
[01:43:58 CEST] <TikityTik> you would pick the one that does the most work and is the slowest
[01:44:16 CEST] <GuiToris> not preset
[01:44:17 CEST] <TikityTik> so it focuses as much as it can on getting compression and good quality
[01:44:37 CEST] <GuiToris> base, main, high, high10, high422 and high444
[01:44:43 CEST] <GuiToris> that's the list
[01:44:52 CEST] <GuiToris> I'm thinking about using high422
[01:45:01 CEST] <GuiToris> 444 doesn't work anywhere
[01:45:28 CEST] <another> GuiToris: is your content 422?
[01:45:40 CEST] <another> or 10 bit?
[01:45:44 CEST] <GuiToris> another, no, they are probably 444
[01:45:48 CEST] <GuiToris> they are currently png files
[01:46:27 CEST] <another> you can leave the profile
[01:46:36 CEST] <another> it will automatically choose one
[01:47:22 CEST] <GuiToris> I saw it's optional but it will choose 444, right?
[01:48:14 CEST] <GuiToris> I just created a fast x264 file and we couldn't play it back with quicktime
[01:48:16 CEST] <another> probably
[01:48:45 CEST] <GuiToris> I modified it to base and yuv420p
[01:48:54 CEST] <GuiToris> do you think yuv420p is enough?
[01:49:03 CEST] <another> depends on your use case
[01:49:16 CEST] <another> most video out there is 420
[01:50:08 CEST] <GuiToris> if I set -pix_fmt yuv420p, do I still have to care about the profile option?
[01:50:24 CEST] <GuiToris> or is it kind of same thing?
[01:52:17 CEST] <another> pix_fmt will convert
[01:52:37 CEST] <another> x265 will then choose an apropriate profile
[01:52:52 CEST] <GuiToris> thank you another!
[01:52:59 CEST] <de-facto> quicktime needs -pix_fmt yuv420p i think
[01:53:13 CEST] <GuiToris> yes, it definitely needs!
[01:53:24 CEST] <GuiToris> it wouldn't work without it
[01:53:40 CEST] <TikityTik> does -bf 16 or -max-intra-rate 0 do anything for libvpx? i'm noticing same file size and quality
[01:53:41 CEST] <GuiToris> the screen was blank
[01:54:55 CEST] <TikityTik> i wonder if it's because i'm setting the average bitrate target before
[01:54:57 CEST] <de-facto> afaik profile will be choosen automatically, but if you want to restrict yourself to specific device capabilities it might make sense i guess
[01:55:02 CEST] <GuiToris> another, before I go just a little feedback. Last time when I was here I had issues with premiere and lossless x264 files. I altered the -level but it didn't help. I created those files with -preset veryslow and I had to reencode them with -preset ultrafast and they worked properly
[01:56:10 CEST] <de-facto> GuiToris, i meant this here: https://trac.ffmpeg.org/wiki/Encode/H.264#iOS
[01:56:45 CEST] <de-facto> lots of useful hints in there
[01:57:16 CEST] <GuiToris> de-facto, that's why I included 420p because I coulnd't play it back and I saw this chart here
[01:57:23 CEST] <GuiToris> but thank you
[01:57:36 CEST] <GuiToris> I should be leaving it's really late here
[01:57:39 CEST] <GuiToris> thanks again
[01:57:41 CEST] <GuiToris> see you all
[01:58:45 CEST] <another2> good to know
[02:05:12 CEST] <TikityTik> another: is it even worth it to tinker with I,P, and B frame settings and GOP size?
[02:06:22 CEST] <another> with gop maybe
[02:06:41 CEST] <another> i personally wouldn't bother with the others
[02:06:53 CEST] <another> but if you want to, go ahead
[02:06:56 CEST] <TikityTik> so when would i want low GOP and high GOP?
[02:07:14 CEST] <TikityTik> high GOP means lower file size and same quality i assume
[02:07:47 CEST] <another> a large GOP means it's seeking gets harder
[02:08:08 CEST] <TikityTik> i don't care about seeking
[02:08:43 CEST] <another> then i guess you can increase it
[02:09:15 CEST] <c_14> it'll probably still throw in i-frames on scene changes anyway
[02:09:26 CEST] <c_14> but try increasing it and see what happens
[02:09:41 CEST] <TikityTik> not noticing any change
[02:10:13 CEST] <another> afaik -g set a maximum GOP size
[02:12:15 CEST] <dablitz> good evening. I am doing an RTP, but i have multiple network adapters. what would be the command in ffmpeg to use a specific network adapter to stream out?
[02:12:46 CEST] <c_14> you use the unicast (or multicast) address which routes over that adapter
[02:16:42 CEST] <another> dablitz: what OS?
[02:17:11 CEST] <another> are the adapters in the same network?
[02:19:05 CEST] <dablitz> another: raspbian and no I have multiple networks 1 eth0 on dhcp from ISP, 1 hostapd wifi on 10. network, one lte on dhcp from cell provider and 1 ad-hoc on 192.168
[02:19:29 CEST] <dablitz> i want to stream rtp from ad-hoc
[02:19:43 CEST] <dablitz> to ad-hoc
[02:21:07 CEST] <another> as c_14 said your os will select the right one
[02:21:52 CEST] <c_14> if it's picking the wrong one, add a route with a higher priority to your routing tables
[02:22:11 CEST] <c_14> (for that specific address/subnet)
[02:24:01 CEST] <another> or use policy routing
[02:24:23 CEST] <another> but that probably won't be necessary
[02:39:31 CEST] <Kaedenn> Can -vf be specified multiple times, or do I need to specify it once with everything separated by commas?
[02:48:03 CEST] <Kaedenn> Also, will "ffmpeg -i f1.mp4 -i f2.mp4 -i f3.mp4 out.mp4" just concatenate those files?
[02:48:16 CEST] <another> no
[02:48:30 CEST] <c_14> you can only have -vf once
[02:48:51 CEST] <c_14> you either need to use commas ore -filter_complex and a graph
[02:48:57 CEST] <another> for concat see: https://ffmpeg.org/faq.html#How-can-I-concatenate-video-files_003f and https://trac.ffmpeg.org/wiki/Concatenate
[02:48:57 CEST] <Kaedenn> ok
[03:08:17 CEST] <dablitz> ok my rtp stream work but I can only seem to make it go with libmp3lame. I want to use a codec like pcm_mulaw but nothing happens when I change codec only. sub libmp3lame with pcm_mulaw and no stream, put libmp3lame back and it stream....insights???
[03:09:54 CEST] <another> commandline and outputß
[03:09:57 CEST] <another> ?
[03:10:38 CEST] <dablitz> arecord -f cd -D plughw:1,0 | ffmpeg -i - -acodec libmp3lame -ac 1 -f rtp rtp://192.168.254.31:123
[03:11:33 CEST] <dablitz> other side --> ffplay rtp://192.168.254.30:1234
[03:12:42 CEST] <dablitz> now just sub libmp3lame with pcm_mulaw and nothing happens
[03:13:33 CEST] <another> what's the output?
[03:14:02 CEST] <TikityTik> another: ah i see you're German
[03:15:02 CEST] <TikityTik> just basing off that typo with that character
[03:16:30 CEST] <dablitz> when I ffplay rtp://192.168.254.30:1234 i get full audio stream, even have that display come up. but when I pcm_mulaw it hangs like this --> 0 aq=    0KB vq=    0KB sq=    0B f=0/0   =0/0
[03:18:07 CEST] <dablitz> also uses a lot of cpu
[03:19:04 CEST] <TikityTik> try different presets and/or profiles to have less CPU usage
[03:19:08 CEST] <dablitz> almost pins the cpu
[03:20:28 CEST] <another> TikityTik: you mean the might Eszett ߞ ?
[03:20:33 CEST] <Kaedenn> Anyway, thank you everyone. I've managed to figure out how to do what I need to do.
[03:20:34 CEST] <TikityTik> another: yeah
[03:20:59 CEST] <another> dablitz: what's the output of ffmpeg?
[03:21:19 CEST] <dablitz> ok let me pastbin it
[03:22:50 CEST] <dablitz> https://pastebin.com/2fK1be1f
[03:28:55 CEST] <another> what the output when you use mulaw?
[03:30:24 CEST] <dablitz> one sec
[03:31:35 CEST] <dablitz> https://pastebin.com/5DdXjSYV
[03:36:36 CEST] <another> hmm.. does this work?   ffmpeg -i rtp://192.168.254.30:1234 -c copy out.wav
[03:38:17 CEST] <dablitz> i will try
[03:48:36 CEST] <dablitz> another: did nothing
[03:50:43 CEST] <another> what do you mean nothing?
[03:53:56 CEST] <dablitz> it just hangs
[03:54:10 CEST] <dablitz> cpu maxes out and hangs
[03:59:53 CEST] <another> alright, i gotta go. here are my thoughts:
[04:00:30 CEST] <another> * figure out if ffmpeg is actually transmitting stuff (wireshark)
[04:01:53 CEST] <another> * i'm not sure if you need some protocol for exchanging session information. things to maybe look into: SDP, RTSP
[04:02:24 CEST] <another> * are you dead set on µlaw? maybe choose a different codec. maybe opus
[04:03:00 CEST] <dablitz> tied opus, mulaw, speex,
[04:03:18 CEST] <dablitz> all the same, but thankyou for your help. I will keep googling
[04:03:24 CEST] <another> anyway, good luck
[04:39:52 CEST] <fling> How to use quicksync? Do I only need to enable vaapi?
[04:54:29 CEST] <TikityTik> i love how i made a 150 line bash script just for ffmpeg encoding lol
[05:30:47 CEST] <Montana> what is cheaper solution? (if you don't mind using used)   ddr4-am4  or  ddr3-x79    64GB ram
[06:03:51 CEST] <TikityTik> if i want 2-pass abr for a webm using libvpx and libvorbis, can i get away with just using -b? Or do I need to specify specific bitrate for both video and audio?
[11:26:23 CEST] <aruns> Hey guys, I am working on a website for a client where I have some HTML5 videos whose background colour (i.e. within the video itself, not the background colour of the video's parent element / container) needs to blend in with the background colour of the page.
[11:26:41 CEST] <aruns> I seem to get inconsistent results depending on browser / OS.
[11:26:48 CEST] <aruns> I think GPU can also affect this, from what I have read.
[11:28:05 CEST] <aruns> This is the information I have on a sample video (video amended for privacy reasons): https://bpaste.net/show/cd17e62c2be1
[11:28:31 CEST] <aruns> I am not too au fait with colour profiles, but am I better off converting from using the YUV colour profile to RGB?
[11:33:38 CEST] <ritsuka> if you have already set the color primaries/transfer function/yuv matrix/color range in the file, the only way to resolve it is either wait for browsers to be all color manager and use the same standard, or check the browser/os and use a different background color
[11:35:55 CEST] <ritsuka> the second solution is a bit more realistic&
[11:36:59 CEST] <ritsuka> for example apple shipped macs with 10bit monitor years ago, and css colors in firefox are still utterly broken.
[11:40:20 CEST] <aruns> I see.
[11:40:39 CEST] <aruns> Maybe I am best off trying to see if I can dynamically set the background colour.
[12:58:10 CEST] <snooky> hi all
[13:07:59 CEST] <snooky> Hello. which cpu is better for ffmpeg? an i3 or i5 intel?
[13:08:18 CEST] <durandal_1707> i9
[13:09:05 CEST] <rnderr> G>BCB
[13:09:36 CEST] <snooky> i can only have an i3 or i5
[13:10:22 CEST] <durandal_1707> one with more cores and cycles
[13:11:30 CEST] <JEEB> cache, cores, GHz, AVX2
[13:11:30 CEST] <snooky> Intel  Xeon W3520 	4c/8t
[13:30:59 CEST] <rnderr> AMD Athlon(tm) 64 X2 Dual Core Processor 3800+
[13:31:26 CEST] <durandal_1707> dual core? that is too weak
[13:34:34 CEST] <rnderr> :)
[13:35:20 CEST] <rnderr> all old
[13:35:36 CEST] <rnderr> and I
[13:35:40 CEST] <rnderr> :)
[13:37:31 CEST] <rnderr> Asus M2N32 WS Pro
[13:45:41 CEST] <snooky> Intel® Xeon® W-2145 Octa-Core
[13:45:42 CEST] <snooky> ?
[14:31:10 CEST] <kepstin> you joke, but my athlon64 x2 3800+ box still works, and it does have ffmpeg installed on it :/
[16:19:44 CEST] <jlw4049> wondering if I could get a little help using FFMPEG with x265 HDR content?
[16:26:19 CEST] <rnderr> libvpx-vp9?
[18:08:59 CEST] <hannes_> hey, https://ffmpeg.org/ffmpeg-codecs.html#Supported-Pixel-Formats says "it is not possible to build one FFmpeg with multiple versions of x264 with different bit depths" but it seems to work for me on arch. -pix_fmt yuv420p10le told me "[libx264 @ 0x563bac96b300] profile High 10, level 3.1, 4:2:0, 10-bit", while -pix_fmt yuv420p said "[libx264 @ 0x562601ec7300] profile High, level 3.1, 4:2:0, 8-bit"
[18:13:20 CEST] <JEEB> hannes_: latest versions have supported the new API in x264 that enables both 8bit and 10bit at the same time for x264
[18:13:33 CEST] <JEEB> before you had to build separate libs for it
[18:18:33 CEST] <hannes_> nice :)
[19:47:49 CEST] <TikityTik> good morning my ffmpeg lords
[19:48:25 CEST] <TikityTik> i am wondering if i could 2-pass abr a webm using libvpx and libvorbis where I can get away with just -b? Or is it needed to set specific bitrates for each codec
[19:48:57 CEST] <TikityTik> Also, does libvorbix have something like qmax? Is there a bias bitrate setting when using just -b to prefer more video by a given percentage?
[20:03:15 CEST] <relaxed> libvorbis will have a default bitrate if -b:a is omitted, by why not set it?
[20:03:28 CEST] <relaxed> but why*
[20:06:58 CEST] <TikityTik> i was hoping to achieve a certain bitrate with by specifying it once for the encoding and use some bias setting between the video and audio stream
[20:07:18 CEST] <TikityTik> total bitrate i mean
[20:07:25 CEST] <furq> yeah that's not a thing
[20:07:39 CEST] <TikityTik> is there at least something like qmin or qmax for libvorbis?
[20:07:56 CEST] <furq> there's -q
[20:08:32 CEST] <TikityTik> i tried adjusting bitrate usage via -q:a and i noticed it would tend to use too much bitrate still
[20:08:50 CEST] <relaxed> if you need a specific size use -b:a, -q:a is variable
[20:09:14 CEST] <TikityTik> and are the values 0-10 or -2 to 10? it's contradictory to this http://wiki.hydrogenaud.io/index.php?title=Recommended_Ogg_Vorbis#Recommended_Encoder_Settings
[20:09:26 CEST] <TikityTik> for -q:a
[20:10:19 CEST] <furq> it's the same as libvorbis
[20:10:43 CEST] <relaxed> man oggenc lists -1 through 10
[21:27:29 CEST] <TikityTik> is there a reason why -fs doesn't stop ffmpeg accurately?
[21:27:43 CEST] <TikityTik> it seems like it either overshoot or undershoots
[21:31:09 CEST] <furq> it should always overshoot
[21:31:24 CEST] <furq> obviously it can't hit exactly or else it'd stop in the middle of a frame/packet
[21:34:32 CEST] <TikityTik> can i limit ffmpeg from using more than 25% cpu but using slowest encoding?
[21:35:19 CEST] <TikityTik> i could use -threads 1, but it would make 1 core use up a lot of CPU.
[21:35:34 CEST] <TikityTik> And is it possible that using high CPU from encoding can damage my CPU even if temps are not high?
[21:35:45 CEST] <TikityTik> where it will wear out over time?
[21:39:03 CEST] <mozzarella> TikityTik: you can use cpulimit
[21:39:05 CEST] <mozzarella> that's what I do
[21:40:13 CEST] <furq> 20:35:34 ( TikityTik) And is it possible that using high CPU from encoding can damage my CPU even if temps are not high?
[21:40:16 CEST] <furq> no
[21:40:46 CEST] <TikityTik> i remember really messing up my overclocked CPU where I was encoding with high CPU usage
[21:40:53 CEST] <furq> yeah because it was overclocked
[21:41:02 CEST] <TikityTik> and i remember reading that the CPU wears out from high usage
[21:41:05 CEST] <furq> i assume it was overvolted as well
[21:41:08 CEST] <TikityTik> yeah
[21:41:12 CEST] <furq> overvolting in particular will stress out a cpu
[21:41:37 CEST] <TikityTik> i got a 7700k and i really don't want to break it...yet
[21:41:48 CEST] <furq> it's literally designed to do this
[21:41:56 CEST] <TikityTik> eh? it is?
[21:42:09 CEST] <furq> this is like buying a ferrari and not shifting it into third gear
[21:42:19 CEST] <TikityTik> i'm not sure if it's turbo freq of 4.5 GHz will cause it to wear down
[21:42:22 CEST] <TikityTik> its*
[21:43:13 CEST] <TikityTik> the last cpu i had was a 4670k i think, i overclocked it to 4.5 GHz
[23:22:06 CEST] <TikityTik> does ffmpeg report in kiB or kB?
[23:52:13 CEST] <Montana> everybody:   did you know that intel system does not support  high-density ram:  where AMD system does
[23:53:31 CEST] <BtbN> Didn't you say the exact same thing a few days ago, and then failed to explain what High Density RAM is even supposed to be?
[23:53:54 CEST] <BtbN> Cause our servers with Intel Xeons support pretty dense registered RAM just fine.
[23:53:57 CEST] <Montana> btbn high density =  higher than 256M on a chip
[23:54:16 CEST] <Montana> btbn  you are right, it's for UDIMM only
[23:54:56 CEST] <Montana> btbn are you clear now?
[23:55:21 CEST] <BtbN> Aren't current DRAM Chips somewhere in the region of several Gbit per chip?
[23:56:46 CEST] <Montana> btbn sorry i should have typed...
[23:56:55 CEST] <Montana> everybody:   did you know that intel system does not support  high-density ram  for ddr3 and UDIMM:  where AMD system does
[00:00:00 CEST] --- Wed May  1 2019