[Ffmpeg-devel-irc] ffmpeg.log.20150405

burek burek021 at gmail.com
Mon Apr 6 02:05:01 CEST 2015


[00:00:15 CEST] <Mavrik> if by ".ogv" you mean theora
[00:00:29 CEST] <Mavrik> it's about lightyears behind in quality
[00:00:42 CEST] <Mavrik> and several real years behind in technology, it's a dead codec
[00:01:02 CEST] <Mavrik> and PLEASE... PLEASE... do learn the difference between a container and a codec before starting out with digital video -_-
[00:01:07 CEST] <plujon> Well, it's only a screencast, not a movie...
[00:01:19 CEST] <plujon> Heh, I really wish I wasn't starting out with digital video.
[00:01:34 CEST] <plujon> I just want to "share my screen" to show people how to use a program.
[00:02:52 CEST] <plujon> I know next to nothing about codecs, containers, and video and audio formats in general.
[00:04:29 CEST] <KeitaroNL> I've got trouble encoding from MKV to MP4. I keep getting the error [matroska @ 0x2805670]Unknown entry 0xB2. Anyone else have any experience with this?
[02:14:15 CEST] <electronrotoscop> can anyone point me to a good guide for compiling ffmpeg for Win 64 on a 64 bit linux system? Or a good place to ask questions like that?
[02:20:59 CEST] <c_14> I'm not entirely sure if the guide is good, but there's https://trac.ffmpeg.org/wiki/CompilationGuide/CrossCompilingForWindows, and you can probably ask around here http://ffmpeg.zeranoe.com/forum/. Though if you have specific questions or errors you can always ask here or on the user mailing list
[02:22:03 CEST] <electronrotoscop> I tried reading the guide and looking around the forum, but really what I'm curious about is what flags to use to tell Mingw-w64 to create a 64-bit static exe
[02:22:16 CEST] <electronrotoscop> all the guides I find with cli examples compile for 32bit
[02:23:23 CEST] <c_14> Use a 64-bit compiler and make sure you set the arch and cross-prefix correctly?
[02:24:22 CEST] <electronrotoscop> that sounds like the kind of information I need, but I unfortunately know too little to know how to tell what the right arch and cross-prefix settings are
[02:31:03 CEST] <c_14> Well, you can either use one of the scripts/build environments linked on that page, or install a 64bit version of mingw. Then run the configure under "compile" just with arch=x86_64 target-os=mingw64 cross-prefix=x86_64-w64-mingw64 (probably)
[02:31:40 CEST] <electronrotoscop> okay thanks I'll give that a shot
[02:35:13 CEST] <electronrotoscop> it gave me x86_64-w64-mingw64gcc is unable to create an executable file.
[02:35:13 CEST] <electronrotoscop> C compiler test failed.
[02:37:51 CEST] <Prelude2004c> Hey everyone, can anyone have a look at this and see if there are ny issues with this output ? -c:v libx264 -preset $5 -r 30 -g 20 -keyint_min 90 -bf 0 -b:v ${bitrate} -minrate 0k -maxrate ${maxrate} -bufsize 1000k -s ${resolution} -flags +loop -me_method hex -g 250 -qcomp 0.6 -qmin 10 -qmax 51 -qdiff 4 -bf 16 -b_strategy 1 -i_qfactor 0.71 -cmp +chroma -subq 1 -me_range 16 -coder 1 -sc_threshold 40
[02:37:52 CEST] <Prelude2004c>   -keyint_min 25 -refs 1 -trellis 0 -partitions -parti8x8-parti4x4-partp8x8-partp4x4-partb8x8-an .. What do you guys think of this output ? anyting i should be worried about ? I see some clipping on fast motion though as though there is image clipping or something.. not sure why.
[02:40:39 CEST] <c_14> electronrotoscop: try finding out what the name of the compiler you installed is
[02:41:14 CEST] <electronrotoscop> I think it's mingw-w64 but I don't know if there are other parts to the name. I just ran the script from http://ffmpeg.zeranoe.com/blog/?p=400
[02:44:06 CEST] <c_14> electronrotoscop: it should have compiled something, somwehere, probably. Most likely something with 'gcc' in the name somewhere
[02:45:04 CEST] <electronrotoscop> Once it compiled it told me to run: export PATH="/home/u64/software/mingw-w64/mingw-w64-x86_64/bin:$PATH" which I did
[02:45:31 CEST] <c_14> Prelude2004c: other than the fact that you're using a lot of x264 suboptions, not really
[02:45:37 CEST] <electronrotoscop> I tried going to /home/u64/software/mingw-w64/mingw-w64-x86_64/bin and there's a bunch of I guess binaries in there, I'm trying them now in place of =x86_64-w64-mingw64 under cross-prefix but no luck so far
[02:45:56 CEST] <c_14> mhm then the prefix is probably mingw-w64-x86_64
[02:46:28 CEST] <electronrotoscop> ./configure --enable-memalign-hack --arch=x86_64 --target-os=mingw64 --cross-prefix=mingw-w64-x86_64 --pkg-config=pkg-config --prefix=/home/u64/builds/test1
[02:46:28 CEST] <electronrotoscop> mingw-w64-x86_64gcc is unable to create an executable file.
[02:46:33 CEST] <electronrotoscop> maybe something else is wrong?
[02:48:43 CEST] <BtbN> The missing - is wrong
[02:50:41 CEST] <electronrotoscop> BtbN: could you elaborate? Or are you referring to something else? sorry I don't understand
[02:50:51 CEST] <BtbN> mingw-w64-x86_64gcc
[02:50:55 CEST] <BtbN> is clearly missing the last -
[02:51:19 CEST] <electronrotoscop> ./configure --enable-memalign-hack --arch=x86_64 --target-os=mingw64 --cross-prefix=mingw-w64-x86_64- --pkg-config=pkg-config --prefix=/home/u64/builds/test1
[02:51:19 CEST] <electronrotoscop> mingw-w64-x86_64-gcc is unable to create an executable file.
[02:51:19 CEST] <electronrotoscop> C compiler test failed.
[02:51:31 CEST] <electronrotoscop> adding a final - to cross-prefix doesn't seem to help
[02:51:55 CEST] <BtbN> Is thas actualy the name of the compiler binary?
[02:52:03 CEST] <BtbN> And is it in your PATH?
[02:52:11 CEST] <Raptors> Hey guys
[02:52:21 CEST] <Raptors> Is there any video format that you can save to that is editable while record?
[02:52:33 CEST] <BtbN> editable?
[02:52:57 CEST] <Raptors> Like I can open it up in a video player or avidmuxer/other video editing software
[02:53:11 CEST] <BtbN> That's not possible.
[02:53:11 CEST] <electronrotoscop> Raptors: I've heard that Amberfin's old iCR software could do that with an MXF that wrote frames that you could access while it was recording, but it's unusual
[02:53:28 CEST] <Raptors> k
[02:53:29 CEST] <BtbN> You can certainly play streamable formats while still appending to the end, but not do arbitrary edits.
[02:53:48 CEST] <Raptors> Like I just want to grab clips from what is already recorded
[02:53:52 CEST] <Raptors> not make full on edits
[02:54:00 CEST] <electronrotoscop> Raptors: also of course an image sequence based editor like a Quantel can do that sort of thing. It requires a hell of a raid though
[02:54:17 CEST] <BtbN> Just record in segments?
[02:54:34 CEST] <BtbN> a bunch of 30 minute segments, and once a segment is done, you can do whatever you like with it.
[02:55:10 CEST] <Raptors> ya, that was what I was planning on doing if there is no other option. It sucks you can't just open up what's recorded already
[02:55:17 CEST] <Raptors> k I guess I'll just do that
[02:55:20 CEST] <Raptors> thanks
[02:55:26 CEST] <electronrotoscop> Depending on the format you can take a copy as it records
[02:55:45 CEST] <Raptors> hmm
[02:55:50 CEST] <Raptors> That is smart
[02:55:53 CEST] <electronrotoscop> something all-i-frame like an uncompressed avi or DNxHD sometimes works. It's always tricky though
[02:55:53 CEST] <Raptors> I'll try that
[02:56:22 CEST] <electronrotoscop> andsometimes you have to take the copy, remux in ffmpeg or vlc, then you can actually load it into your NLE
[02:56:48 CEST] <electronrotoscop> BtbN it looks like x86_64-w64-mingw32- was for whatever reason the magic word, thanks for the help
[02:57:22 CEST] <electronrotoscop> I looked in the /bin directory that the build script made and it had a x86_64-w64-mingw32-gcc in it
[02:57:34 CEST] <electronrotoscop> now it's telling me Unknown OS 'mingw64'
[02:57:45 CEST] <electronrotoscop> ha ha sorry can you tell I'm incredibly new to the whole compiling thing
[02:57:53 CEST] <BtbN> mingw32
[02:58:00 CEST] <BtbN> should work
[02:58:03 CEST] <BtbN> or just win32
[02:58:09 CEST] <electronrotoscop> even for a 64-bit everything?
[02:58:21 CEST] <BtbN> mingw64 and mingw32 can both be both.
[02:58:52 CEST] <electronrotoscop> sweet that made it not throw up an immedate error at least!!
[02:59:10 CEST] <electronrotoscop> I guess I am still saying arch=x86_64
[03:01:47 CEST] <Raptors> Is there a way to tell FFMPEG to just detect the quality of the stream and match it?
[03:02:01 CEST] <Raptors> The man page is blowing my mind
[03:02:05 CEST] <Raptors> so many options
[03:02:34 CEST] <c_14> In short, no
[03:02:43 CEST] <electronrotoscop> for x264 to make an output stream that looks the same as the input stream the best way I know is crf 0 but that's probably not what you want
[03:03:05 CEST] <Raptors> it's a rtmp stream
[03:03:09 CEST] <Raptors> idk if that helps
[03:03:21 CEST] <Raptors> I'm using livestreamer
[03:03:31 CEST] <Raptors> I guess I'll try vcodec copy
[03:03:47 CEST] <electronrotoscop> it's certainly the easiest solution, if it works!
[03:03:56 CEST] <electronrotoscop> lower load on the cpu that way ha ha
[03:04:01 CEST] <BtbN> Well, if a stream copy is ok, that will of course give you the exact same quality.
[03:05:55 CEST] <electronrotoscop> how can I tell that a final binary is 64bit or 32bit without stress testing it?
[03:06:46 CEST] <c_14> electronrotoscop: `file binary'
[03:08:54 CEST] <electronrotoscop> what if it's in windows ha ha
[03:09:29 CEST] <c_14> You're cross-compiling anyway, right? Can't you just check before you copy it over to Windows?
[03:09:43 CEST] <electronrotoscop> oh sure, if that works on exes yeah for sure. I'll give it a shot
[03:16:29 CEST] <Psi-Jack> I have an odd issue I don't understand. ffprobe -i "filename.flac" -show_format 2>/dev/null | sed -nr -e 's/TAG:(.*)=(.*)/\1 \2/p' -- Shows me various important metadata has odd information on a particular batch of files, and I don't see why. for TITLE, I get Track01;Song Name, for ARTIST I get Unknown Artist;Band Name, for ALBUM, I get Unknown Title;Album Name, where the data after the ; is correct, but Unknown * is obviously not correct.
[03:17:00 CEST] <Psi-Jack> I'm editing the metadata with Kid3, but I see no reason why it's doing this.
[03:18:49 CEST] <Psi-Jack> Heh, even DISC is 1/1;1 ...
[03:25:19 CEST] <c_14> Psi-Jack: what version do you have?
[03:25:39 CEST] <Psi-Jack> 2.4.7
[03:25:58 CEST] <c_14> Can you try with a git version?
[03:26:09 CEST] <c_14> Maybe with a static build
[03:26:11 CEST] <c_14> http://johnvansickle.com/ffmpeg/
[03:26:50 CEST] <Psi-Jack> This is only happening with one batch of flac files, so far, but yeah, I can test it.
[03:27:15 CEST] <c_14> Yeah, it might be what I think it is.
[03:27:50 CEST] <Psi-Jack> Still doing it in that linked version.
[03:28:00 CEST] <c_14> mhm
[03:28:36 CEST] <c_14> did that patch ever get applied&
[03:30:04 CEST] <Psi-Jack> What's the actual problem? I mean it acts like it's reading from multiple tag containers, but there's only the vorbis container tag information that I can see.
[03:31:23 CEST] <c_14> flac files can actually contain both ID3 tags and Vorbis tags.
[03:31:43 CEST] <c_14> And since the patch was never applied (to the best of my knowledge) it reads both.
[03:33:09 CEST] <Psi-Jack> Hmm
[03:33:23 CEST] <c_14> https://trac.ffmpeg.org/ticket/3799
[03:33:48 CEST] <Psi-Jack> Well, Kid3 shows me 2 sets of tags, but 1 is empty, 2 is all the Vorbis tags.
[03:35:09 CEST] <Psi-Jack> But yeah.. id3v2 does confirm there's that.
[03:35:40 CEST] <Raptors> what do I do about this error "Codec for stream x does not user global header but container format requires global header?
[03:35:43 CEST] <Raptors> what do I do about this error "Codec for stream x does not user global header but container format requires global header"?
[03:36:09 CEST] <c_14> Psi-Jack: https://ffmpeg.org/pipermail/ffmpeg-devel/2015-February/168509.html <- thread with patch
[03:36:25 CEST] <Raptors> I am using option -flags -global_header
[03:36:28 CEST] <c_14> If you feel up to it, you can try reviving it with user input, ie saying what you would expect to occur.
[03:36:28 CEST] <Raptors> but I still get that
[03:36:35 CEST] <c_14> Raptors: -fflags iirc
[03:36:50 CEST] <c_14> And you need +global_header
[03:37:04 CEST] <Raptors> so I just add +global header?
[03:37:11 CEST] <c_14> -fflags +global_header
[03:37:36 CEST] <Psi-Jack> Hmm, well, for now, I can at least fix this issue by removing the id3 tags. :)
[03:39:08 CEST] <Raptors> c_14, http://i.imgur.com/xsz2hzo.png
[03:39:13 CEST] <Raptors> :/
[03:39:31 CEST] <c_14> Psi-Jack: If you won't ping the thread, I might. See what the holdup is, etc.
[03:40:16 CEST] <c_14> Raptors: eh, I was wrong. It is -flags +global_header
[03:40:48 CEST] <Raptors> thanks
[04:26:35 CEST] <Prelude2004c> hey, anyone knows the best way to keep sound in sync?
[04:26:39 CEST] <Prelude2004c> my stuff keeps drifting
[04:27:43 CEST] <electronrotoscop> usually when I have that problem it's because I wrote out the audio and video into elemental files instead of .mov or .mp4 containers
[04:28:59 CEST] <Prelude2004c> i am outputting to HLS
[04:31:50 CEST] <electronrotoscop> ah
[04:52:21 CEST] <richard> name is richard
[04:54:16 CEST] <Guest3872> If I am looking to cut some time off of a video, what command is used?
[04:56:50 CEST] <electronrotoscop> do you want to start partway into the video or do you want to stop before it's reached the end?
[04:58:11 CEST] <electronrotoscop> cutting time off of the beginnig of the video is usually refered to as seeking and it's the -ss flag, cutting time off the end is usually referred to as setting the duration and it's the -t flag. Guide here https://trac.ffmpeg.org/wiki/Seeking
[05:24:34 CEST] <WickedP> Happy Easter folks! I'm trying to build ffmpeg into a c++ plugin but am having a difficult time getting it to work. The software builds (vs 2010 Exp. 32bit) and is using the static lib links. The software appears to build fine, but as soon as I start using ffmpeg, e.g. av_register_all(), the software fails: I get the busy mouse icon, the main working thread seems to get caught in an endless loop an
[05:24:34 CEST] <WickedP> d never does anything. Any ideas?
[12:56:20 CEST] <xintron> What codec would you guys recommend for HTTP streaming where I need to do transcoding on the input source and get the bitrate down without loosing too much quality? VP9 + vorbis? (client support is not an issue)
[12:57:12 CEST] <Mavrik> H.264 + AAC of course
[12:57:27 CEST] <Mavrik> unless you're rolling in piles of CPU power and free electricity
[12:57:38 CEST] <__jack__> if you are not restricted by tech or client codec: x265 / vp9
[12:57:41 CEST] <Mavrik> and love hearing your clients fans spin up and batteries die
[13:00:03 CEST] <loki_> hello, where i can read about adding deps for system lib in ffmpeg?
[13:00:05 CEST] <xintron> Been running h264 + aac a while and been quite happy with that but thought I should look into alternatives (as bandwidth could become a problem going forward)
[13:00:17 CEST] <loki_> i need at to linkage cudart.so
[13:00:26 CEST] <loki_> at -> add
[13:00:26 CEST] <xintron> But it seems like vp9 could probably be too slow for what I need
[13:00:55 CEST] <Mavrik> xintron, both VP9 and x265 are still very slow encoders
[13:01:11 CEST] <Mavrik> xintron, and you can usually do way better by switching for slower x264 profile for streaming
[13:01:17 CEST] <Mavrik> unless you really have a huge CPU overhead
[13:01:40 CEST] <Mavrik> also having HW acceleration for display on target usually hugely helps
[13:01:41 CEST] <xintron> My plan is to see if I can get something running ok on a Raspberry PI 2.
[13:01:47 CEST] <Mavrik> ugh
[13:01:54 CEST] <xintron> Just as a test :)
[13:02:06 CEST] <Mavrik> you'll get like 1 frame per 5 seconds on that with x265
[13:02:18 CEST] <Mavrik> xintron, use HW encoder because everything else will be prohibitevly slow
[13:02:47 CEST] <loki_> lol v9 + vorbis on pi
[13:05:46 CEST] <__jack__> he said: pi2, isn't that one with multicore @2GHz or something ? I guess it can decode all that stuff
[13:06:33 CEST] <Mavrik> __jack__, huh, no
[13:06:57 CEST] <xintron> 900MHz quad core (so not the beefiest machine, no)
[13:06:59 CEST] <Mavrik> it's a 900MHz quad ARMv7
[13:07:50 CEST] <Mavrik> __jack__, it's still order of magnitude slower than anything Intel stamped
[13:10:59 CEST] <__jack__> owh
[13:11:29 CEST] <Mavrik> I think new Intels do like 2-3x as much per clock as ARMv7
[13:16:19 CEST] <rcombs> rip can decode H.264 to some degree, but I wouldn't rely on it
[13:16:36 CEST] <rcombs> *rpi2
[13:16:49 CEST] <rcombs> H.265, definitely wouldn't bet on it
[13:20:53 CEST] <loki_> guys, where i can ask about ffmpeg dev?
[13:21:00 CEST] <loki_> where is better place for this?
[13:21:42 CEST] <__jack__> loki_: #ffmpeg-devel
[13:21:58 CEST] <__jack__> or the mailing-list
[15:20:26 CEST] <mrkie> hey
[15:21:49 CEST] <mrkie> i have a question, i use subtitle workshop and i have a bunch of mp4 files, i set each mp4 file as being a movie fragment, now i merge all subtitles of each mp4 file behind it to one subtitle file, are the time values only calculated of my mp4 files in seconds as being movie fragments or also as being in seconds and miliseconds as being moviefragments?
[15:23:24 CEST] <mrkie> i ask this because i converted mp4 files as being movie fragments to avi files as being moviefragments, my subtitle workshop only reads avi files as being movie fragments, after the reencoding the avi don't have the same length in ms as being new moviefragments
[15:33:42 CEST] <kate_12> Hi,
[15:34:07 CEST] <kate_12> I am trying to concatenate two mp4 files. One has video, the other has both video and audio.
[15:34:15 CEST] <kate_12> Here is my CMD output
[15:34:16 CEST] <kate_12> http://pastebin.com/eXTCnwHE
[15:34:57 CEST] <kate_12> In short, I am using the following command ffmpeg -i 5.mp4 -i 6.mp4 -filter_ complex "[0:0][1:0][1:1] concat=n=2:v=1:a=1 [v][a]" -map "[v]" -map "[a]" output .mp4 and getting this error: Stream specifier ':0' in filtergraph description [0:0][1:0][1:1] concat=n=2:v=1: a=1 [v][a] matches no streams.
[15:43:44 CEST] <kate_12> Anyone?
[15:43:49 CEST] <kate_12> Please help...
[15:43:54 CEST] <kate_12> Been stuck at it for hours now.
[15:59:04 CEST] <mrkie> ffmpeg -i file.mp4 > output.txt
[15:59:08 CEST] <mrkie> why that is not working?
[16:11:09 CEST] <__jack__> mrkie: what means "not working" ?
[16:11:54 CEST] <__jack__> mrkie: maybe you want that: ffmpeg -i file.mp4 &> output.txt
[16:12:19 CEST] <__jack__> ffmpeg -i file.mp4 2> output.txt if you prefere
[16:13:42 CEST] <kate_12> got anything for me, jack?
[16:19:39 CEST] <mrkie> __jack__: i already found out how i use 2> output.extension
[16:19:45 CEST] <__jack__> kate_12: hum, what do you expect as a result ? no sound for the first part of the result ? or repeated sound ?
[16:21:49 CEST] <mrkie> how can i calculate 00:20:04.51 to seconds like x.51
[16:21:58 CEST] <kate_12> no sound.
[16:22:14 CEST] <kate_12> The first file has only the video stream while the second one has both,
[16:22:35 CEST] <mrkie> i need a formula for that
[16:23:10 CEST] <kate_12> both audio and video
[16:23:12 CEST] <klaxa> you can use 00:20:04.51 in that format for ffmpeg
[16:23:50 CEST] <mrkie> yes, but i use Google2SRT and i need the moviefragement in seconds according this page: http://google2srt.sourceforge.net/en/help.html
[16:24:16 CEST] <klaxa> if you need to calculate it to seconds, a minute has 60 seconds, if you multiply the minutes by 60 seconds you get the total seconds
[16:24:19 CEST] <mrkie> i have used ffmpeg to get the duration of each moviefragment, now i need to recalculate all my durations to seconds
[16:25:09 CEST] <mrkie> klaxa: do you have a formula for that?
[16:25:24 CEST] <klaxa> for example 00:20:04.51 is 20 * 60 + 4.51 = 1200 + 4.51 = 1204.51 seconds
[16:25:46 CEST] <mrkie> i know
[16:25:59 CEST] <klaxa> good
[16:26:05 CEST] <mrkie> but i need a formula for my sheet program (excel or libreoffice calc) :)
[16:26:09 CEST] <klaxa> that's your formula if you want to call it like that
[16:26:27 CEST] <mrkie> i really have a bunch of durations
[16:26:44 CEST] <klaxa> mmh i'm really not that familiar with spreadsheet programs
[16:27:03 CEST] <klaxa> maybe google, this is also no longer ffmpeg-related really
[16:28:21 CEST] <kate_12> try this
[16:28:22 CEST] <kate_12> HOUR(A1)*60+MINUTE(A1)+SECOND(A1)/60
[16:31:45 CEST] <mrkie> kate: i did, not working
[16:32:33 CEST] <kate_12> I'll post again. How can I concatenate two media files with different number of streams? File 1 has only the video stream while file 2 has both audio and video streams. I want no sound while the first part plays. Here is what I have tried"
[16:32:47 CEST] <kate_12> http://pastebin.com/eXTCnwHE
[16:32:56 CEST] <kate_12> Are you using excel?
[16:33:01 CEST] <kate_12> It works in excel.
[16:33:40 CEST] <mrkie> =UUR(B2)*60+MINUUT(B2)+SECONDE(B2)/60
[16:33:42 CEST] <mrkie> i did
[16:33:45 CEST] <kate_12> Updated paste. http://pastebin.com/rU6LK63c
[16:33:46 CEST] <mrkie> since i have a dutch excel version
[16:33:58 CEST] <__jack__> kate_12: found a not awesome but still working solution
[16:34:13 CEST] <__jack__> add a silent stream :D
[16:34:32 CEST] <__jack__> ffmpeg -f lavfi -t 00:00:20 -i aevalsrc=0 -i video.mkv  -i both.mkv -filter_complex '[1:0][0:0][2:0][2:1] concat=n=2:v=1:a=1 [v][a]' -map "[v]" -map "[a]" output.mkv
[16:34:46 CEST] <__jack__> (need the duration of the video-only file)
[16:35:44 CEST] <kate_12> Ah! I was trying to avoid that. But looks like that's the only way. Anyways, thanks jack. much love.
[16:35:55 CEST] <mrkie> kate_12: libre office calc also doesn't work
[16:36:09 CEST] <kate_12> mrkie, my bad.
[16:36:17 CEST] <kate_12> you have minutes, seconds and microseconds.
[16:36:25 CEST] <kate_12> so, the formula will be different
[16:36:27 CEST] <kate_12> give me a sec.
[16:36:53 CEST] <mrkie> sure
[16:43:16 CEST] <kate_12> I don't understand why the earlier formula didn't work.
[16:43:25 CEST] <kate_12> Here is a more expanded but cumbersome formula
[16:43:26 CEST] <kate_12> =mid(b2,7,2)+(mid(b2,4,2)*60)+(left(a2,2)*3600)
[16:43:32 CEST] <kate_12> It should help.
[16:45:50 CEST] <mrkie> kate_12: did you test it? because i don't see a result coming up
[16:46:49 CEST] <mrkie> #VALUE! it shows up
[16:48:21 CEST] <mrkie> you also use a2 i see now
[16:48:24 CEST] <mrkie> i only have b2 input
[17:26:28 CEST] <kate_12> yeah, I might have got it wrong.
[17:26:31 CEST] <kate_12> Works fine now?
[17:35:22 CEST] <mrkie> kate_12: did a party manually, it wouldn't work here, atleast your formula woudln't work at my excel
[17:54:38 CEST] <johnny_|_> hi, how do I compile from source ffmpeg with libfaac? Where to get libfaac-dev? I am trying to install either on debian or kali linux.
[17:55:47 CEST] <klaxa> apparently you need the debian-multimedia repos to get libfaac (http://crunchbang.org/forums/viewtopic.php?id=23488)
[17:56:09 CEST] <johnny_|_> will this work with kali linux as well?
[17:56:18 CEST] <klaxa> you can substitute libfaac with libfdk-aac though
[17:56:46 CEST] <klaxa> which seems to be in the default repositories
[17:57:20 CEST] <klaxa> and afaik is currently the best option to encode aac with ffmpeg
[18:01:52 CEST] <__jack__> libfaac is in the repo too
[18:02:08 CEST] <johnny_|_> but not for kali linux it seems
[20:18:47 CEST] <seasc> __jack__, if you're on a rpm based distro, visit rpmfusion.org
[20:20:44 CEST] <seasc> n/m seen its deb. idk4t
[22:18:02 CEST] <faLUCE> Hello, I have two pictures (jpeg) and a mp3 music file. How can I make a very simple video with these two pictures + music ? The first picture must show from the beginning to 3 minutes and 2 seconds, while the second from 3 minutes and 2 seconds to the end of the music...
[22:19:26 CEST] <c_14> ffmpeg -framerate 1/182 -i one.jpg -loop 1 two.jpg -i mp3 -c:a copy out.mkv
[22:19:28 CEST] <c_14> might work
[22:19:48 CEST] <c_14> Might need to use concat+trim
[22:20:53 CEST] <faLUCE> c_14: tanks, but how can I specify 3 minutes and 2 seconds in your command?
[22:21:10 CEST] <c_14> That's (hopefully) what the framerate is for
[22:21:49 CEST] <faLUCE> c_14: many thanks let's try
[22:27:19 CEST] <DragonsLordKERNE> Hello to everybody
[22:28:01 CEST] <DragonsLordKERNE> I hope someone can help me ... How can I do a frame-accurate cutting of a .MP4 file?
[22:28:16 CEST] <c_14> cuts are frame-accurate by default unless you use -codec copy
[22:28:48 CEST] <DragonsLordKERNE> Yes, I need to copy ... I mean no re-encoding
[22:29:03 CEST] <DragonsLordKERNE> From my understanding a need to choose one I-frame
[22:29:06 CEST] <c_14> Then it isn't possible.
[22:29:07 CEST] <DragonsLordKERNE> Is it true?
[22:29:10 CEST] <c_14> yes
[22:29:22 CEST] <c_14> With codec copy, cuts are to the nearest i-frame
[22:29:26 CEST] <DragonsLordKERNE> Now my question is ... How can I detect an I-frame?
[22:30:15 CEST] <mrkie> hey
[22:30:17 CEST] <DragonsLordKERNE> (thank you, c_14)
[22:30:25 CEST] <DragonsLordKERNE> Hi mrkie
[22:30:43 CEST] <mrkie> Google2SRT is really a fine tool to work with
[22:31:05 CEST] <c_14> DragonsLordKERNE: probably showinfo and look at the type
[22:34:02 CEST] <DragonsLordKERNE> c_14, do you know who is responsible to set the I-frame interval?
[22:34:11 CEST] <c_14> The encoder
[22:35:02 CEST] <DragonsLordKERNE> c_14, I'm asking this because I use HD PVR Rocket to record from and HDMI source ... and I-frames seems to be too far between themselves
[22:35:31 CEST] <c_14> Look for a gop size (group of pictures) or something about keyframes
[22:35:39 CEST] <DragonsLordKERNE> But is it something related to the FPS, to the format ... or it's "my" choice when I encode one file?
[22:36:04 CEST] <c_14> It's "your" choice when you encode a file (in most cases)
[22:38:07 CEST] <DragonsLordKERNE> Do you think MediaInfo should tell me something more about the I-frame interval?
[22:38:54 CEST] <c_14> I have no clue. If you could upload the output to a pastebin site, I might be able to pick something out.
[22:42:58 CEST] <DragonsLordKERNE> c_14, here it is > http://pastebin.com/svQkWYSD <
[22:43:03 CEST] <DragonsLordKERNE> Thank you in advance
[22:44:52 CEST] <c_14> Looks like an around 50 frame gop
[22:45:02 CEST] <c_14> at 25 fps that's once every 2 seconds
[22:46:13 CEST] <DragonsLordKERNE> Then you looked at this ...
[22:46:16 CEST] <DragonsLordKERNE> > Format settings, GOP                     : M=1, N=50 <
[22:46:26 CEST] <c_14> yep
[22:46:29 CEST] <DragonsLordKERNE> and this
[22:46:32 CEST] <DragonsLordKERNE> > Frame rate                               : 25.000 fps <
[22:47:06 CEST] <DragonsLordKERNE> N = 50 means one I-frame each 50 frames
[22:47:21 CEST] <DragonsLordKERNE> with 25 fps it means 25x2 so 2 seconds
[22:47:22 CEST] <faLUCE> c_14: ffmpeg -framerate 1/119 -i san_pietroburgo1.jpg -loop 1 -i terzetto.jpg -i 14.mp3 -c:a copy 14.avi   ----> the command doesn't stop, and it adds only san_pietroburgo.jpg to the video
[22:47:23 CEST] <DragonsLordKERNE> right?
[22:48:30 CEST] <c_14> DragonsLordKERNE: ye
[22:48:37 CEST] <c_14> faLUCE: right, add -shortest
[22:48:41 CEST] <c_14> totally forgot about that
[22:48:44 CEST] <c_14> as an output option
[22:48:44 CEST] <faLUCE> c_14: where?
[22:48:53 CEST] <c_14> (before 14.avi, after 14.mp3
[22:48:55 CEST] <c_14> )
[22:49:45 CEST] <DragonsLordKERNE> c_14, so the best request to do vs the supplier of my HDMI recorder should be to leave the user set the GOP size?
[22:54:15 CEST] <c_14> That, or get it to produce intra-only output
[22:54:47 CEST] <DragonsLordKERNE> c_14, could you please elaborate on this?
[22:54:57 CEST] <DragonsLordKERNE> I did not understand :(
[22:55:15 CEST] <c_14> With intra-only encoding, every single frame is an i-frame.
[22:57:27 CEST] <faLUCE> c_14: now the video is not endless, but it adds only the second picture to the music
[22:58:36 CEST] <c_14> Ok, will probably have to go with the trim route then, although lemme see: ffmpeg -loop 1 -t 3:02 1.jpg -loop 1 -i 2.jpg -i mp3 -shortest -c:a copy out.avi
[22:58:38 CEST] <c_14> try that
[22:58:39 CEST] <DragonsLordKERNE> c_14 ... does it have any cons? (I don't know ... extra-space needed for the output or something like this?)
[22:58:47 CEST] <c_14> Extra space needed
[23:00:12 CEST] <DragonsLordKERNE> More or less ... how much?
[23:00:15 CEST] <DragonsLordKERNE> 10% more?
[23:00:18 CEST] <DragonsLordKERNE> 50% more?
[23:01:21 CEST] <c_14> 10x (on my system on one test video)
[23:02:47 CEST] <DragonsLordKERNE> Do you mean 10 GB output instead of 1 GB?
[23:03:14 CEST] <c_14> In this case 22M instead of 2.7M, so yes.
[23:03:27 CEST] <DragonsLordKERNE> O_o
[23:03:31 CEST] <faLUCE> thanks c_14 lets try
[23:03:56 CEST] <DragonsLordKERNE> They will never accept my request for enhancement :'(
[23:09:37 CEST] <DragonsLordKERNE> c_14, do you often use FFMpeg?
[23:09:43 CEST] <c_14> relatively
[23:10:00 CEST] <DragonsLordKERNE> Do you use it from the command line
[23:10:05 CEST] <c_14> yep
[23:10:12 CEST] <DragonsLordKERNE> O_o
[23:11:45 CEST] <DragonsLordKERNE> Do you know how I can convert an MP4 into an AVI (XviD)?
[23:11:59 CEST] <c_14> Is the mp4 xvid or something else?
[23:12:07 CEST] <DragonsLordKERNE> This is what I "built":
[23:12:07 CEST] <DragonsLordKERNE> ffmpeg -i input.mp4 -codec:v libxvid -vtag DIVX -vf fps=fps=25 -codec:a libmp3lame -b:a 128k -ar 48000 -ac 2 -joint_stereo 0 output.avi
[23:12:35 CEST] <DragonsLordKERNE> input mp4 is the one I pastbinned before
[23:12:48 CEST] <DragonsLordKERNE> it shoud be H.264 + AAC into an MP4 container
[23:12:50 CEST] <c_14> That looks fine. Might want to set a visual quality though, also you probably don't need the fps filter
[23:13:18 CEST] <DragonsLordKERNE> I would like to do a dual bass
[23:13:32 CEST] <c_14> A wat?
[23:13:35 CEST] <c_14> *what
[23:13:43 CEST] <DragonsLordKERNE> I would like to do a dual *pass recoding
[23:13:55 CEST] <c_14> https://trac.ffmpeg.org/wiki/Encode/MPEG-4
[23:13:58 CEST] <c_14> near the bottom
[23:13:58 CEST] <DragonsLordKERNE> 1500 kbps of variable bitrate
[23:14:42 CEST] <DragonsLordKERNE> Well, I have already found that
[23:14:43 CEST] <DragonsLordKERNE> But my issue is
[23:15:02 CEST] <DragonsLordKERNE> ffmpeg -y -i input.avi -c:v mpeg4 -vtag xvid -b:v 555k -pass 1 -an -f avi /dev/null > should not produce something to elaborate during the second pass?
[23:15:20 CEST] <c_14> It creates a passlogfile
[23:15:25 CEST] <c_14> in cwd
[23:17:23 CEST] <DragonsLordKERNE> so my instructions should became
[23:17:29 CEST] <DragonsLordKERNE> ffmpeg -y -i input.mp4 -codec:v libxvid -vtag DIVX -vf fps=fps=25 -b:v 555k -pass 1 -an -f avi /dev/null?
[23:18:01 CEST] <ChocolateArmpits> Are you using linux or windows?
[23:18:24 CEST] <DragonsLordKERNE> Windows ChocolateArmpits, at the moment (Hi!)
[23:18:39 CEST] <ChocolateArmpits> then replace "/dev/null" with "NUL"
[23:18:51 CEST] <DragonsLordKERNE> YEAH, I have just read thank you
[23:19:13 CEST] <ChocolateArmpits> do you need to fix the framerate too with the fps command ?
[23:19:28 CEST] <DragonsLordKERNE> Yes, I need
[23:19:33 CEST] <ChocolateArmpits> k then that's good
[23:19:57 CEST] <DragonsLordKERNE> And I also would like to set Motion search to 6 - High (for Xvid)
[23:20:18 CEST] <ChocolateArmpits> just don't change the directory after the 1st pass because the passlogfile will be saved to your current directory
[23:20:38 CEST] <DragonsLordKERNE> And I also would like to set VHQ Mode to 4 - Wide search (for Xvid)
[23:20:44 CEST] <ChocolateArmpits> you can manually point where to save it via "-passlogfile" command
[23:21:34 CEST] <DragonsLordKERNE> I am basically trying to replace Xmedia Recode with FFmpeg because probably Xmedia Recode did a wrong "implementation" of FFMPEG
[23:21:45 CEST] <DragonsLordKERNE> and the resulting output is a bit out of sync
[23:22:03 CEST] <DragonsLordKERNE> while with AVIdemux the sync is nearest to the source
[23:22:14 CEST] <DragonsLordKERNE> so I would like to give FFMPEG a try
[23:24:32 CEST] <DragonsLordKERNE> Does anyone know how I can set "Motion search" and "VHQ Mode" for XviD within Xmedia Recode?
[23:25:23 CEST] <c_14> DragonsLordKERNE: -me_method full -mbd 2
[23:26:44 CEST] <DragonsLordKERNE> Is full = Ultra High (6) for Motion search precision
[23:26:45 CEST] <DragonsLordKERNE> and
[23:27:21 CEST] <DragonsLordKERNE> Is mbd = 2 Wide search (4) for VHQ Mode
[23:27:26 CEST] <DragonsLordKERNE> as far as you know?
[23:28:08 CEST] <c_14> As far as I know, yes
[23:36:01 CEST] <DragonsLordKERNE> c_14, are you lookig at some documentation I can also look at?
[23:36:55 CEST] <c_14> I got the full from the -codecs manpage: https://ffmpeg.org/ffmpeg-codecs.html#libxvid
[23:37:03 CEST] <c_14> And the other one from here: https://ffmpeg.org/pipermail/ffmpeg-user/2013-March/013938.html
[23:37:08 CEST] <c_14> about halfway down
[23:44:55 CEST] <DragonsLordKERNE> What about setting Quantization type to H.263 instead of MPEG?
[23:47:03 CEST] <c_14> -mpeg_quant 0 (I'm pretty sure it's off by default and I'm not sure that syntax is valid)
[23:48:07 CEST] <DragonsLordKERNE> Chroma motion estimation is -cmp chroma right?
[23:48:48 CEST] <c_14> ye, but I think -me_method full already enables that
[23:49:29 CEST] <DragonsLordKERNE> and to set max_bframes to 0 is -bf 0?
[23:50:28 CEST] <c_14> probably
[23:50:37 CEST] <DragonsLordKERNE> I will try
[23:51:25 CEST] <DragonsLordKERNE> Do you also know how to set the "famous" Keyframe interval to 250?
[23:52:20 CEST] <c_14> -g 250
[23:52:58 CEST] <DragonsLordKERNE> is the order important?
[23:53:14 CEST] <c_14> As long as it's an output option, no.
[23:56:02 CEST] <DragonsLordKERNE> Probably I should use
[23:56:08 CEST] <DragonsLordKERNE> -trellis 1
[23:56:10 CEST] <DragonsLordKERNE> instead of
[23:56:11 CEST] <DragonsLordKERNE> -mpeg_quant 0
[23:58:10 CEST] <DragonsLordKERNE> -af 'volume=4' should increase the volume by a factor of 4, right?'
[23:58:20 CEST] <DragonsLordKERNE> I mean 4X higher?
[00:00:00 CEST] --- Mon Apr  6 2015


More information about the Ffmpeg-devel-irc mailing list