[Ffmpeg-devel-irc] ffmpeg.log.20171226

[burek]

[03:48:54 CET] <mechanist> hello
[03:50:06 CET] <mechanist> when using this command "ffmpeg -ss 0:07:40.8 -i a.mp4 -t 00:18:18.8 -c copy out.mp4" the video produced starts with a strange green tint on VLC, it means that something was wrong
[03:50:22 CET] <mechanist> how can I solve it?
[03:54:12 CET] <JEEB> mechanist: not necessarily. which vlc version?
[03:54:28 CET] <JEEB> highly likely that it just fails at edit lists
[03:54:59 CET] <phryk> ffmpeg seems to be generating corrupt video files for me& Not quite sure what's wrong, but neither firefox nor omxplayer eat the output
[03:55:15 CET] <JEEB> mechanist: also what version of FFmpeg for the record
[03:55:19 CET] <mechanist> JEEB: 2.2.6 Umbrella
[03:55:34 CET] <JEEB> try the 3.0 rcs
[03:55:39 CET] <mechanist> JEEB: the latest one
[03:55:58 CET] <JEEB> 2.2 is very old, unfortunately
[03:56:12 CET] <phryk> ffmpeg -i ente.gif -preset veryslow -crf 22 -c:v libx264 ente.mp4 <- is what i tried, the gif can be found at http://phryk.net/upload/image/ente
[03:57:08 CET] <JEEB> mechanist: if the issue doesn't happen with mpv then first poke the 3.0 release candidate and then if that breaks report it @ videolan's trac
[03:57:26 CET] <mechanist> JEEB: i can't find any 3.0 rcs
[03:57:40 CET] <JEEB> uhh, they released them weeks ago
[03:58:02 CET] <phryk> output works with mpv tho, and mpv isn't displaying any warnings either. used version is 3.4.1 on freebsd
[03:58:12 CET] <mechanist> just checked, it says I have the latest version..
[03:58:21 CET] <kepstin> phryk: firefox on what os? if it can't load system codecs, it uses openh264 which can only decode baseline
[03:58:37 CET] <phryk> kepstin: "baseline"?
[03:58:46 CET] <phryk> also freebsd
[03:58:48 CET] <kepstin> phryk: on linux you can usually fix that by installing gstreamer 1.0 h264 plugin
[03:58:58 CET] <kepstin> might work on freebsd too? :/
[03:59:13 CET] <phryk> I don't want it to work, I want to generate video files that omxplayer will eat
[03:59:35 CET] <kepstin> ah, i don't know about what omxplayer supports :/
[03:59:50 CET] <JEEB> mechanist: poke #videolan I have no idea whay they fon't link it there
[03:59:50 CET] <phryk> yeah, and it's not giving any error messages either :F
[03:59:58 CET] <phryk> kepstin: how do i create a "baseline" video file?
[04:00:11 CET] <kepstin> phryk if it's a really simple player, it might fail on files with no audio track
[04:00:21 CET] <JEEB> phryk: -level 41 or so? if it's rpi it should in theory take high profile
[04:00:35 CET] <JEEB> also check which pix_fmt is getting used
[04:00:45 CET] <JEEB> (aka post logs on pastebin)
[04:00:51 CET] <kepstin> oh, right, gif source means it's probably encoding to yuv444p
[04:02:00 CET] <phryk> oh, then lemme try something converted from another video file
[04:02:03 CET] <kepstin> phryk: it would still be nice if you give us logs, but adding "-pix_fmt yuv420p" to the encode commandline will probably fix it
[04:04:09 CET] <phryk> https://paste.xinu.at/1wQwlm/
[04:04:51 CET] <phryk> Option pixel_format not found.
[04:06:19 CET] <kepstin> it's an output option, not an input option
[04:06:45 CET] <JEEB> mechanist: https://nightlies.videolan.org/build/win32/vlc-3.0.0-rc2-20171225-0301/
[04:06:50 CET] <JEEB> this should do
[04:07:17 CET] <JEEB> complained on #videolan that the official rcs are hard to find
[04:07:41 CET] <mechanist> lol
[04:07:46 CET] <mechanist> thank you
[04:08:31 CET] <JEEB> and yes, I am in a darn bus moving to boonies, japan looking at a terminal on my phone
[04:09:41 CET] <mechanist> windows detecting it as a malware
[04:09:45 CET] <mechanist> what a mockery...
[04:11:05 CET] <CoreX> JEEB does vlc support 10bit now ?
[04:19:39 CET] <phryk> kepstin: thanks, works on firefox now. gonna text omxplayer next :)
[04:19:45 CET] <phryk> test*
[04:34:16 CET] <phryk> omxplayer can't handle a video that short. so I want to loop it. there are multiple non-working loop options documented in the man page, one of them saying, that stuff is deprecated, one should use -loop
[04:34:21 CET] <phryk> but -loop is not documented&
[04:35:58 CET] <phryk> -stream_loop made the video longer, but filled all the extra time with only the last frame of the gif&
[04:37:07 CET] <phryk> -loop is apparently ignored, at least it's not looping anything&^^
[06:13:27 CET] <ruto> Hello,  anybody can help me with actual tutorial
[06:13:37 CET] <ruto> sorry? draft message
[07:31:37 CET] <ruto> Hello. I find in documentation (http://www.ffmpeg.org/ffmpeg-devices.html#dshow) usefull comman for me? for example it is: "fmpeg -list_devices true -f dshow -i dummy". I use it with util and it is work fine. Now i want to use it from my own application, by ffmpeg library without util. How i can execute that ot another same command from my code?
[10:37:38 CET] <diverdude> durandal_1707: hello. Do you think you could point me to the parts of ffmpeg.c i need to be able to do what this command is doing: ffmpeg -i Video3.mp4 video3frames/frame_%04d.png ?
[10:46:29 CET] <durandal_1707> diverdude: demuxing/decoding and muxing/encoding , there are others examples
[10:49:58 CET] <durandal_1707> and that example doesnt crash for me
[11:38:57 CET] <adnn> I don't know if my question is related to ffmpeg, or whether the question itself is wrong. But I'm trying to understand something ...
[11:38:58 CET] <adnn> Say I have a video with 55 seconds of a static image and 5 seconds of lots of pictures. Does that mean that the video bitrate will peak at these 5 seconds? Or is the bitrate the same throughout the video?
[11:41:25 CET] <sfan5> that depends on how your video was encoded
[11:42:01 CET] <sfan5> if it was encoded in a (strict) CBR mode, bitrate will always be the same regardless of video content
[11:42:33 CET] <adnn> hmmm
[11:42:45 CET] <adnn> should I look for "CBR" in the video info to know if that[ the case ?
[11:43:28 CET] <sfan5> in VBR mode (crf or 2pass) it will be like you described: less bits are allocated for the static image, more bits are allocated for the rest
[11:44:32 CET] <sfan5> determining whether a video stream is cbr isn't as easy, sometimes you have access to the encoding params that directly indicate this
[11:44:46 CET] <sfan5> but if you don't, looking at the bitrate over time is pretty much the only way
[11:44:51 CET] <adnn> Is it safe to assume that most videos are vbr ?
[11:44:53 CET] <sfan5> yes
[11:45:04 CET] <adnn> neat
[11:45:08 CET] <sfan5> well, web video
[11:45:13 CET] <sfan5> broadcasting is entirely different
[11:45:24 CET] <adnn> "live broadcasting"?
[11:46:03 CET] <sfan5> by broadcasting i meant TV and the likes
[11:46:09 CET] <adnn> ah ok
[11:46:31 CET] <adnn> I'm worried about web video for now, thankfully
[11:47:32 CET] <kepstin> satelite, cable, and over-the-air signals have a fixed transmission rate, so video bitrate can't be allowed to peak over the limit.
[11:47:51 CET] <kepstin> web video's not so bad, worst case is the viewer has to buffer a bit
[11:52:07 CET] <adnn> is there a command to convert vbr video to cbr in ffmpeg ?
[11:52:20 CET] <sfan5> you'd have to re-encode it
[11:52:56 CET] <sfan5> https://trac.ffmpeg.org/wiki/Encode/H.264#twopass
[11:54:21 CET] <furq> https://trac.ffmpeg.org/wiki/Encode/H.264#AdditionalInformationTips
[11:54:23 CET] <furq> you mean that
[11:54:41 CET] <sfan5> ah yes, picked the wrong section
[11:54:55 CET] <adnn> I almost got lost
[11:55:56 CET] <Darshan> Hello everyone, and merry christmas
[11:56:59 CET] <Darshan> I was wondering if there is a specific reason why ffmpeg do not support the loading of external filter and you have to make a new build ( or use Frei0r). Is it a security problem?
[11:58:20 CET] <iive> you don't want cbr video, nobody uses cbr video and for a good reason. even dvb where you have fixed bitrate channel.
[12:01:24 CET] <adnn> thanks for the warning but I'll have to ignore it if it solves my problem
[12:10:14 CET] <kepstin> well, what's the actual problem you're having?
[12:10:30 CET] <kepstin> all we can tell you right now is that cbr is probably not the solution :)
[12:12:17 CET] <adnn> you're right, I'm doing a horrendous workaround I shouldn't even be doing
[12:12:38 CET] <adnn> the thing is, I'm required to track how many seconds of a video a user has watched, without using Javascript (in an email client)
[12:12:53 CET] <adnn> So the server has to guess how many seconds the user has played based on the amount of transferred data
[12:12:57 CET] <kepstin> well, that's impossible
[12:13:20 CET] <kepstin> typical email usage will have the video loaded through a proxy server that'll prefetch the entire thing
[12:13:24 CET] <kepstin> assuming it does video at all
[12:13:25 CET] <aphirst> adnn, you're going to have to do that a different way
[12:13:30 CET] <adnn> kepstin: I wrote that in the report
[12:13:38 CET] <adnn> aphirst: suggestions?
[12:14:00 CET] <aphirst> proper content-aware media streaming framework, so there's no "guessing"? :P
[12:14:17 CET] <kepstin> aphirst: honestly? put an image in the email with a play button that is actually just a link to a player on a real webpage
[12:14:45 CET] <aphirst> email? oh right i missed that part
[12:14:48 CET] <kepstin> will work with more email clients more reliably, and let you get view tracking info
[12:14:50 CET] <adnn> kepstin: they already have this feature. BTW only a few email clients read videos .. But apparently it'ß a feature they want to incorporate
[12:15:06 CET] <adnn> for the few clients that do support it
[12:15:35 CET] <adnn> and everytime I open a video in the mail client on iOS, a server reads it and blows up the statistics
[12:15:52 CET] <adnn> But I have to do it and then tell them it won't work
[12:17:31 CET] Action: kepstin notes that all the popular webmail clients proxy media as well, so you're limited only to people using desktop clients... most of which hide media by default...
[12:17:42 CET] <furq> i can already tell you how much of a video i'll watch if someone embeds it in an email instead of linking to it
[12:17:45 CET] <furq> it's 0 seconds
[12:17:47 CET] <kepstin> i dunno what clients that would actually work in, tbh
[12:18:32 CET] <kepstin> an email with a <video> tag is a sure sign of spam. that should just get hardcoded in spam filters, imo :)
[12:18:42 CET] <furq> yeah i'm amazed that's even a thing
[12:18:48 CET] <adnn> furq: ikr
[12:19:23 CET] <furq> i'm not amazed it's possible, i'm just amazed there are mail clients that actually let that happen
[12:19:38 CET] <adnn> iOS mail, Apple mail on OSX
[12:19:40 CET] <furq> i would've thought that would always be filtered
[12:19:42 CET] <furq> oh.
[12:19:48 CET] <furq> well then maybe i'm not so amazed any more
[12:19:52 CET] <adnn> hahahaha
[12:20:46 CET] <kepstin> email clients always scare me, they handle html parsing & media & stuff, but typically run older browser engines without sandboxing :/
[12:22:56 CET] <adnn> furq: thunderbird supports video playback too
[12:22:58 CET] <adnn> mind: blown
[12:23:17 CET] <furq> nice
[12:23:23 CET] <furq> i assume that's policy blocked by default though
[12:23:24 CET] <kepstin> like, the reason desktop clients usually block images by default isn't just to avoid email tracking pixels but also to avoid images exploiting image parser bugs for code execution :/
[12:24:17 CET] <adnn> I haven't changed any of the default settings
[12:24:29 CET] <furq> weird
[12:24:39 CET] <furq> i've never changed them and it blocks remote images by default
[12:25:02 CET] <adnn> hmm
[12:25:14 CET] <furq> i last set it up a long time ago though
[12:25:16 CET] <adnn> ah, maybe because I've already accepted remote sources for this particular email
[12:25:29 CET] <adnn> checked an unread one and got the warning
[16:23:56 CET] <klinikot> looking for a ffmpeg command to convert videos to a format compatible with old hardware players
[16:41:52 CET] <nostrora> Hello ! Can we use 2 different file (one .mp3 and one .ogg) to make another file with better quality ?
[16:42:12 CET] <nostrora> take the better of these 2 sources*
[16:57:59 CET] <klinikot> how to convert to format compatible with old hardware players?
[18:17:14 CET] <alexpigment> klinikot: what format?
[18:18:09 CET] <DHE> klaxa: way more information required. if it's h264, definitely yuv420p pix_fmt, and probably main profile (or baseline for cellphones and stuff).
[18:18:12 CET] <DHE> whoops
[18:18:16 CET] <DHE> klinikot: ^^
[18:18:19 CET] <klaxa> :P
[18:18:19 CET] <DHE> (sorry)
[18:47:00 CET] <artushin> https://trac.ffmpeg.org/wiki/CompilationGuide/Ubuntu currently throws a build error because of https://git.videolan.org/?p=x264.git;a=commit;h=71ed44c7312438fac7c5c5301e45522e57127db4
[18:47:16 CET] <artushin> Should that build guide maybe use x264's stable branch?
[18:47:37 CET] <artushin> Haven't tested building with the stable branch yet, doing it now
[18:55:26 CET] <artushin> build succeeded. Propose changing `git -C x264 pull 2> /dev/null || git clone --depth 1 http://git.videolan.org/git/x264 && \` to `git -C x264 pull 2> /dev/null || git clone --depth 1 -b stable http://git.videolan.org/git/x264 && \` if the purpose of the guide is to have something that consistently builds
[18:56:22 CET] <klinikot> alexpigment: i have different input formats, and i need to convert them into format that would be for sure supported on old hardware players
[18:56:45 CET] <alexpigment> klinikot: what players?
[18:56:55 CET] <alexpigment> like a DVD player from 2001?
[18:57:00 CET] <alexpigment> a cell phone from 2009?
[18:57:08 CET] <klinikot> we do not have this player so we could not test
[18:57:13 CET] <alexpigment> what is the player
[18:58:38 CET] <klinikot> as far as i know it is old dvd player that have usb interface
[18:58:46 CET] <alexpigment> ok, so divx then
[18:59:20 CET] <klinikot> probubly divx or xvid?
[19:00:32 CET] <alexpigment> ffmpeg -i [input] -vf scale=640:-2 -c:v libxvid -b:v 1500 -c:a libmp3lame -b:a 160000 [output]
[19:00:37 CET] <alexpigment> try something like that
[19:01:06 CET] <therage3> if compatibility is your issue, then make sure when you have "-vtag xvid" somewhere in there
[19:01:21 CET] <therage3> a lot of old hardware will refuse to play things unless something like that is there, i've run into that issue myself
[19:01:34 CET] <alexpigment> i'll trust therage3 on this one
[19:03:16 CET] <therage3> See for instance https://ubuntuforums.org/showthread.php?t=1906751
[19:03:34 CET] <alexpigment> therage3: that's using the native mpeg4 encoder
[19:03:53 CET] <alexpigment> which is fine, but i think needing to specify vtag xvid is because it's -c:v mpeg4
[19:04:00 CET] <alexpigment> rather than -c:v libxvid
[19:04:06 CET] <therage3> right
[19:04:14 CET] <alexpigment> fwiw, -c:v mpeg4 is faster in my experience, but the quality isn't as good
[19:05:30 CET] <therage3> huh? i thought if you left it out, it uses "fourcc"
[19:05:49 CET] <alexpigment> fourcc is just four letters that identify the codec
[19:05:55 CET] <alexpigment> like XVID
[19:05:59 CET] <alexpigment> or DIVX
[19:06:02 CET] <alexpigment> or DV50
[19:06:15 CET] <therage3> woops, I mean "fmp4"
[19:06:19 CET] <alexpigment> yeah
[19:06:25 CET] <alexpigment> for the native mpeg4 encoder
[19:06:33 CET] <alexpigment> https://trac.ffmpeg.org/wiki/Encode/MPEG-4
[19:06:41 CET] <therage3> i know, I had the wrong thing in copy paste buffer
[19:06:41 CET] <therage3> i meant to paste fmp4
[19:06:52 CET] <alexpigment> yeah
[19:06:56 CET] <therage3> point is, if you use the native mpeg4 encoder without that explicit -vtag, it'll default to fmp4 afaik
[19:07:00 CET] <alexpigment> true
[19:07:05 CET] <therage3> and with that default, a lot of players refuse to play it
[19:07:12 CET] <alexpigment> that's a part of the reason I told him to use libxvid above
[19:07:41 CET] <alexpigment> either way, not a huge deal. you can redundantly use -vtag xvid
[19:07:49 CET] <therage3> hm, I tried to use libxvid before and not the native one, and an old media drive I have here refused to play it
[19:07:59 CET] <therage3> but with the native encoder and with the -vtag declared explicitly, it worked
[19:08:04 CET] <alexpigment> gotcha
[19:08:14 CET] <alexpigment> i'll take your word on this. it's been quite a few years since i've messed with xvid
[19:08:25 CET] <therage3> yeah, we have a lot of old junk around here :D
[19:08:53 CET] <therage3> hell tomorrow I receive a new charger for my old Core2Duo laptop to resurrect it
[19:09:05 CET] <alexpigment> nice
[19:09:11 CET] <alexpigment> 2006, here we come! :0
[19:09:34 CET] <alexpigment> honestly, a core 2 duo mac is still not too shabby
[19:09:47 CET] <alexpigment> i don't know how Windows fares on a core 2 duo anymore
[19:10:01 CET] <therage3> don't know, don't care. i'm going to slap either Ubuntu or Manjaro on it
[19:10:02 CET] <therage3> :D
[19:10:11 CET] <alexpigment> should be pretty smooth then
[19:10:14 CET] <therage3> yep
[19:10:56 CET] <klinikot> thanks a lot, i will try your suggestions
[19:11:02 CET] <therage3> good luck
[19:32:29 CET] <klinikot> alexpigment: shoul i enter for output *.avi ?
[19:32:43 CET] <alexpigment> yeah
[00:00:00 CET] --- Wed Dec 27 2017