[Ffmpeg-devel-irc] ffmpeg.log.20150426

[burek]

[02:03:34 CEST] <kyleogrg> hey
[02:03:43 CEST] <kyleogrg> curious about doing 25p to 24p
[02:03:54 CEST] <kyleogrg> what's the most common way to do this?
[02:09:50 CEST] <BtbN> Why do you want to do that?
[02:10:59 CEST] <kyleogrg> curiousity, mostly
[02:13:37 CEST] <kyleogrg> I downloaded some 25p video, and it made me wonder if the standard process is to slow down the video to 24p
[02:25:48 CEST] <BtbN> Why would you want to do that?
[02:25:57 CEST] <BtbN> Is it running to fast, and actualy intended to be 24 fps?
[02:26:43 CEST] <kyleogrg> no, nothing's wrong with it
[02:27:05 CEST] <kyleogrg> just wondering what someone would do IF they wanted to make it 24p
[02:47:20 CEST] <prelude2004c> hi everyone.. anyone here familiar with pimd ?
[03:17:12 CEST] <c_14> kyleogrg: either drop frames or slow the video down. Not much else you can do
[03:19:07 CEST] <kyleogrg> c_14: sure
[03:33:07 CEST] <kyleogrg> if i'm going from 60i to 60p (as opposed to 60i to 30p), should i have any special deinterlace settings?
[03:33:25 CEST] <kyleogrg> or is there no difference, just whatever deinterlacer i want to use
[03:37:40 CEST] <c_14> I think you need to tell the deinterlacer to output a frame per field instead of a frame per frame. For yadif this is with mode=1 or 3 iirc
[03:38:42 CEST] <kyleogrg> ok, that's what i thought.  so to do 60i to 60p, is it sufficient to do yadif (one field) and then set -r 59.94?  is that all?
[03:39:13 CEST] <c_14> you shouldn't even have to set the -r
[03:39:34 CEST] <kyleogrg> ok, that yadif setting alone will do it...
[03:40:09 CEST] <kyleogrg> you think mcdeint could be useful?
[03:41:59 CEST] <kyleogrg> that is, is it just as useful for 60i to 60p as it is for 60i to 30p....
[03:43:05 CEST] <c_14> Depends on the amount of motion probably.
[03:45:21 CEST] <kyleogrg> hmm, well, it can't hurt, right?  the only downside would be unnecessary slowness?
[03:49:04 CEST] <c_14> Depends on the content. Test on a sample of the video.
[03:49:11 CEST] <kyleogrg> ok, thanks a lot
[04:18:29 CEST] <hi117> what is that bar that goes down a survelance video called and how do i get rid of it?
[06:02:03 CEST] <kyleogrg> hey
[06:02:19 CEST] <kyleogrg> i'm trying to do a batch two-pass x264 encode
[06:02:42 CEST] <kyleogrg> Here's my command for one-pass so far.  (it takes all the AVIs in a folder) how can i adjust it?
[06:02:45 CEST] <kyleogrg> for %%a in ("*.avi") do ffmpeg.exe -i "%%a" -y -t 00:00:10 -vf "yadif=1:1:0,mcdeint=0:1:1,scale=iw*sar:ih" -sws_flags lanczos -pix_fmt yuv420p -movflags +faststart -c:v libx264 -preset veryslow -c:a libfdk_aac -vbr 4 "%%~na.mp4"
[06:03:05 CEST] <kyleogrg> ignore the -t
[06:10:41 CEST] <kyleogrg> Or, simply put, how can I do TWO commands (Windows) for every AVI, instead of the one command pasted above?
[06:17:50 CEST] <kyleogrg> never mind...
[07:00:30 CEST] <unknownloner> Hello. I'm trying to record video and audio from my desktop, and the audio has a strange desync where it thinks more time has elapsed than really has. When I play the video with mplayer I see that the audio timestamp runs far ahead of the video timestamp. while recording the "time" value is far greater than the actual time I've been recording, matching the final audio timstamp. Example mplayer output: A:
[07:00:33 CEST] <unknownloner> 66.9 V:  16.8 A-V: 50.097. Anything anyone can suggest based on that or should I pastebin my entire command?
[07:02:28 CEST] <unknownloner> Looks like it's related to my somewhat strange conversion process so I'll paste the whole thing
[07:03:56 CEST] <unknownloner> http://pastie.org/10114231
[07:04:41 CEST] <unknownloner> Yes there's a lack of -r in the conversion command, I've tried with and without it in there and it doesn't make a difference
[07:05:56 CEST] <unknownloner> When I use the 'record()' function in there on it's own and write to /dev/null the Time number in the progress display is correct, so it's only after it goes through the FLV encoding that the time is off
[07:06:16 CEST] <unknownloner> Also I've noticed I use -f flv and save to an mp4, whatever the case mplayer figures it out and plays it so w/e
[07:28:36 CEST] <JamJams> Is there a way to change the Encoder metadata tag when remuxing?
[07:28:54 CEST] <JamJams> I tried -metadata encoder="test" but it still writes "Lavf56.30.100"
[09:58:17 CEST] <techtopia> anyone know how i can work out the location of an .ismv file on a http server
[09:59:42 CEST] <techtopia> like the manifest file is here
[09:59:44 CEST] <techtopia> http://lb-ngx.p-01.streamgate.nl/smartvod/video/videoland/138182_TL_Zwarte-Tulp-1.01---Het-Corso_FE.ism/138182_TL_Zwarte-Tulp-1.01---Het-Corso_FE.ismc
[10:00:16 CEST] <techtopia> so the ismv file should be in the same dir
[10:00:26 CEST] <techtopia> but with a diffrent name
[10:00:35 CEST] <techtopia> anyone got any ideas how i can find it
[11:39:03 CEST] <mariangabriel> Hey guys, I got a problem when I want to concatenate 15 videos. Console http://pastebin.com/e336YYAt and the result https://www.youtube.com/watch?v=DItmfKIrtPQ&t=19s
[11:39:09 CEST] <mariangabriel> wird pixels
[11:39:13 CEST] <mariangabriel> wierd
[11:45:50 CEST] <mariangabriel> console dump when I want to add a background http://pastebin.com/ctc1ZsEx . The errors start to 19s
[17:40:54 CEST] <DeadSix27> for testing purposes what would be the way, to under windows, pipe from ffmpeg, to ffplay?
[17:41:32 CEST] <DeadSix27> (if there is one)
[17:41:33 CEST] <BtbN> The exact same way you'd do it everywhere else?
[17:41:39 CEST] <jarainf> ^this
[17:43:00 CEST] <DeadSix27> well, i have no idea how to do it anywhere else
[17:43:56 CEST] <DeadSix27> prob a combination of dev/stdout and the "|" syntax, but, under windows?.. idk.
[17:45:04 CEST] <Anoia> balh | frob
[17:45:15 CEST] <c_14> 'pipe:1 | ffplay -' probably
[17:45:20 CEST] <Anoia> stdout of balh will be attached to stdin of frob
[17:45:42 CEST] <Anoia> I odn;t think - is standardised on Windows to be stdout/stdin
[17:45:48 CEST] <DeadSix27> note: on windows
[17:45:53 CEST] <DeadSix27> ye.
[17:46:03 CEST] <DeadSix27> the | syntay for sure doesnt work
[17:46:07 CEST] <DeadSix27> syntax*
[17:46:13 CEST] <BtbN> the | syntax for sure does work.
[17:46:36 CEST] <c_14> 'pipe:1 | ffplay pipe:0' then
[17:46:53 CEST] <DeadSix27> hm
[17:47:23 CEST] <DeadSix27> uh
[17:47:31 CEST] <DeadSix27> i should definitely work more with the windows console
[17:47:36 CEST] <DeadSix27> c_14: thanks that worked
[17:48:14 CEST] <BtbN> ffmpeg -i something.m4a -c copy -f flv - | ffplay -
[17:48:17 CEST] <BtbN> works fine for me.
[17:48:36 CEST] <DeadSix27> didnt work for me BtbN
[17:48:39 CEST] <DeadSix27> but the c_14 method worked
[17:48:57 CEST] <BtbN> I'm running that right now, it definitely does work, with exactly that command, in the windows cmd.exe
[17:48:58 CEST] <DeadSix27> (maybe i mistyped it the first time idk, but the pipe:1 & pipe:0 method did the trick.
[17:49:11 CEST] <squeegily> Is there an "auto passthrough"? I want to write a shell script: ffmpeg -i "$1" -vf "subtitles='$1'" -c:a copy_if_possible_otherwise_idgaf "$2"
[17:49:58 CEST] <DeadSix27> BtbN: like i assumed, the console was relatively small, so the "-" behind ffplay was right after a linebreak, and i wasnt sure if its there or not, and it probably wasnt the first few times
[17:50:20 CEST] <c_14> squeegily: no
[17:50:35 CEST] <c_14> And as soon as you add a video filter you won't be able to copy anyway.
[17:50:51 CEST] <squeegily> You can still audio copy
[17:51:03 CEST] <squeegily> Saves a tiny bit of CPU and audio quality
[17:51:25 CEST] <squeegily> When transcoding fansubs for the PS3, I usually use c:a copy since the audio is usually AAC
[17:51:33 CEST] <c_14> Oh, didn't see that :a
[17:52:07 CEST] <c_14> The only way would be to use a scripting language of some sort.
[18:19:22 CEST] <jookiyaya> why does so many dvds have black bars on all sides? like this  http://www.imagebam.com/image/eab05d406028359
[18:21:26 CEST] <BtbN> Because TVs overscan anyway, and that way the codec can compress the actual image a little more efficiently.
[18:22:31 CEST] <jookiyaya> btbn but not all dvds have blackbars
[18:22:32 CEST] <Anoia> hangover form the analogue times :|
[18:23:11 CEST] <jookiyaya> btbn what if i buy bluray for same title, would it have black bars?
[18:23:31 CEST] <BtbN> If the producers feel like doing it.
[18:23:43 CEST] <Anoia> i'd have thought it's down to the production company
[18:23:52 CEST] <BtbN> There is no technical reason for it with digital video.
[18:24:06 CEST] <Anoia> it seems stupid to take high res video, down samepl it a bit for TV to upsamepl it again
[18:24:32 CEST] <jookiyaya> have you ever seen blackbars for bluray video?
[19:22:25 CEST] <retard> uh, aren't black bars around the image super bad for codec compression?
[19:22:35 CEST] <retard> with mpeg2 anyway
[19:27:29 CEST] <squeegily> Ugh my "HDTV" is 1366x768 and overscans freaking everything
[19:27:48 CEST] <squeegily> 720p gets scaled up so it can crop ~20px off the edges
[19:28:11 CEST] <RobotsOnDrugs> unless your tv is really that terrible, you should be able to make it display properly
[19:28:51 CEST] <RobotsOnDrugs> probably an option in the picture settings somewhere
[19:29:25 CEST] <squeegily> Regrettably, there is nothing I can do.. I've scoured the menus and the manual and tweaked all the options
[19:30:05 CEST] <squeegily> Windows XP is the only OS I've seen that actually detects 1366x768 as an available option
[19:30:33 CEST] <squeegily> It's probably an OS-level workaround since XP's era overlaps with the "HD-Ready TV" era
[19:30:47 CEST] <squeegily> It was convenient as fuqq
[19:46:24 CEST] <myth_> hi
[19:50:06 CEST] <myth_> For a project I a video with "~/bin/ffmpeg -y -i my.avi -codec copy -map 0 -f segment out\%07d.nut", encode part with "~/bin/ffmpeg -y -i <my segment001> -vcodec libx264 -vb 1000k -acodec copy -f mp4 001.ts" then for segment002... etc. When I concat the result it works fine until I let the "acodec copy". If I re-encode the audio, I can hear the split
[19:51:02 CEST] <myth_> there is a something to avoid this noise ?
[19:54:51 CEST] <techtopia> yes
[19:55:05 CEST] <techtopia> demux the audio
[19:55:48 CEST] <techtopia> and "copy /b" all the audio segments into a single file, then mux it back in to the concatted video file
[19:56:17 CEST] <c_14> Just mux the video over from the original file?
[20:55:28 CEST] <retard> squeegily resolution like that is a function of driver support
[20:55:46 CEST] <retard> it's not a part of the os at all
[22:38:22 CEST] <JoeDawson> Hey, is anyone about to answer a question or two?
[22:39:47 CEST] <c_14> Just ask, if someone can help you they will.
[22:41:09 CEST] <JoeDawson> Thanks - my first is regarding visualizations. Is this possible, bottom left? https://www.youtube.com/watch?v=fEx1k2zW6SA
[22:42:51 CEST] <JoeDawson> My second question would be, if not possible - how about a single horizontal line? The width of the video. I've done some looking around from google and on the ffmpeg docs, but I'm yet to come across something at least similar to what I want.
[22:44:20 CEST] <c_14> 1: If you have that as a video with alpha, the overlay filter should work afaik.
[22:46:03 CEST] <c_14> With single horizontal line do you mean something like showwaves?
[22:48:33 CEST] <JoeDawson> With the first question, I'd ideally like it to be created depending on the audio I pass to it - it might not be clear in that video as the visualization is pretty small, but it's matched to the audio. WIth the single line, showwaves looks like it may be what I'm after - can it be tweaked to look similar to the video above?
[22:49:45 CEST] <JoeDawson> Just come across this actually re: showwaves - going to have a little play now; http://looooooooop.blog35.fc2.com/blog-entry-1098.html
[23:00:40 CEST] <WhiteBunny> Hello everyone, i have a question about .m2v file how can i re-encode it to a bigger size ?
[23:00:58 CEST] <c_14> You have a file and you want to make it bigger?
[23:02:00 CEST] <WhiteBunny> The thing is i have a DVD9 source and i want to make it to a DVD5 source without unwanted materials like FR audio and stuff like that :D
[23:02:40 CEST] <c_14> Isn't DVD9 bigger than DVD5?
[23:03:00 CEST] <WhiteBunny> Yes it is
[23:03:18 CEST] <BtbN> iirc dvd9 is double layer, dvd5 single layer. So a double layer dvd has double capacity.
[23:03:21 CEST] <sfan5> re-encoding MPEG2 video to make it bigger won't make the quality any better
[23:03:29 CEST] <c_14> So wouldn't you want to make it smaller, not larger?
[23:04:17 CEST] <WhiteBunny> But after i have remove all unwanted materials and remove all unwanted audio stream i have a DVD size of 3.37gb :/
[23:05:32 CEST] <c_14> Making it bigger won't increase quality (it'll actually make it worse). Just mux/write what you have.
[23:05:33 CEST] <BtbN> so you are well below the limit of a single layer dvd
[23:08:14 CEST] <WhiteBunny> c_14 if i say it this way, i want to know how the ppl from scene are doing when they take a DVD9 source and re-encoding it to a DVD5 source with only English audio stream and have blank/stripp all Extras and warnings
[23:12:05 CEST] <WhiteBunny> This release as a sample -> The.Interview.MULTISUBS.PAL.DVDR-WILDER i find the DVD9 source that they work on and i tried to stripp all unwanted materials and audio streams and i get a final dvd with English and Italian audio + An video sample of 50mb to a final size of 4.36gb :/
[23:13:05 CEST] <flurant> are you saying that the scene encode is larger?
[23:14:29 CEST] <WhiteBunny> i dont know, but something are they doing to make it to a finally size of a release to 4.42 - 4.46GB :/
[23:14:42 CEST] <flurant> how do you know they are doing that if you dont know the size of it
[23:17:51 CEST] <WhiteBunny> i can easy download the release and demux .m2v file and see how large it is and compare to the .m2v file i get
[23:19:27 CEST] <c_14> Making the file bigger is easier, there's just no real reason to do so as it doesn't (positively) affect the quality.
[23:20:21 CEST] <trodis_> what causes this 5 second delay when i use ffmpeg -> ffserver <-> ffplay
[23:20:54 CEST] <c_14> buffering presumably. Either in the encoder, the muxer, or ffserver or ffplay
[23:20:58 CEST] <trodis_> on ffplay side i turned off buffering with -fflag nobuffer on the server side i set the preroll to 1
[23:21:47 CEST] <trodis_> can u help me reduce this as much as possible, im just streaming audio
[23:22:16 CEST] <trodis_> im encoding with the opus codec on the ffmpeg side
[23:23:10 CEST] <c_14> try -application lowdelay ?
[23:25:19 CEST] <trodis_> the werid part ist when i use libmp3lame the delay is reduced to 1 sec.
[23:25:27 CEST] <trodis_> i thought opus was made for low latency
[23:26:04 CEST] <c_14> Are you using opus or libopus ?
[23:26:10 CEST] <trodis_> libopus
[23:26:20 CEST] <c_14> Did you try adding -application lowdelay?
[23:26:32 CEST] <trodis_> let me try
[23:29:29 CEST] <trodis_> yeah i tried
[23:29:34 CEST] <trodis_> still not improvement
[23:29:38 CEST] <trodis_> with libopus
[23:30:15 CEST] <trodis_> http://pastebin.com/bnVDyvUS ffmpeg
[23:30:27 CEST] <trodis_> http://ix.io/i7U
[23:30:32 CEST] <trodis_> my server conf
[23:32:01 CEST] <c_14> Oh, adding -application lowdelay there won't actually do anything
[23:32:18 CEST] <trodis_> where do i need to add it
[23:32:23 CEST] <c_14> I think you need to add AVOptionAudio application lowdelay to ffserver.conf
[23:33:51 CEST] <c_14> In the stream block
[23:34:19 CEST] <trodis_> yeah i did now ffplay is not playing anything
[23:45:41 CEST] <c_14> ok, I tested on my system. The 5 seconds of delay are because of the ogg format
[23:45:50 CEST] <c_14> If I switch to matroska the delay is much lower
[23:46:06 CEST] <trodis_> yeah thats what i noticed with libmp3lame
[23:46:16 CEST] <trodis_> hmm any suggestion?
[23:46:26 CEST] <trodis_> can i change the format
[23:46:26 CEST] <c_14> Change the stream to test.mkv
[23:46:31 CEST] <trodis_> ok
[23:46:34 CEST] <c_14> Assuming your audio player can handle matroska
[23:46:43 CEST] <trodis_> ffplay should be able to
[23:46:55 CEST] <trodis_> for long term im going to use mplayer or mpc
[23:51:03 CEST] <c_14> Don't know if there are any other containers that can hold opus...
[23:53:35 CEST] <trodis_> even when i use mkv the delay is the same
[23:53:45 CEST] <trodis_> i dont get it srsly so frustrating
[23:53:53 CEST] <c_14> For me mkv has 4s less delay.
[23:54:05 CEST] <trodis_> can i see ur ffserver.conf pls
[23:54:09 CEST] <c_14> I was opening a raw tcp socket with ffmpeg though
[23:54:32 CEST] <c_14> `ffmpeg -re -i foo -c:a libopus -application lowdelay -f oga tcp://127.0.0.1:5555?listen'
[23:54:51 CEST] <trodis_> ah ok
[23:54:54 CEST] <c_14> eh -f matroska
[23:55:14 CEST] <c_14> Then open tcp://127.0.0.1:5555 and check the delay
[23:55:16 CEST] <trodis_> what is the -re option for
[23:56:46 CEST] <c_14> Reads the input in "real-time". You won't need it if you're using a live source like a microphone or something.
[00:00:00 CEST] --- Mon Apr 27 2015