[Ffmpeg-devel-irc] ffmpeg.log.20191116

burek burek at teamnet.rs
Sun Nov 17 03:05:02 EET 2019


[00:28:49 CET] <arnaud-ro> assuming I have this
[00:28:52 CET] <arnaud-ro> Program 0
[00:28:56 CET] <arnaud-ro> : Other (qad)
[00:29:31 CET] <arnaud-ro> how would I correctly map  the program and the audio video I want to chose, knowing that I output to flv so I can't have several audio tracks
[00:30:03 CET] <arnaud-ro> -map p:0 -map a:0 -map v:3  doesn't work
[00:30:29 CET] <arnaud-ro> (there are 7 programs so I need to select one)
[00:30:50 CET] <DHE> how about a pastebin?
[00:30:55 CET] <arnaud-ro> ok
[00:33:31 CET] <arnaud-ro> https://pastebin.com/DpdrKXkj
[00:33:57 CET] <arnaud-ro> if I only select the program, then it doesn't work because there are 3 audio tracks and flv output doesn't allow it
[00:34:15 CET] <arnaud-ro> so i m not sure how I should write the command..
[00:34:44 CET] <DHE> you might be best off first selecting a specific variant rather than the master stream, then use -map 0:v -map 0:1 or whatever audio stream you want from the set...
[00:36:08 CET] <arnaud-ro> i m not aware of such variant, just the master playlist
[00:40:03 CET] <DHE> if you just run 'curl https://streams._hd' it should print all the variants and you can select what you're actually interested in
[00:52:33 CET] <arnaud-ro> no it doesn't work
[01:19:52 CET] <arnaud-ro> irony is when you don't map anything it just works fine (ie select only one audio stream on his own)
[01:30:12 CET] <arnaud-ro> p:6:0 p:6:4  that works..
[07:15:14 CET] <gp> anyone know by chance when the hls master_pl_name option was added? trying to find minimum version for it
[07:15:35 CET] <gp> 3.4.6 errors with unrecognized option
[07:15:48 CET] <gp> couldn't find it in the changelog
[07:16:09 CET] <JEEB> if you are trying to check if the feature is available, I think it makes sense to check for the feature
[07:16:16 CET] <JEEB> `-h muxer=hls`
[07:16:28 CET] <JEEB> if you get a match for that string, that binary has that feature
[07:16:45 CET] <JEEB> unless you're doing API, in which case it should be possible to list the available AVOptions as well
[07:16:48 CET] <JEEB> :)
[07:17:25 CET] <JEEB> I could go into the git history and try to figure where it was added, but as an application I feel it is more useful to check for the availability of an option rather than check by version
[07:18:17 CET] <gp> JEEB: no problem. My parser script is breaking when I upgrade to ffmpeg 4 so I was hoping to get the master option in an earlier one. so i could delay the major version bump
[07:19:34 CET] <JEEB> also the "major" bump in v4 was not really major :P
[07:19:42 CET] <JEEB> it's just like with the linux kernel
[07:19:52 CET] <JEEB> when numbers get big enough the major gets bumped
[07:20:07 CET] <JEEB> so whatever breakage you have might have been utilizing something that wasn't about to be stable to begin with
[07:20:36 CET] <JEEB> (or it's a bug, of course, but not a whole lot of things from the cli stdout/err are standardized)
[07:20:54 CET] <gp> JEEB: no it's my problem from parsing the cli output
[07:21:33 CET] <JEEB> but there is nothing in between 3.4 and 4 as far as releases go :P and new features do not get added to release branches
[07:21:56 CET] <gp> JEEB: this is probably worth the regex nightmare https://github.com/FFmpeg/FFmpeg/commit/606879ec42980368b8bfe523c90ad859f448b18b
[07:23:10 CET] <gp> Just felt comfy using the ubuntu version hah
[07:24:19 CET] <JEEB> I basically would not parse too much of what ffmpeg.c outputs
[07:25:49 CET] <JEEB> if actual interfaces are needed, they should be done (or APIs used that are available)
[07:47:58 CET] <JustLandedOnMars> hi
[07:48:13 CET] <JustLandedOnMars> is there a difference between using "-vcodec libx264" and "-vcodec h264" ?
[07:54:11 CET] <JEEB> JustLandedOnMars: h264 just picks *some* encoder that fits the bill
[07:54:29 CET] <JEEB> setting -c:v libx264 (or some other video encoder) then picks a *specific* encoder
[10:45:14 CET] <solars> hi, I'm recording a HTTP stream but the playback is way too fast. I guess input differs from output fps. How can I fix this? The command is: ffmpeg -i http://url -codec copy -map 0 -f segment -segment_time 900 -segment_atclocktime 1 out%03d.mp4
[10:47:31 CET] <JEEB> FFmpeg APIs and ffmpeg.c will go as fast as it can
[10:50:17 CET] <JEEB> there's a -re option if you want to simulate real-time, but if your input timestamps go wee-wee, then that will cause issues
[10:50:34 CET] <JEEB> basically, if you need the real world clock to be a thing, you need to make your own API client
[11:09:36 CET] <solars> JEEB: hmmm is there nothing I can use to record an http stream in the actual time?
[17:06:33 CET] <JustLandedOnMars> JEEB: so which option is "better" ?
[17:12:48 CET] <microchip_> JustLandedOnMars: libx264
[17:15:40 CET] <Compressionz> Hey all, I'm experimenting with adding custom prediction methods to the png encoder/decoder. I was curious if the decoder uses a specific prediction method by default or if it uses whichever one was used on compression?
[20:22:47 CET] <gp> How do I get a codec string like "avc1.64002a" from a given video file?  I inspected the metadata with ffprobe and it only seems to give me the string "avc1"
[20:22:59 CET] <gp> Is there some sort of lookup somewhere? Or do I have to know how to map profile and level?
[20:29:50 CET] <DHE> the level and stuff is encoded in there. 2a has hex value 42 which means level 4.2 so I am going to take a guess, 1080p at 60fps
[21:02:23 CET] <gp> DHE: spot on =) trying to go the other direction though. Wrapping my head around adding formats and the CODECS. Was expecting to just specify the master playlist file for HLS and see it work
[21:02:39 CET] <gp> but vp9 and h265 do not get a CODEC listed
[21:04:25 CET] <gp> The first half of the string like "avc1" is straightfoward but not sure how to turn level 4.2, 1080p, and 60fps to 64002a.  Or if it is consistent between codecs - probably not?
[21:08:53 CET] <furq> gp: https://wiki.whatwg.org/wiki/Video_type_parameters#Video_Codecs_3
[21:10:40 CET] <gp> furq: thank you
[21:11:20 CET] <furq> not entirely sure that's correct
[21:11:42 CET] <furq> looks like the first byte is the profile (0x64 = 100 = high profile) and the last byte is the level as DHE said
[21:11:45 CET] <furq> so idk what the middle byte is
[21:13:07 CET] <DHE> I don't know the meaning, but it encodes baseline/main/high somehow..
[21:14:49 CET] <furq> apparently it's the constraint set
[21:15:01 CET] <furq> not sure why that whatwg example has it as 0x40 for main since that doesn't imply any constraint set flags
[21:15:20 CET] <furq> https://tools.ietf.org/html/rfc6381#section-3.3
[22:04:01 CET] <void09> any way to get screenshots from a certain timestamp of a DVD (folders on hdd, not disc) using just ffmpeg ?
[22:04:48 CET] <void09> the main video track of a dvd
[22:06:04 CET] <void09> without writing code to get the duration of all the main .vob files and other code to figure out which vob file at which timestamp is the timestamp i am looking for
[00:00:00 CET] --- Sun Nov 17 2019


More information about the Ffmpeg-devel-irc mailing list