[Ffmpeg-devel-irc] ffmpeg.log.20150513

[burek]

[00:04:19 CEST] <Renari> Is there a way to enable an encoder after build? Or do you have to rebuild everything?
[00:04:31 CEST] <c_14> rebuild everything
[00:04:38 CEST] <c_14> well
[00:04:44 CEST] <c_14> rebuild everything you need to rebuild, anyway
[00:05:09 CEST] <Renari> Damn :\ only the decoder was enabled for VP9
[00:14:41 CEST] <bawNg> c_14: If I output the constructed stream to a flv file, and play that in my VLC over HTTP, it is perfect. If I stream it to my nginx over RTMP and then play it over RTMP or HLS, it stops randomly, has artifacts and buffer periods. If I relay the RTMP from nginx directly to twitch using nginx configuration, the same issues with it freezing and having huge artifacts occurs on Twitch
[00:14:58 CEST] <bawNg> Any idea what I could try to resolve this?
[00:22:53 CEST] <c_14> no clue; try updating ffmpeg or nginx-rtmp ?
[00:26:52 CEST] <bawNg> They're both the latest version, installed less than 48 hours ago
[00:27:31 CEST] <c_14> no clue, sorry. always worked for me so far
[00:27:40 CEST] <bawNg> Damn :/
[00:27:56 CEST] <bawNg> Well I'm just about at the point on giving up on this
[00:28:16 CEST] <bawNg> It's already taken more time than it should have, considering it was only meant to be a 2 day hack job
[00:28:52 CEST] <bawNg> It's totally unusable like this, I've tried increasing buffers too, but nothing seems to help
[00:31:36 CEST] <bawNg> I've seen a bunch of posts like this: http://ffmpeg.org/pipermail/ffmpeg-user/2014-January/019344.html
[00:31:49 CEST] <bawNg> Seems like the issue is not that uncommon, but so far there are no solutions
[00:56:57 CEST] <bawNg> c_14: The freezing over RTMP is limited to the video stream, I added an audio stream and that does not freeze, but video constantly freezes
[00:57:36 CEST] <bawNg> But oh well, I'm done with this, I'll consider it a failed experiment
[00:57:46 CEST] <bawNg> Why exactly it failed, we shall never know
[01:33:05 CEST] <]R[> is there any way to override the max analyzeduration ? because for now if any track starts after 2147483647µs (~35min) it can't be detected (and processed).
[01:42:44 CEST] <maslen1> How can I create a second output stream so I can watch the video either locally or remotely (either is fine for me), with minimal overhead?
[01:43:01 CEST] <maslen1> I'm looking at https://trac.ffmpeg.org/wiki/Creating%20multiple%20outputs and https://trac.ffmpeg.org/wiki/StreamingGuide for details
[01:43:25 CEST] <maslen1> The video is being streamed from a real-time source, so no slowing down is necessary
[01:50:44 CEST] <Sazpaimon_> maslen1, did you get my reply to your question from earlier (regarding my use case for -rc)
[01:50:57 CEST] <maslen1> Sorry, I think I may have missed it
[01:51:13 CEST] <Sazpaimon_> *-re
[01:51:27 CEST] <Sazpaimon_> maslen1, i want to stream a playlist of webm files to icecast using ezstream, but ezstream doesn't support timing data for webm, only theora, so I need to stream data to it in realtime
[01:51:56 CEST] <Sazpaimon_> i couldnt find a tool that would do that for me, so I resigned myself to streaming the data
[01:56:32 CEST] <maslen1> I'm just trying to find a way to stream it, either locally or remotely.
[01:56:37 CEST] <maslen1> VLC can't seem to handle it though :/
[02:52:21 CEST] <BenMcLean> how do i write my ffmpeg command to ask for a vdieo to go through HQX ?
[02:56:39 CEST] <BenMcLean> https://ffmpeg.org/ffmpeg-filters.html#hqx
[02:57:03 CEST] <BenMcLean> i am getting error [NULL @ 038ff520] Requested output format 'hqx' is not a suitable output format 4: Invalid argument
[03:04:12 CEST] <BenMcLean> fflogger http://pastebin.com/dCLdTd6f
[03:04:25 CEST] <BenMcLean> %1 is the name of the video
[03:05:03 CEST] <BenMcLean> oops i should have said c_14
[03:05:28 CEST] <c_14> Missing console output
[03:05:48 CEST] <BenMcLean> anyway, i'm trying to blow up DOSBOX captures for editing in Sony Vegas and then deployment to YouTube
[03:06:02 CEST] <BenMcLean> "blow up" meaning scale up, keeping the pixels sharp
[03:07:03 CEST] <BenMcLean> the tiny AVI files DOSBOX produces look like ass on youtube, and i'm not into butts
[03:11:20 CEST] <BenMcLean> so my question is still, "How do I write the command to get ffmpeg to scale up a video with HQX?"
[03:11:58 CEST] <c_14> And I'm still missing the complete console output
[03:12:34 CEST] <BenMcLean> oh ... ok
[03:13:05 CEST] <BenMcLean> c_14, how would that help? I'm not asking why my particular command doesn't work. I'm asking what the correct command would be
[03:13:25 CEST] <BenMcLean> Then I could try it, and provide the output if it doesn't work
[03:13:49 CEST] <BenMcLean> I'm pretty sure that the commands I've tried (like a dozen since I posted that pastebin) are wrong
[03:14:19 CEST] <c_14> `ffmpeg -i file -vf hqx out.mkv' <- works just fine for me
[03:14:29 CEST] <BenMcLean> OK let me try taht
[03:17:39 CEST] <BenMcLean> c_14, here are results: http://pastebin.com/0B6rrm9U
[03:18:21 CEST] <BenMcLean> oh ... it says 2013. that must be old
[03:18:37 CEST] Action: c_14 was just about to check when the hqx filter got ported
[03:19:02 CEST] <BenMcLean> i think recently
[03:19:55 CEST] <c_14> Jun 7 2013
[03:20:23 CEST] <BenMcLean> my build is from march. no idea how the hell i'm running such an old ffmpeg
[03:27:44 CEST] <BenMcLean> i can't figure out where the hell that old ffmpeg is coming from. i've checked every folder in my system path
[03:28:04 CEST] <BenMcLean> for now, i'm just gonna copy a new fmpeg.exe into the same folder with my video file
[03:28:07 CEST] <c_14> which ffmpeg
[03:28:14 CEST] <c_14> oh, windows?
[03:28:18 CEST] <BenMcLean> yeah
[03:28:31 CEST] <c_14> magic"
[03:28:38 CEST] <BenMcLean> unfortunately i sometimes have to use DRM'd crap
[03:29:13 CEST] <BenMcLean> maybe sometime i'll convert to the #LinuxMasterRace if you guys have cookies
[03:29:47 CEST] <c_14> No cookies, but plenty of cookie recipes.
[03:29:51 CEST] <BenMcLean> worse still, i'm on Windows 8, so i can't figure out where the hell the search box went
[03:30:39 CEST] <BenMcLean> found it. i am gonna update this ffmpeg one way or another ...
[03:33:44 CEST] <Nemo377307289> hi
[03:33:47 CEST] <Nemo377307289> I know how to add watermark for the whole video.
[03:33:48 CEST] <Nemo377307289> ffmpeg -i test.mp4 -vf "movie=long.png[watermark];[in][watermark] overlay=10:10" output.mp4
[03:33:48 CEST] <Nemo377307289> But I don't know how to add watermark in the range of time.For example: I hope to add watermark for the "test.mp4" between 10s and 30s.
[03:35:25 CEST] <BenMcLean> seems to be working with the new ffmpeg. thanks! :D
[03:36:11 CEST] <c_14> Nemo377307289: split, overlay, concat
[03:37:37 CEST] <Nemo377307289> i see, c_14
[03:38:31 CEST] <c_14> or to be more precise: split, trim, overlay, concat
[03:38:52 CEST] <c_14> split (into 2 or three copies) trim each copy, overlay the one you want overlayed, then concat
[03:43:16 CEST] <Nemo377307289> why need trim? c_14
[03:43:53 CEST] <c_14> split doesn't do what it sounds like, split is more like clone, trim is what actually trims the video into chunks. The chunk[s] you don't want an overlay on and the chunk[s] you want an overlay on
[03:46:28 CEST] <Nemo377307289> i see, thanks a lot, c_14
[04:14:49 CEST] <grepper> damn, libav doesn't have a subtitle filter either, wish I didn't start supporting both with our scripts
[04:17:00 CEST] <krizz_> I really don't understand this conversion thing. I just want my 1080i camcorder video to be 720p because 1080i seems to scare kdenlive. :(
[04:17:48 CEST] <relaxed> krizz_: use the yadif and scale filters
[04:18:03 CEST] <BenMcLean> its weird, but when i specify hqx=4, i don't get a bigger video. instead, i get some kind of crap where similar colored pixels are combined with diagonal lines
[04:18:39 CEST] <BenMcLean> i want sharp pixel-for-pixel scaling
[04:18:48 CEST] <relaxed> grepper: have your script grab the most recent ffmpeg static build :)
[04:21:21 CEST] <grepper> relaxed: not a bad idea :) But a bit intrusive maybe. How long has the subtitle filter been in ffmpeg ?
[04:21:46 CEST] <relaxed> for a while, not sure of the exact date
[04:23:02 CEST] <relaxed> you could prompt the user with a message saying certain filters require it
[04:23:48 CEST] <relaxed> BenMcLean: to increase the frame size you need the scale filter
[04:24:21 CEST] <grepper> I might put a little ad for Johnvansickle.com :) , but I have a fallback in any case, mplayer/mpv piped to ffmpeg/libav, so either way it should work out
[04:25:19 CEST] <grepper> only bad combo would be mpv and libav, as postproc is not in libav and mpv is often compiled against libav, and even when compiled against ffmpeg it postproc seems to get left out often
[04:25:45 CEST] <grepper> s/it//
[04:28:01 CEST] <relaxed> you're piping audio and video when using mplayer/mpv? if not, that could introduce sync issues.
[04:29:38 CEST] <grepper> yeah, sync is the issue and why I use ffmpeg for both whenever possible
[04:30:38 CEST] <grepper> if they are using mpv and libav they can always test with -slice 0-100 or so in the script and pass an -async X option to mplex
[08:17:19 CEST] <anshul_mahe> I am trying to play encrypted hls using ffplay, it gives me error. while I am able to play encrypted segment
[08:17:22 CEST] <anshul_mahe> http://pastebin.com/x0yWMrNp
[08:34:16 CEST] <anshul_mahe> I debugged in code, it looks like passing key is not the way I have passed
[08:34:41 CEST] <anshul_mahe> does any one know or done passing key in hls
[08:50:21 CEST] <anshul_mahe> ok, I got it its working now
[11:59:57 CEST] <Dark-knight> hey, know its been awhile since i've been in here, just saying hello and asking a quick question.
[11:59:59 CEST] <Dark-knight> whats better?
[12:00:00 CEST] <Dark-knight> video stream Format profile: High at L4.0 with Bit depth: 8 bits
[12:00:02 CEST] <Dark-knight> or
[12:00:03 CEST] <Dark-knight> video stream Format profile: High 10 at L4.0 with Bit depth: 10 bits
[12:00:32 CEST] <Dark-knight> both are .mkv files
[12:01:40 CEST] <rcombs> who the hell encodes High 10 at level 4.0
[12:02:25 CEST] <Dark-knight> its a Hellsing OVA episode.
[12:03:10 CEST] <Dark-knight> idk, it was the only one i could find and I already downloaded both of them, just trying to figure out which one to delete
[12:04:26 CEST] <Dark-knight> same exact video, different numbers
[12:04:41 CEST] <Dark-knight> have no clue whats better
[12:32:23 CEST] <Dark-knight> so nobody awake?
[13:10:26 CEST] <anshul_mahe> is it possible to start rtmp in listen mode, and user connecting it authenticate by username and password
[13:13:04 CEST] <anshul_mahe> do ffmpeg have any authentication system if we start listening with any streaming protocol, (rtsp, rtmp,hls)
[14:24:59 CEST] <thesub6> does anyone know if it is possible to change the bitdepth of an image2 output? i need to change the bitdepth to 10 bit. my input is a 16 bit exr my output should be a 10 bit dpx.
[16:05:02 CEST] <antiatom> Hi all, I am trying to do a 2-pass x264 encode on the CLI, but it fails saying unstuitable output format for /dev/null:
[16:05:05 CEST] <antiatom> ffmpeg -y -i IMG_3035.MOV -c:v libx264 -preset medium -b:v 1200k -pass 1 -c:a libopus -b:a 96k /dev/null && ffmpeg -i input -c:v libx264 -preset medium -b:v 1200k -pass 2 -c:a libopus -b:a 96k output test.mkv
[16:25:26 CEST] <antiatom> Ah I guess it is not -f mkv but -f matroska
[17:46:45 CEST] <rjp421> streaming media east 2015 is live http://www.streamingmedia.com
[17:46:53 CEST] <rjp421> any min now..
[17:51:27 CEST] <LordyLord> Hi
[17:52:05 CEST] <LordyLord> I'm looking to transcode a live 128kbps MP3 stream to 48kbps AAC+ - possible to do tbis with ffmpeg?
[17:52:16 CEST] <LordyLord> s/tbis/this
[17:53:00 CEST] <c_14> ffmpeg -i protocol://where -b:a 48k -c:a aac -strict -2 out.m4a
[17:53:09 CEST] <c_14> At that bitrate you might want to consider libfdk-aac though
[17:53:17 CEST] <Mavrik_> that's not going to create an AAC+ output
[17:53:39 CEST] <LordyLord> c_14: yes, I have recompiled with libfdk-aac
[17:53:53 CEST] <Mavrik_> LordyLord, https://trac.ffmpeg.org/wiki/Encode/AAC#fdk_he
[17:54:15 CEST] <c_14> Aah, I overlooked the +
[17:56:10 CEST] <LordyLord> Maybe the output I want is AAC and not AAC+ - bear with me
[17:56:55 CEST] <LordyLord> I tried the following (which si wrong):
[17:56:55 CEST] <LordyLord> /usr/bin/ffmpeg -re -user-agent 42 -i
[17:56:55 CEST] <LordyLord> http://example1.com:8002/stream.mp3 -c:a libfaac -ar 44100 -ab 48k
[17:56:55 CEST] <LordyLord> -acodec copy http://username:password@example2.com:8099/stream.aac -v
[17:56:55 CEST] <LordyLord> debug
[17:57:05 CEST] <c_14> get rid of the -re
[17:57:13 CEST] <c_14> and the -acodec copy
[17:57:16 CEST] <LordyLord> I've been advised to remove -acodec copy
[17:57:45 CEST] <LordyLord> Thnx - and get rid of -re too ? I thought I need that bcuz it was a lice stream...?
[17:57:52 CEST] <LordyLord> s/lice/live
[17:58:16 CEST] <c_14> You only need that if you have a file and are trying to create a live stream, using it when the source is a live stream is only liable to make problems.
[17:59:21 CEST] <LordyLord> c_14: kk, thnx. Ok, I'll try all this now. BRB. Thnx.
[18:02:18 CEST] <LordyLord> Ok, silly question (apologies). How do I enable libfdk-aac ? Like this ???
[18:02:21 CEST] <LordyLord> /usr/bin/ffmpeg -user-agent 42 -i
[18:02:21 CEST] <LordyLord> http://example1.com:8002/stream.mp3 -c:a libfdk-aac -ar 44100 -ab 48k
[18:02:21 CEST] <LordyLord> http://username:password@example2.com:8099/stream.aac -v debug
[18:02:33 CEST] <LordyLord> I'll try it...
[18:02:48 CEST] <c_14> I think it's libfdk_aac
[18:03:10 CEST] <LordyLord> Yup...
[18:05:12 CEST] <LordyLord> ok, errors. Is it ok to post 9 lines of error here ???
[18:05:28 CEST] <c_14> Use a pastebin service
[18:05:33 CEST] <LordyLord> kk
[18:06:58 CEST] <LordyLord> Ok, I have some sort of errors thus: http://pastebin.com/hLPLFTyz
[18:07:08 CEST] <LordyLord> Thanks for your inout BTW
[18:07:21 CEST] <LordyLord> s/inout/input
[18:07:35 CEST] <seanDunc> hi, I'm trying out the dither abilities in ffmpeg.  I am running the following two commands
[18:07:44 CEST] <seanDunc> ffmpeg -i "1080p_prores_422_10bit_yuv422p10le_test.mov" -pix_fmt yuv420p -an -threads 36 -vf scale=852:480:sws_dither=bayer 8Bit_raw_bayer.yuv
[18:07:57 CEST] <seanDunc> ffmpeg -i "1080p_prores_422_10bit_yuv422p10le_test.mov" -pix_fmt yuv420p -an -threads 36 -vf scale=852:480:sws_dither=x_dither 8Bit_raw_x-dither.yuv
[18:08:22 CEST] <c_14> LordyLord: can you play the input stream in a video/audio player? (also in future, the complete console output is appreciated in case you missed another error somewhere)
[18:08:45 CEST] <seanDunc> however, I don't see a visual difference between 8Bit_raw_x-dither.yuv and 8Bit_raw_x-dither.yuv
[18:09:04 CEST] <seanDunc> I can play .yuv files in yuv player just fine
[18:09:13 CEST] <seanDunc> hold on, I'll get console output
[18:09:20 CEST] <LordyLord> c_14: I think the adts might be an issue - I was advised on the ffmpeg mailing list that "You're actually streaming unchanged MP3 into an ADTS stream named with
[18:09:20 CEST] <LordyLord> the suffix "aac".
[18:09:42 CEST] <seanDunc> oh, just realized you're not talking to me.  sorry
[18:10:32 CEST] <LordyLord> c_14: kk. yes, can play input stream in audio player (e.g. VLC). Will post entire error dump. BRB.
[18:10:46 CEST] <LordyLord> sorry seanDunc - c_14 was talking to me.
[18:11:07 CEST] <seanDunc> when I do a md5 of the two .yuv files, I get the same number on both .yuv files
[18:14:10 CEST] <LordyLord> c_14 - full error dump here: http://pastebin.com/9eRGkg1g
[18:18:41 CEST] <LordyLord> c_14 - of course I have amended the relevant URLs to protect the innocent...
[18:19:41 CEST] <c_14> seanDunc: no clue. looks like it might be a bug though. Might want to check on trac and if you can't find an issue about it create one
[18:20:10 CEST] <c_14> LordyLord: your version of ffmpeg is reeeally outdated
[18:20:27 CEST] <c_14> You should probably build from git.
[18:22:44 CEST] <LordyLord> c_14 - thnx. How can you tell? I rebuilt it yesterday followinjg the instructions from here: https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu
[18:24:27 CEST] <LordyLord> c_14 - ok, version here: ffmpeg version 1.0.10 Copyright (c) 2000-2014 the FFmpeg developers
[18:24:27 CEST] <LordyLord>   built on Jul 25 2014 07:50:40 with gcc 4.7 (Debian 4.7.2-5)
[18:24:53 CEST] <LordyLord> c_14 - yeah, it's old. kk, back to the drawing board...
[18:25:35 CEST] <LordyLord> c_14 - thanks for the advice. I'll be back in about 24 hours then...
[18:26:13 CEST] <c_14> You followed that guide? Cause that should use a recent git snapshot.
[18:27:19 CEST] <c_14> Do you maybe just have another ffmpeg binary in your path that's overshadowing the one you built?
[18:29:07 CEST] <LordyLord> c_14 - yeah, I followed that guide.
[18:29:36 CEST] <LordyLord> c_14 - the the best of my 'ability'. But no complie errors or anything.
[18:30:04 CEST] <Dark-knight> whats better?
[18:30:05 CEST] <Dark-knight> video stream Format profile: High at L4.0 with Bit depth: 8 bits
[18:30:07 CEST] <Dark-knight> or video stream Format profile: High 10 at L4.0 with Bit depth: 10 bits
[18:30:25 CEST] <Dark-knight> both are .mkv files
[18:30:32 CEST] <LordyLord> c_14 - this is the critical line: wget http://ffmpeg.org/releases/ffmpeg-snapshot.tar.bz2
[18:30:58 CEST] <c_14> LordyLord: try `~/bin/ffmpeg -version' and see what that returns
[18:31:48 CEST] <c_14> Dark-knight: assuming everything else is the same 10bit is usually better.
[18:31:56 CEST] <LordyLord> c_14 - yeah, it still comes back as old: ffmpeg version 1.0.10
[18:31:57 CEST] <LordyLord> built on Jul 25 2014 07:50:40 with gcc 4.7 (Debian 4.7.2-5)
[18:32:23 CEST] <c_14> LordyLord: then something probably went wrong during install
[18:32:27 CEST] <LordyLord> c_14 - current version is 2.6.2....
[18:32:31 CEST] <c_14> What does `which ffmpeg' return?
[18:32:41 CEST] <LordyLord> c_14 - ok, thnx, I'll try again.
[18:32:43 CEST] <Dark-knight> c_14: why is the 8bit file bigger?
[18:33:04 CEST] <LordyLord> c_14: /usr/bin/ffmpeg
[18:33:16 CEST] <c_14> Dark-knight: higher bitrate
[18:33:47 CEST] <LordyLord> c_14 - oops, I make a mistake....
[18:34:31 CEST] <LordyLord> c_14 - I have the updated version. See:
[18:34:32 CEST] <LordyLord> user at machine:~# ~/bin/ffmpeg -version
[18:34:32 CEST] <LordyLord> ffmpeg version 2.6.git Copyright (c) 2000-2015 the FFmpeg developers
[18:35:12 CEST] <LordyLord> c_14 - ok, so I have two versions of ffmpeg installed. I will invokee ffmpeg using ~/bin/ffmpeg and see what happens
[18:35:35 CEST] <LordyLord> s/invokee/invoke
[18:36:07 CEST] <Dark-knight> 8bit file = Overall bit rate: 1,117 Kbps
[18:36:09 CEST] <Dark-knight> 10bit file = Overall bit rate: 867 Kbps
[18:36:58 CEST] <Dark-knight> but remember 8bit file = High at L4.0. and 10bit file = High 10 at L4.0
[18:37:14 CEST] <Dark-knight> for these particular mkv files
[18:37:57 CEST] <c_14> Dark-knight: yeah sure, so? higher bitrate = larger file
[18:38:16 CEST] <Dark-knight> 8 is higher than 10?
[18:38:26 CEST] <Dark-knight> no wonder i failed math
[18:38:30 CEST] <Dark-knight> oh wait
[18:38:30 CEST] <c_14> No, 1,117 is higher than 867
[18:38:44 CEST] <c_14> bit depth != bitrate
[18:39:25 CEST] <Dark-knight> ok so for these 2 videos, I should keep the 10bit version?
[18:39:35 CEST] <Dark-knight> yes no maybe?
[18:39:35 CEST] <kepstin-laptop> Dark-knight, keep the one that looks better.
[18:39:47 CEST] <Dark-knight> its an hour long
[18:39:57 CEST] <Dark-knight> not watching both just to find out
[18:40:04 CEST] <kepstin-laptop> then it doesn't really matter
[18:40:52 CEST] <c_14> You can just look at a small part with lots of movement or with gradients to check for banding etc.
[18:41:26 CEST] <Dark-knight> is 10bit at level 4 bad?
[18:41:48 CEST] <Dark-knight> someone earlier said made a comment about that
[18:42:14 CEST] <kepstin-laptop> the only thing that might be bad about it is that it could require more cpu to decode, and it isn't supported by many (any?) hardware players.
[18:42:50 CEST] <kepstin-laptop> if you're only watching it on a computer, and your computer plays it smoothly, neither of those is an issue.
[18:43:17 CEST] <LordyLord> c_14 - I think this progressed further: http://pastebin.com/5VJkLuk0
[18:43:28 CEST] <Dark-knight> ok so c_14 is telling me to choose 10bit, but your info is telling me to choose 8bit
[18:43:46 CEST] <LordyLord> c_14 - I think the error might now be on the relay server ??? '404 not found'
[18:44:17 CEST] <c_14> Dark-knight: I never said to always choose 10bit, I said to choose 10bit when the other factors are equal.
[18:44:32 CEST] <rjp421> streaming media east 2015 is live http://www.streamingmedia.com How To: Producing and Distributing HEVC by Jan Ozer is live in a few min
[18:45:16 CEST] <Dark-knight> ok thanks that answer my question
[18:45:27 CEST] <Dark-knight> answered*
[18:46:54 CEST] <c_14> LordyLord: what are you streaming to? Doesn't look like the server liked the post
[18:50:35 CEST] <LordyLord> c_14 - not a 404 error. It was a 400 error. I am streaming to an Icecast server which should work (it works with Liquidsoap). I want to use ffmepg for lower serever load and also I don't think Liquidsoap can do CBR.
[18:50:59 CEST] <LordyLord> s/serever/server
[18:51:00 CEST] <c_14> LordyLord: https://ffmpeg.org/ffmpeg-protocols.html#Icecast
[18:51:09 CEST] <c_14> You need to use the icecast protocol for streaming to icecast
[18:51:33 CEST] <rjp421> apparently my time is off, live in an hour
[18:51:43 CEST] <LordyLord> c_14 - many thanks. Will try, fingers crossed...
[18:56:41 CEST] <LordyLord> c_14 - nearly there. Got this: [http @ 0x3c29ae0] HTTP error 401 Authentication Required
[18:57:18 CEST] <LordyLord> c_14 - it's an Icecast server implemented in CentovaCast (which does stats and account management).
[18:57:24 CEST] <c_14> LordyLord: did you set the username and password?
[18:57:52 CEST] <LordyLord> c_14 - yes, set it, checked it, and changed to something simple for it for good measure
[19:00:05 CEST] <LordyLord> c_14 - I'll check that I can stream to the server using, for example, Edcast and using same authentication. Maybe not an ffmpeg issue now but Icecast/authentication. BRB.
[19:05:08 CEST] <LordyLord> c_14 - kk, it was an authentication issue. But still have problem. I get this now:
[19:05:09 CEST] <LordyLord> icecast://example1.com:8002/stream.mp3: End of file
[19:05:09 CEST] <LordyLord> [AVIOContext @ 0x36a7240] Statistics: 131600 bytes read, 0 seeks
[19:05:28 CEST] <LordyLord> c-14 - and thge process just stops
[19:05:54 CEST] <LordyLord> s/thge/the
[19:06:04 CEST] <LordyLord> c_14 - I'll do a pastebin. Thnx
[19:12:37 CEST] <LordyLord> c_14 - pastebin: http://pastebin.com/ic4cvU86
[19:12:43 CEST] <LordyLord> c_14 - many thnx
[19:22:50 CEST] <LordyLord> c_14 - I must go soon. Have a meeting to get to. I'll try posting on the ffmpeg mailing list. Many, many thanks for your help. This has moved forward a lot!
[19:42:59 CEST] <LordyLord> kk, must go. AFK. Many thanks: c_14
[19:56:39 CEST] <rjp421> streaming media east 2015 is live http://www.streamingmedia.com How To: Producing and Distributing HEVC by Jan Ozer
[20:27:26 CEST] <Zucca> Hi. I'm getting some seriously GOOD results using zlib lossless video codec. Is there any downsides using it? So far I've seen about 15% lose in compression to libx264 lossless.
[20:28:13 CEST] <Zucca> Also I've read that zlib video codec has two compression levels. How can I change them
[20:28:15 CEST] <Zucca> ?
[20:29:57 CEST] <Zucca> (15% compression losing == libx264 lossless on placebo is about 15% smaller than zlib compressed)
[20:44:27 CEST] <llogan> Zucca: in my single, shitty test zlib came out to be more than 4.5x larger than lossless, medium.
[20:45:40 CEST] <llogan> of course "it depends" on the content
[21:10:24 CEST] <Zucca> llogan: Ah ok. I have screencasting. That might explain.
[21:11:16 CEST] <Zucca> Well. It's very suitable for screencasting at least.
[21:12:40 CEST] <Zucca> Beats ffv1 in every way... although for ffv1 (as for x264) the pixel format I used wasn't available. zlib codec has brg24.
[21:15:55 CEST] <kepstin-laptop> I wonder how zlib compares to png for that use case. theoretically png should do better, it can do transforms prior to the zlib compression to increase the ratios.
[21:33:02 CEST] <Zucca> kepstin-laptop: The paper says something about png filter even.
[21:33:40 CEST] <Zucca> kepstin-laptop: http://multimedia.cx/lcl.txt
[21:36:30 CEST] <kepstin-laptop> heh, from that document "I don't know why it's called png filter since it has nothing to do with filters used in pnglib."
[21:36:40 CEST] <Zucca> :P
[21:40:59 CEST] <kepstin-laptop> huh, apparently my laptop can't manage 30fps at 1080p with -c:v png. Wonder if it's a cpu issue or disk bandwidth.
[21:41:07 CEST] <kepstin-laptop> 188mbit/s is a fair bit :)
[21:43:00 CEST] <thesub6> disk bandwidth should not be the problem
[21:43:17 CEST] <thesub6> u could monitor those values
[21:43:47 CEST] <thesub6> htop under linux. task manager under windows
[21:45:04 CEST] <kepstin-laptop> hmm. looks cpu limited :)
[21:50:34 CEST] <kepstin-laptop> fun. in some quick testing, png does get better compression than zlib on screen capture, but neither is fast enough to to 1080p30 capture on my laptop :)
[23:14:34 CEST] <Zucca> kepstin-laptop: I had to set higher -thread_queue_size for video and audio to get my screencast working. Although I didn't capture my whole screen. The fps was 60.
[23:15:51 CEST] <Zucca> Another codec that gives me pretty good results is ffvhuff. But the filesize is MUCH bigger.
[23:21:30 CEST] <JEEBsv> ffvhuff is a simple huffman, Ut Video is similar
[23:28:18 CEST] <Zucca> i wonder if youtube can accept zlib codec. I believe they use ffmpeg as well...
[23:28:52 CEST] <Zucca> Although uploading 7GB might go over the limits...
[23:30:20 CEST] <kepstin-laptop> if your final encode is gonna be lossy, it might make sense just capturing to a high quality with a lossy codec; the encoder's probably gonna be faster...
[23:31:10 CEST] <Zucca> I'll try one more test: x265 lossless using veryslow preset. (My server is idling, so why not?)
[23:31:25 CEST] <Zucca> I wan't to see how small I can get lossless.
[23:32:08 CEST] <Zucca> kepstin-laptop: I'll upload it to youtube as well. I'll store the lossless version on my server or somewhere else.
[23:33:17 CEST] <Zucca> I'll capture with zlib as long as I can get decent framerates.
[23:33:59 CEST] <Zucca> If I cannot, then I'll try x264 lossy maybe. I haven't tried snow at all...
[23:36:25 CEST] <Zucca> Oh man. 8 threads running. 60 fps 18 minute video. Under 1 fps encoding using libx265. I might stop this and loosen the compression.
[23:37:14 CEST] <kepstin-laptop> nah, x265 is just slow.
[23:38:09 CEST] <Zucca> Yeah. And super slow if using veryslow preset.
[23:38:59 CEST] <Zucca> I heard some new Intel CPUs have h.265 encoding acceleration built-in.
[23:41:00 CEST] <kepstin-laptop> well, the only thing really to say about that is that it'll probably have better quality per bitrate than the h.264 encoder on the chip.
[23:41:17 CEST] <Zucca> I wonder if CPUs have something that accelerates zlib functions. Since it's really fast here. Or at least feels so.
[23:41:39 CEST] <kepstin-laptop> not really, it's just that modern cpus are really, really fast ;)
[23:42:40 CEST] <kepstin-laptop> i wouldn't be surprised if the microcode designers try to make sure the branch prediction works well on zlib tho.
[23:44:32 CEST] <Zucca> hexa-core AMD 2.8GHz on desktop. But still not enought for 1920x1200 30fps screencasting using ffv1 at least. Acatually it's quite tasking process with any codec I tried.
[23:45:14 CEST] <kepstin-laptop> you can probably do it if you use a lossy codec. lossless codecs have to process ridiculous amounts of data for minimal gains :/
[23:45:42 CEST] <Zucca> I think I need to use ffvhuff if I want full screen casting. I looked promising.
[23:46:02 CEST] <Zucca> Sadly lagarith isn't possibility as I have only Linux machines.
[23:46:06 CEST] <kepstin-laptop> if you have the disk space, yeah. just recompress it smaller after if you like.
[23:47:13 CEST] <Zucca> For that disk space I need to clean a lot, or attach eSATA drive.
[23:51:10 CEST] <foonix> you can try utvideo, its pretty good imo
[23:52:01 CEST] <foonix> but ffv1 has better compression ratio if you dont mind more cycles
[23:53:01 CEST] <Zucca> foonix: How's utvideo compared to ffvhuff?
[23:53:12 CEST] <JEEBsv> you're much better with x264 and long gops and stuff than ffv1 IMHO
[23:53:14 CEST] <Zucca> I believe utvideo is pretty new?
[23:53:24 CEST] <JEEBsv> because ffv1 the last I checked was slow
[23:53:48 CEST] <Zucca> Yeah. It is.
[23:54:01 CEST] <JEEBsv> decoding is especially, but encoding IIRC as well
[23:54:16 CEST] <Zucca> And compresses less than zlib on my castings.
[23:54:17 CEST] <JEEBsv> x264 is much more optimized in that sense, and I would believe it will give you better results
[23:54:48 CEST] <Zucca> JEEBsv: I'll try that as time comes for it.
[23:54:58 CEST] <foonix> Zucca: better compression, but you need to try..
[23:55:07 CEST] <JEEBsv> ut video is close to ffvhuff on the very general level, although I haven't tested it  :P
[23:55:10 CEST] <JEEBsv> I just wrote the encoder
[23:55:40 CEST] <JEEBsv> ut video is mostly useful because there's a plugin for all major multimedia frameworks for it
[23:55:53 CEST] <JEEBsv> so for video editors and such it's quite useful
[23:56:01 CEST] <foonix> JEEBsv: i replaced ffvhuff with utvideo, yes the support is briliant
[23:56:34 CEST] <Zucca> I gotta run tests with utvideo asap.
[23:57:19 CEST] <JEEBsv> well it doesn't really have much asm (except for the median prediction, which IIRC later someone found from lavc)
[23:57:26 CEST] <foonix> aka good compression + perfect support everywhere but the best archival codec is ffv1
[23:57:32 CEST] <foonix> dunno if there's something better atm
[23:57:35 CEST] <JEEBsv> uhh
[23:57:48 CEST] <JEEBsv> I dunno, I really prefer stuff that's currently properly specified
[23:57:51 CEST] <JEEBsv> like AVC
[23:58:03 CEST] <JEEBsv> libx264's lossless is quite spiffy and all
[23:58:14 CEST] <JEEBsv> supports everything from RGB/4:4:4 to 4:2:0
[23:58:26 CEST] <kepstin-laptop> hey, ffv1 should be properly specified pretty soon, some work being done there right now :)
[23:58:44 CEST] <JEEBsv> I know of that, which is why I had *currently* there
[23:58:53 CEST] <JEEBsv> also it will not make it much faster I will guess
[00:00:00 CEST] --- Thu May 14 2015