[Ffmpeg-devel-irc] ffmpeg.log.20141202

[burek]

[00:37] <iwkse> hi, is it possible with the -ss option in ffmpeg to give the time as, 02:10 ? And..there's an alternative to the -t option and have instead something like "end position"?
[00:38] <llogan> yes. -to
[00:38] <iwkse> i basically have a list of times, in the way of "01:10 - 02:20" ecc and I want to automate this
[00:38] <iwkse> ah..cool
[00:39] <llogan> https://trac.ffmpeg.org/wiki/Seeking%20with%20FFmpeg
[00:40] <iwkse> thank you llogan
[00:45] <iwkse> llogan: i'm using the option -c copy because I don't need any compression but doing this way I cannot hear the sound. The file is in VOB format, any idea?
[00:46] <iwkse> i use it like this, ffmpeg -ss 00:05:50 -i bigfile.vob -to 00:05:59 -c copy cut.mpeg
[00:47] <iwkse> understood..
[00:47] <iwkse> should be cut.vob
[00:51] <iwkse> there's still something wrong with the cut..when I use ffmpeg -ss 00:05:50 -i bigfile.vob -to 00:05:59 -c copy cut.vob the length of the cut.vob file is 00:05:57 instead of 9 seconds
[00:53] <iwkse>  ffmpeg -i bigfile.vob -ss 00:05:50 -to 00:05:59 -c copy cut.vob , i don't know why but this works :)
[00:54] <iwkse> ops..i know why..i didn't read the page properly
[01:03] <wyatt8740> so no one could make sense of my error?
[01:03] <wyatt8740> okay
[01:03] <wyatt8740> looks like it's bug report time...
[01:09] <iwkse> wyatt8740: which error?
[01:12] <wyatt8740> http://pastebin.com/vZW2Fysg
[01:12] <wyatt8740> also tried with just the one -i and without the mapping
[01:16] <iwkse> wyatt8740: did you try to change according to line 50-51 ?
[03:54] <peko> Hello!
[03:59] <peko> Guys, I have small questions about ffmpeg`s filter_complex/blend.  There is a way to make ffmpeg respect alpha channel of overlay source? I post an detailed description on video.stackexchange.com - http://goo.gl/cYXz7Q
[05:16] <alec-b> Hello :) I'm using x11grab to record a screen that is playing a movie with a framerate of 25. For some reason if I grab at a rate of 25 or 30, frames are lost and the result isn't as smooth as what's on screen. I need to grab at about 60 fps, any idea why?
[05:18] <alec-b> Which begs a second question. If I have a movie that has 60 fps, and I know it has a ton of duplicate frames and I want to reduce the framerate to a lower value, is there any way I can do so while only dropping duplicated frames and never any unique ones?
[06:59] <share> how to extract the audio from many files at once while preserving filenames?
[07:02] <share> the easy solution is WinFF
[09:35] <croco> hello
[09:35] <croco> i have this line command: 'ffmpeg -f pulse -i default -ac 2 -f x11grab -s 800x600+0+0 -r 30 -i :0.0 -vcodec libx264 -strict -2 ./out.mp4'
[09:35] <croco> and give the error like to http://software.techassistbox.com/all-variants-streaming-with-ffmpeg-unable-to-parse-option-value-1920x108000-as-image-size_69577.html
[09:36] <croco> how to fix?
[09:41] <croco> any ideas?
[09:41] <croco> or all is sleeping?
[09:42] <croco> i use '$(xwininfo -root | awk '/geometry/ {print $2}')' with same result...
[09:43] <croco> how to get offset with x11grab? helpme!
[09:44] <croco> i get width and height but none position x and y
[09:48] <croco> please, helpme :(
[09:49] <croco> i none encounter guide for this, and browser none encounter in the search
[10:14] <pmallek> hello, i have noob question, when i make 2 pass conversion (linux), first one is just /dev/null so it change my input file or ffmpeg "somehow" remember data from first pass and use it for second pass?
[13:54] <anshul_mahe> I am trying to get  SCTE-35 cue insertion message from bmd device, but I always get closed caption SDID and DID.
[14:26] <termos> I have a folder with a lot of small .ts files and I want to convert it into a mp4 file. Is there an easy way to do this?
[14:34] <Mavrik> termos, ffmpeg -i "concat:<file1>.ts|<file2>.ts|..." -c copy output.mp4 should do the trick
[14:40] <sacarasc> termos: cat *.ts > all.ts && ffmpeg -i all.ts -c copy all.mp4
[14:43] <termos> nice, thanks
[14:54] <mco> can someone explain to me why we have 2 fputc in this example ? -> https://www.ffmpeg.org/doxygen/2.4/filtering__audio_8c_source.html#l00182
[14:55] <mco> the output is all good but I dont quite understand why we need 2
[15:34] <pereba> hi
[15:38] <pereba> I need some hint to keep colors as original video, searching the web they said that problem is on playback (decoder), but I already tried many things, nothing fixed it.
[15:38] <pereba> http://1.grn.cc/temp3/2014-12-02_12-08-25.avi http://1.grn.cc/temp3/2014-12-02_12-10-34.mp4  my example, difference on colors is very noticeable
[15:41] <Mavrik> ah, colorspace converson issues :)
[15:44] <Mavrik> pereba, what are your output requirements?
[15:50] <pereba> Mavrik: -y -rtbufsize 100M -f dshow -framerate $fps$ -i video="screen-capture-recorder" -c:v libx264 -r $fps$ -crf 10 -preset medium -tune zerolatency -pix_fmt yuv420p "$output$"
[15:50] <pereba> removing the red part fixes the issue, but doesn't play in browsers, only in MPC :\
[15:50] <Mavrik> I don't see colors :P
[15:51] <Mavrik> you removed pix_fmt? :)
[15:51] <pereba> ok, this part -pix_fmt yuv420p
[15:51] <Mavrik> yes, that's because doing chroma subsampling (e.g. converting RGB screen color format to YUV420) causes the red shift
[15:51] <Mavrik> however only YUV420 is required to be supported by players so hence your problems :)
[15:51] <pereba> hmm how I can fix it and same time make it playable in browsers?
[15:51] <Mavrik> that's why I asked about output requirements - nost players can only play 420
[15:53] <Mavrik> pereba, I don't think there's anything you can do - doing chroma subsampling will lose your color data
[15:54] <pereba> no way... Youtube conversion (html5 player) has the same problem?
[19:40] <troy_s> Anyone in?
[19:40] <troy_s> av_find_input_format() seems to always return null.
[19:40] <troy_s> I need to specifically _avoid_ registering all formats.
[20:03] <troy_s> micechal_: Any secret to only registering a single file format or codec set? I cannot figure out a single method to do so. av_find_input_format applies only to registered codecs and such.
[20:03] <troy_s> (To be clear, I want to control exactly what input formats ffmpeg can handle. av_register_input_format needs an AVInputFormat, and I can't figure out how to get a properly filled structure to pass to it.)
[20:07] <anshul_mahe> using ./configure --disable-all
[20:09] <JEEB> troy_s, avformat.h, line 44 and forward
[20:09] <troy_s> JEEB: Thanks. Looking.
[20:10] <troy_s> anshul_mahe: No control over library configurations.
[20:11] <JEEB> basically, give av_iformat_next() a NULL to get the first one, and if that isn't what you want, keep iterating by giving the last you got to it
[20:11] <troy_s> JEEB: Doesn't appear to give the answer. It lists only registered ones
[20:11] <troy_s> That's for registered codecs and muxers though, right?
[20:11] <JEEB> hmm
[20:11] <troy_s> I'm looking to keep everything unregistered and only register the values the library wants ffmpeg to handle.
[20:11] <JEEB> yeah, I think I overlooked that
[20:12] <troy_s> Which _does_ work if you pass it an AVInputFormat that you fill out, but the problem is the 'official' ones (ff_mov_demuxer for example) are buried in c files.
[20:12] <JEEB> it just dealt with the same kind of data structure :P
[20:12] <troy_s> Basically I'm simply trying to neuter ffmpeg's ability to handle formats programatically.
[20:13] <micechal_> troy_s: why did you mention me?
[20:13] <troy_s> Was hoping you might have an answer as to how to register a specific codec.
[20:13] <troy_s> (or demuxer)
[20:13] <troy_s> Wrong handle. :)
[20:13] <troy_s> Apologies.
[20:14] <micechal_> no problem mate
[20:14] <micechal_> I just found it a bit weird
[20:14] <micechal_> like you had any reason I could be the person to know this or did you pick me at random? :P
[20:15] <troy_s> Uh... the lead developer shares a handle that is deadly close to yours, but my tab key doesn't know that.
[20:15] <troy_s> ;)
[20:15] <micechal_> I see now
[20:16] <troy_s> av_guess_format seems closer.
[20:17] <troy_s> Nope. Still for all registered ones.
[20:17] <troy_s> Dammit.
[20:17] <JEEB> that's what gets called when you actually start to demux
[20:17] <JEEB> you can still say "nope" at that point if it's not something you want to support
[20:17] <troy_s> JEEB: Seems a very tough nut to crack. All the tutorials point to av_register_input_format, but not how to do it.
[20:17] <troy_s> I'm hoping for a more elegant solution such as "register mov, avi, and a few codecs"
[20:20] <JEEB> I have a feeling that those ff_xx_demuxer things are global pointers
[20:21] <JEEB> looking at allformats.c
[20:22] <JEEB> no idea if the symbol gets exported outside of lavf, though
[20:27] <troy_s> JEEB: Yeah if I use extern AVInputFormat ff_blah_demuxe, I get unresolved symbols
[20:28] <troy_s> I _think_ that is the approach, but not entirely sure what library to link to.
[20:28] <JEEB> then it's just not exported and is private (just as the ff_ prefix notes)
[21:47] <mntmn_> hey all! noob question: does ffmpeg somehow have support for 10- or 12-bit-per-channel rgb? ("deep color")
[21:49] <llogan> mntmn_: see "ffmpeg -pix_fmts"
[21:51] <TheJoker2244> Hi guys, I am having trouble with ffmpeg. Here is what we are trying to accomplish: we have a premade opener video which is joined to a recorded video and then joined to a premade ending video. Premade-recorded-Premade. This is being done via an Amazon server. The problem we are having is when we are trying to swap out the opener video with another one. Right now we an opener video that works,
[21:51] <TheJoker2244> but we want to change it. When we make this change, the new opener video plays fine but all the video and audio after becomes corrupted, playing flashing gray boxes and such. I made sure to encode this new video exactly the same as the one that works. Me and my web developer are stumped for why we can't simply change the video out without corrupting the rest. I hope this made sense. I can post
[21:51] <TheJoker2244> the code if necessary.
[21:54] <TheJoker2244> The code we are using here: http://pastebin.com/dCQqSiRr
[21:55] <llogan> do you have complete console outputs? can you show info for each input: ffmpeg -i input0 -i input1 -i input2
[21:57] <TheJoker2244> I'm not sure what you mean, my ffmpeg experience is limited to none. My developer wrote that code.
[21:58] <llogan> when you run ffmpeg commands it will output information in the console
[21:59] <mntmn_> llogan: ok, i only see stuff like gbrp10be and YUV formats, wonder what gbrp is supposed to mean
[21:59] <JEEB> rgb just that it's planar and it's internally not in the r,g,b order
[22:00] <kepstin-laptop> right now you're using the concat demuxer, that would only work if all the files use the same video codec and encoding settings
[22:00] <kepstin-laptop> since you're reencoding anyways, maybe switch to the concat filter?
[22:00] <TheJoker2244> Having my dev check the ffmpeg logs atm
[22:01] <mntmn_> JEEB: ah thx, i see now "planar GBR"
[22:01] <kepstin-laptop> TheJoker2244: it might save some time if you just get your dev in here to talk directly :)
[22:02] <TheJoker2244> He lives in India unfortunatley and I'm working with his assistant right now.
[22:02] <TheJoker2244> Kepstin, what line would this change to the concat filter be made at? according to my pastebin link
[22:04] <llogan> see http://ffmpeg.org/ffmpeg-filters.html#concat
[22:04] <TheJoker2244> Line 12 I would assume. -y -f concat -i would be changed to what?
[22:05] <llogan> see http://ffmpeg.org/ffmpeg-filters.html#concat
[22:05] <TheJoker2244> Thanks, taking a look now
[22:05] <kepstin-laptop> TheJoker2244: final command would be something along the lines of https://gist.githubusercontent.com/kepstin/c1f9628823695b8d3146/raw/33db336be3868735caddb7d58ab0f47319016a4a/gistfile1.txt assuming all files have 1 video and 1 audio track.
[22:06] <kepstin-laptop> and also assuming I made no typos, which is not the best assumption one could make.
[22:10] <TheJoker2244> Thank you Kep, I'll have my guy take a look when he wakes up. So the source of the problem is that somehow when we are switching out the video, the encoding is not consistent and with the way they are being combined, it is causing the corruption?
[22:11] <kepstin-laptop> probably, yeah. With the current way you're doing it, all the videos get run through the same video decoder
[22:11] <kepstin-laptop> so if the encoding settings of one is different, it'll mess up and you'll get corruption
[22:13] <kepstin-laptop> using the concat demuxer is useful if you're using -c:v copy, since it lets you join multiple video streams without re-encoding. But if you have to re-encode anyways, the filter is a cleaner way to do it, since each video is separately decoded.
[22:15] <TheJoker2244> I see, that makes sense to me. In the context of what we are doing, it should work better. Thank you for your help.
[22:31] <troy_s> michaelni: How can you force ffmpeg to only register particular codecs / formats in a list? av_register_input_format ends up with undefined references for extern ff_mov_demuxer for example.
[22:34] <rcombs> anyone know about how much faster an x86_64 x264 is than an i386 build on the same machine?
[22:36] <klaxa> rcombs: why not just benchmark it yourself?
[22:37] <rcombs> klaxa: because I'm pretty sure somebody's already done it
[22:45] <llogan> rcombs: i did it once 4 years ago because i used the wrong OS disc. my faulty memory says "10% faster" for encoding with x264 using -crf 24 -preset slow, but I'm probably wrong
[22:46] <llogan> with grainy SD content
[22:47] <llogan> i forget which CPU it was
[22:47] <rcombs> some googling is giving me results between 7% and 30% depending on the content and settings
[22:48] <rcombs> with ~10% being pretty common, so that sounds right
[22:49] <llogan> it was worth reinstalling the OS for me since I had 1200 hours of inputs to encode.
[22:50] <BtbN> So my wild guess of 10% was somewhat correct
[22:51] <BtbN> People didn't believe me when i told them the difference is not academic
[22:53] <iive> you can win 1-3% if you use vdpau for decoding too. that's what i measured it with the radeon/mesa3d drivers.
[22:59] <BtbN> yeah, takes a little stress from the cpu
[23:00] <BtbN> Doesn't matter if you are only encoding though
[23:01] <rcombs> BtbN: what would "academic" mean here?
[23:01] <BtbN> 0.1% faster or something like that
[23:01] <rcombs> heh
[23:01] <rcombs> less surprising when you realize the ASM differences
[00:00] --- Wed Dec  3 2014