[Ffmpeg-devel-irc] ffmpeg.log.20131021

burek burek021 at gmail.com
Tue Oct 22 02:05:01 CEST 2013


[00:41] <x__> Hey all. Is there a way I can convert my .mov 1.9 gb files to a smaller mp4 format? like around 700 mb for a two our video?
[00:42] <x__> I tried to use ffmpeg -i input.avi -acodec libfaac -b:a 128k -vcodec mpeg4 -b:v 1200k -flags +aic+mv4 output.mp4 but it isn't any smaller
[00:43] <zennist> JEER: I've now got the resampling function to work. But interestingly only one channel outputs correctly, and the other channel is still noise...
[00:44] <x__> I tried to run it on an AVI file, but it come out bigger?
[01:54] <SSlater> G'day all.
[01:55] <SSlater> I have a question about resampling video. Particularly which options are best to use in ffmpeg.
[01:58] <SSlater> I have security camera video recorded from rtsp stream to mpg4 files at 25fps.  Each day's file is about 30G. The thought was to reduce the file size each day by reducing the framerate, then archiving when down to 1fps.
[01:59] <SSlater> The command is: ffmpeg -i test0.mpg4 -r 20fps test0-20.mpg4
[02:01] <SSlater> I am getting messages about headers not complete; picture size invalid; invalid dts/pts combination and Unable to find a suitable output format for 'test0-20.mpg4'.
[02:02] <SSlater> Is something wrong with the input file?  It plays with audio fine.
[02:56] <SSlater> First problem solved. Changed the command to: ffmpeg -i test0.mpg4 -r 20 test0-20.mp4.
[03:16] <poiupoiu> good evening everyone. I was hoping someone more familiar in linux can help me with a problem...
[03:16] <poiupoiu> I've compiled ffmpeg from the commands given at https://trac.ffmpeg.org/wiki/UbuntuCompilationGuide
[03:17] <poiupoiu> a script runs perfectly fine as my username, but bombs when i run it as another user.
[03:17] <poiupoiu> poke around for a bit and find that as myself, it is using the compiled version, as other users it is using the avconv "ffmpeg"
[03:19] <poiupoiu> does anyone know commands or has a chance to give me a hand with solving this?
[03:21] <poiupoiu> http://askubuntu.com/questions/313217/install-latest-ffmpeg-for-all-users-in-debina-ubuntu <-- I think i found my solution...
[03:58] <poiupoiu> And that got it. Thanks for letting me drop in for a minute!
[07:56] <hans_henrik> ./ffmpeg -i ../../www/writhing.wmv -c:v libx264 pipe:0
[07:56] <hans_henrik> Unable to find a suitable output format for 'pipe:0'
[07:56] <hans_henrik> pipe:0: Invalid argument
[07:57] <JEEB> hans_henrik, you need to specify a container format
[07:57] <JEEB> -f container_of_your_choice
[08:02] <hans_henrik> JEEB, yup that was it! thanks
[08:04] <hans_henrik> can i tell ffmpeg to be otherwise, completely quiet?
[08:04] <hans_henrik> except the output file, which is piped to stdout
[08:04] <hans_henrik> .. or, tell it to pipe all the stuff it normally puts on stdout, to stderr?
[08:04] <hans_henrik> or
[08:05] <hans_henrik> um
[08:05] <hans_henrik> this doesnt have to be anywhere near foolproof, i guess i can just pipe it to stderr
[08:41] <xingchao_> why does av_seek_frame() return non-key frame even i donot set *ANY flag?
[08:41] <xingchao_> i thought it should return the key frame anyway
[08:41] <xingchao_> does anyone know that root cause?
[09:56] <zennist> I'm using a custom AVIO with my AVInputFormat and I found that this causes avformat_open_input to hang on some audio streams i.e. m4a
[09:56] <zennist> Anyone has any idea?
[09:57] <zennist> FFMpeg's default io implementation doesn't cause any problems though
[10:06] <shur> http://stackoverflow.com/questions/19488762
[11:27] <GeorgesLeYeti> Hi
[11:29] <GeorgesLeYeti> I encounter a problem for converting files with ffmpeg.
[11:29] <GeorgesLeYeti> When i try to convert my file on my server it returns me: moov atom not found
[11:30] <sacarasc> That sounds like you have an incomplete MP4 file.
[11:30] <GeorgesLeYeti> But the thing is that in my local machine (with exact same ffmpeg version and file) it works
[11:31] <GeorgesLeYeti> sacarasc: I copy the file from my server to my computer
[11:33] <GeorgesLeYeti> this mean if the file is corrupted on my server it still should be corrupted on my computer
[11:58] <Trax|wrk> maybe the copy is a download or it's an open file, try comparing both files with md5 checksum then you'lre sure they are the same
[12:49] <Macey> hi all, the windows builds of ffmpeg, how different is it to the Ubuntu 13.04 release? I'm trying to transcode a TS into FLV but on windows i get a "Invalid data found when processing input" whereas under Ubuntu she happily transcodes away
[12:50] <JEEB> ubuntu uses libav, not ffmpeg first of all
[12:50] <plm> morning all
[12:50] <JEEB> and since ubuntu only uses release code it's probably somewhat old libav, too :)
[12:51] <JEEB> so yes, pretty different from a build you might be able to catch these days from zeranoe
[12:52] <Macey> ffmpeg version N-57235-g0bf8580 - Windows
[12:52] <JEEB> also if you are going to use libav-based tools, I recommend switching to the avconv tool instead of the ffmpeg tool. Because basically elenril rewrote parts of the ffmpeg tool, and the rewrited/updated one was then called avconv in libav. ffmpeg's ffmpeg does have these changes as well, since ffmpeg merges most if not all that comes from libav :)
[12:52] <Macey> ok cool, i tried avconv aswell
[12:52] <Macey> same issue
[12:53] <JEEB> so the same issue comes up with current ffmpeg and the avconv on ubuntu?
[12:53] <Macey> yeah, ubuntu both ffmpeg and avconv both happily transcode
[12:53] <Macey> though windows versions don't
[12:53] <JEEB> uhh
[12:53] <JEEB> you just said... oh wait, you meant that you grabbed a current libav avconv build and that failed too, while the older versions in ubuntu work fine?
[12:54] <JEEB> anyways, sounds like a bug
[12:54] <Macey> yeah so ubuntu are just apt-get installs
[12:54] <Macey> windows is latest d/ls
[12:54] <JEEB> if it fails with zeranoe's current ffmpeg builds, you can report it in ffmpeg's trac, and libav problems can be reported at their respective bugzilla tracker
[12:54] <JEEB> :)
[12:55] <Macey> i think that the windows versions aren't built with TS support
[12:55] <JEEB> there's nothing special you need for that
[12:55] <JEEB> mpeg-ts is libavformat, part of ffmpeg/libav
[12:56] <JEEB> you would have to specifically disable things for it to be disabled, so no -- that's not it
[12:56] <Macey> hum
[12:56] <JEEB> it's most probably a bug in the current libavformat code
[12:56] <Macey> "Invalid data found when processing input"
[12:56] <JEEB> also you could try building a simple build of ffmpeg or libav on the ubuntu i guess
[12:56] <JEEB> and try that too
[12:57] <JEEB> if it's just reading the input that fails then `ffmpeg|avconv -i welp.ts -c:v rawvideo -f null -` should be all needed to replicate
[12:57] <JEEB> and the build would only need ./configure without extra options set
[13:01] <JEEB> and yeah, you'll need the packages git, build-essentials and yasm to build ffmpeg/libav :)
[15:25] <plm> people, is possible to increase the buffer whe streaming via http? I'm streaming a video em box1 to box2. but between box1 and box2 I have a high delay networking and bandwidth unstable (is 3g). So I was thinking to have a bigger buffer in box2 (before show video) for don't have congeals the video. I know that increase the buffer I will have a bigger delay on video, but this is not problem for me. A 30s delay is just fine.
[15:40] <EchoDev> Anyone here who can help me with compiling ffmpeg on windows?
[15:41] <TheCycoTWO> EchoDev: Would these work for you instead http://ffmpeg.zeranoe.com/builds/
[15:41] <EchoDev> nope :( I need a modified one
[15:42] <EchoDev> WARNING: Unknown C compiler gcc, unable to select optimal CFLAGS <- Im getting this error
[15:42] <EchoDev> I think I'm making a mistake with mingw somewhere
[15:46] <EchoDev> Hmm I have found a complete package for compiling ffmpeg
[15:46] <EchoDev> It seems to be working fine >:)
[15:46] <TheCycoTWO> spiffy
[15:46] <EchoDev> except for the libopencore_amrnb which I can fix
[15:48] <TheCycoTWO> anyone here up on the latest recommendations for libswresample vs libavresample? I read when it was brought in that libavresample would be unmaintained, but is that still the case?
[15:48] <durandal_1707> libav* is maintained by Libav
[15:48] <durandal_1707> *libavresample
[15:49] <durandal_1707> and it had some serious bugs, dunno if it still have as I do not use it
[15:50] <TheCycoTWO> We are currently using libswresample exclusively.  Ubuntu users keep whining in the issue tracker, so I was wondering if the best course was to maintain two code paths or switch.
[15:51] <durandal_1707> how much of your userbase are Ubuntu users?
[15:51] <JEEB> TheCycoTWO, if you want to support both I would recommend you use libavresample, since that is also available in ffmpeg as well
[15:51] <JEEB> although of course libavresample is only maintained by libav, and just merged into ffmpeg
[15:52] <JEEB> that said, there are plenty of projects using it already
[15:52] <TheCycoTWO> Mostly windows users.  Thanks JEEB.
[15:52] <durandal_1707> but it is full of bugs
[15:52] <TheCycoTWO> ah
[15:52] <JEEB> uhh, if it was that full of bugs
[15:52] <JEEB> I'm pretty sure nev wouldn't use it :P
[15:52] <durandal_1707> that is nev problem
[15:53] <JEEB> yes, but I'm just saying that if it was fully borked it wouldn't get any usage
[15:54] <durandal_1707> even worse stuff have bigger usage
[15:54] <JEEB> yes
[15:54] <TheCycoTWO> nice of you guys to track it anyway, not cause a complete rift
[15:55] <JEEB> durandal_1707, just saying that I'm not aware of any completely broken parts in it
[15:55] <durandal_1707> they are so similar that it should be trivial to support both
[15:55] <JEEB> yes, not really surprising as such
[15:55] <JEEB> and that's another alternative
[15:55] <TheCycoTWO> ?
[15:56] <TheCycoTWO> ah, nm
[16:03] <JEEB> durandal_1707, btw not only have you not actually noted any actual bugs (which no doubt there exist, since there is no perfect software -- but "full of bugs" as far as I know is not what it is), but I must say that even some people who wholly recommend ffmpeg usage want it around http://ffmpeg.org/pipermail/ffmpeg-devel/2013-October/149693.html
[16:04] <JEEB> and if core ffmpeg folk have a problem with it I don't really know... you wanted to make some effort of maintaining interoperability, but you don't want to take it to the end?
[16:05] <JEEB> (full interoperability between libav and ffmpeg is probably impossible, but most other things are manage'able)
[16:05] <durandal_1707> i see it other way around: Ubuntu not providing FFmpeg packages
[16:05] <JEEB> that's a whole separate problem of course
[16:06] <JEEB> in a perfect world we would not have forks of ffmpeg, and all software would be perfect, unfortunately that's not the case.
[16:08] <hans_henrik> discussing libav again?
[16:09] <TheCycoTWO> sorry
[16:10] <durandal_1707> you do not need to apologize
[16:10] <TheCycoTWO> no, I'm Canadian, it's an obligation ;)
[16:11] <JEEB> Just mostly noting that he might not actually be looking at it calmly but more from the "we need to keep people our APIs" stance with some superlative-likes :) Which is understandable of course. That said, I can understand people not wanting to have two separate code paths for two different libraries, as well as since I see ffmpeg-using people using libavresample...
[16:11] <JEEB> just had to comment on it :)
[16:12] <durandal_1707> that is littly funny considering that libswresample is older
[16:12] <JEEB> I have no idea about those things and to be honest I do not have much interest in it either :)
[16:13] <JEEB> I mostly care about the fact that there is a visible line of being able to use the same code path with both projects
[16:16] <JEEB> of course there are also people who have opted to support both since avresample is disabled by default in ffmpeg's configure, like the project whose main developer is asking for avresample to be enabled by default :) https://github.com/mpv-player/mpv/blob/master/audio/filter/af_lavrresample.c
[18:56] <TSM> im confused about the SAR/DAR values and if they can be used to resize the input DV file i have out to a file where it does not need to be streached out as when people play on VLC etc it does not do it automaticly with the DV file
[19:02] <thisislawatts> I am looking into overlaying an image onto footage using a multiply blend mode. Wide eyeed noob here, but I have stumbled upon the filter_complex which I believe should do the trick, but I am not sure I am understanding the syntax completely
[19:11] <Hfuy> Is there anything wrong with this:
[19:11] <Hfuy> ffmpeg -i 00000.MOV -c 2 -c:a pcm_s16le -c:v prores -ac 2 00000_2ch_apcn.MOV
[19:11] <Hfuy> All I really want to do is drop two of the audio streams in the input. But I don't seem to be able to do that without reencoding the video.
[19:19] <Hfuy> Hello? Anyone?
[19:20] <durandal_1707> -c:a copy
[19:20] <durandal_1707> or -c:v copy
[19:21] <Hfuy> Instead of what?
[19:22] <durandal_1707> when you do not want to reencode/transcode use copy codec
[19:22] <Hfuy> I'm also not sure why I have -c 2 and -ac 2
[19:22] <Hfuy> but it doesn't seem to work otherwise.
[19:22] <durandal_1707> -ac 2 means make audio 2 channels
[19:23] <Hfuy> The input file has (oddly) four.
[19:23] <Hfuy> Which is the problem. My NLE doesn't like four channel prores. For some reason. Which is beside the point. :)
[19:23] <durandal_1707> -c 2 is strange way to set codec
[19:23] <Hfuy> er, I did.  ffmpeg -i 00000.MOV -c 2 -c:a pcm_s16le -c:v prores -ac 2 00000_2ch_apcn.MOV
[19:24] <durandal_1707> remove '-c 2'
[19:24] <Hfuy> Uhhuh.
[19:25] <Hfuy> Hm. This is rapidly becoming more of an Adobe Premiere problem than an ffmpeg problem.
[19:26] <Hfuy> It does not like ffmpeg's output in that case, to the extent of actually crashing premiere.
[19:26] <durandal_1707> use '-c:v copy' if 00000.MOV is already prores video
[19:26] <Hfuy> durandal_1707: I'm not completely stupid but it's obvious that I'm not understanding something you're saying. What are you asking me to do?
[19:27] <durandal_1707> to copy your full ffmpeg output of ffmpeg command you use
[19:27] <Hfuy> Well, it is prores, but I've a feeling it's prores 4444, or something.
[19:27] <Hfuy> I'm not sure what the default prores profile used by ffmpeg is.
[19:27] <durandal_1707> if you use copy it should not really matter
[19:27] <durandal_1707> well you said you do not want to reencode vide, no?
[19:28] <Hfuy> Well, I may have to, if Premiere doesn't like what ffmpeg is doing.
[19:28] <Hfuy> Or rather doens't like the way it's reordering things/
[19:28] <Hfuy> http://pastebin.com/mkt8ZV6F
[19:28] <Hfuy> Ah. The input is apch
[19:29] <Hfuy> Which is 422 HQ. As expected. Wtf.
[19:30] <Hfuy> The stream ordering is coming out differently, can I control that?
[19:32] <Hfuy> On the input, 0:1 is the video and it's coming out on 0:0,vice versa for the audio. Any idea why ffmpeg is doing that?
[19:32] <Hfuy> (and ideally it would be nice to copy the metadata streams)
[19:41] <Dresk|Laptop> Hey guys, been using ffmpeg a long time to encode stuff (3+ years now), and I still consider myself a novice -> amateur.  Anyway, right now I'm dealing with the -map option, which is new to me.  I basically have an arbitrary input file (contains video and audio stream), where I extract the audio stream into PCM16bit, peak normalize it, and then in one command I want to re-encode the video stream from the original file into MP4 AND take
[19:41] <Dresk|Laptop>  another input file (the normalized wav file) and use that as the audio
[19:41] <Dresk|Laptop> I'm a bit confused as to the order of precedence necessary on the command line to have ffmpeg ignore the audio stream in the original file and instead use this 2nd input file, which is the audio stream peak normalized
[19:42] <Hfuy> Dresk|Laptop: If you find a decent -map reference please tell me.
[19:42] <Dresk|Laptop> Hfuy: Well one of the things I considered is extracting the video as well, and just working with two raw input streams, which would be easier
[19:42] <Dresk|Laptop> Hfuy: But I don't want to do that, I use 2 pass encoding and want to keep the command simple
[19:43] <Hfuy> Not a clue.
[19:43] <Hfuy> It is possible, as I understand it. But I have no idea how.
[19:43] <Hfuy> Personally, I have a situation where a command is reordering the streams in the file when I don't want it to, which I suspect is causing my NLE to fritz out. But I don't know.
[19:46] <Hfuy> If it's any consolation, I have found it very difficult to find up to date, authoritative documentation for this.
[19:46] <Dresk|Laptop> The order in which you write your ffmpeg call does matter, correct?  Or can you provide enough info on every command line option to make the order independent?  I've always found it confusing to work "first with one stream" and then another, sometimes using the same parameter but for two different streams
[19:47] <TSM> it it possable to detect using ffprobe if a DV video is shot as 16:9 instead of 4:3
[19:47] <Hfuy> There are some situations where option ordering on the command line matters. But I'm not sure when or how. Again, I can't find any authoritative docs.
[19:47] <Hfuy> TSM: Probably, see if it outputs a value for pixel aspect ratio (PAR) or display aspect ratio (DAR).
[19:48] <Hfuy> That information is probably duplicated inside the DV codec's payload in each frame, but I don't think ffmpeg can read that.
[19:49] <TSM> Hfuy: no PAR any longer on new version of ffprobe, [SAR 8:9 DAR 4:3],
[19:49] <Hfuy> I suspect S stands for Sample. Which may be the same thing as Pixel. But I'm not sure.
[19:49] <Hfuy> But DAR seems to be what you need anyway.
[19:50] <TSM> so the fact i can see it says 4:3 means that it should be shown as 16:9?
[19:50] <Dresk|Laptop> Hfuy: Take a look at this apparently very straight-forward (but order dependent) command line for multiple stream mapping : ffmpeg -i a.avi -i b.avi -i c.avi -vcodec copy -acodec copy  /
[19:50] <Dresk|Laptop> -vcodec copy -acodec copy out.avi -map 2.0 -map 0.1 /
[19:50] <Dresk|Laptop> -map 1.1 -map 2.1 -acodec copy -newaudio -acodec copy -newaudio
[19:50] <Hfuy> TSM: Is it a 16:9 video?
[19:51] <Dresk|Laptop> 3 input files, the video codec chosen is just to copy, it's only called twice, not sur ehow the third input file comes into play
[19:51] <Hfuy> Dresk|Laptop: Sorry, that's advanced to me.
[19:51] <TSM> Hfuy: yes it is, well its a dvvideo format
[19:51] <TSM> direct from cam
[19:51] <thisislawatts> Hm ok so I've made some more progress on my multiply overlay dilemma, but now getting what looks like just green channels. So I am using -filter_complex "blend=all_mode='multiply':all_opacity=1" full copy/paste --> http://pastie.org/8419218
[19:51] <Dresk|Laptop> Hfuy: I will solve this for us!  Onward!
[19:51] <Hfuy> I simply have a situation where it's mapping 0:1 to 0:0 and 0:0 to 0:1
[19:52] <Hfuy> It's also dropping 0:2 which is metadata I'd like to maintain.
[19:52] <Hfuy> I can't find any uptodate information on what the behaviour is supposed to be, why it's that way, or how to control it.
[19:56] <Hfuy> TSM: Sometimes I've known DV in AVI files where the DAR in the AVI doesn't get set properly.
[19:56] <Hfuy> That way you can end up with the thing assuming PAR=1:1
[19:57] <Hfuy> or any damn thing inbetween
[19:58] <TSM> grrr, how anoying
[19:58] <Hfuy> so you're saying the DAR given by ffmpeg is 4:3 but it should be 16:9?
[19:59] <TSM> essentialy yes, ive done a straight 'ffmpeg -i <in> -aspect 16:9 <out>' and the result is fine and plays without adjustment
[19:59] <Hfuy> That'll be getting reencoded though I assume?
[19:59] <TSM> but otherwise when i play it in VLC without any changes it is squished until i change the aspect ratio
[19:59] <Hfuy> VLC didn't used to be known for obeying things anyway.
[19:59] <TSM> let me do it with -vcodec copy
[20:00] <Hfuy> I'm not sure if you can change DAR with -c:v copy
[20:00] <Hfuy> But again, it's undocumented.
[20:00] <Hfuy> Can anyone explain to me how I should tell ffmpeg to not reorder the input streams in the output?
[20:01] <Hfuy> Or to put it another way, put input 0:0 into output 0:0, input 0:1 into output 0:1, and ideally maintain input 0:2 into output 0:2 unchanged (0:2 is a timecode track)
[20:02] <TSM> does not allow me to do vcodec copy, hmmm
[20:02] <TSM> ile be reencoding the videos anyway so not much of a problem
[20:02] <Hfuy> what is it, quicktime, avi?
[20:04] <TSM> dvvideo
[20:08] <Hfuy> oh, shit. right.
[20:08] <Hfuy> raw DV eh.
[20:08] <Hfuy> Hmm.
[20:08] <Hfuy> And this is captured from a camera, not encoded with software?
[20:09] <Hfuy> If it's from a camera the information will almost certainly be in there.
[20:10] <Hfuy> But I wouldn't be surprised if it didn't appear correctly in ffmpeg, which is probably more commonly used with containers such as quicktime.
[20:10] <Hfuy> there are DV tools around which allow you to read DIF information from raw dv video files, which should reveal what's going on.
[20:11] <Hfuy> http://www.avpreserve.com/avpsresources/tools/ look for DIF analyzer
[20:13] <Hfuy> Anyone know what this means: "Data stream encoding not supported yet (only streamcopy)"
[20:13] <Hfuy> It appears when I insert -map 0:2 into a command dealing with an input file that has quicktime timecode as stream 0:2.
[20:13] <Hfuy> I'd be happy with a copy of the stream but I can't find out how to do that.
[20:14] <TSM> oh the complexity my head is burning
[20:15] <Hfuy> Complexity of what?
[20:15] <TSM> working out video formats etc
[20:15] <Hfuy> Heh. It can be a bit complicated.
[20:15] <TSM> im just compiling up that DV analizer
[20:16] <Hfuy> It doesn't help that the documentation for ffmpeg is... well, practically nonexistent.
[20:16] <Dresk|Laptop> Hfuy: Well, it's definitely not a win, but I solved the issue by extracting the video stream and just using 2 input files, since they were raw streams, so no mapping was needed
[20:16] <TSM> but on the flip side, who the hell still shoots real 4:3 DV video anyway, not in my industry they should not
[20:16] <Hfuy> Dresk|Laptop: the way it seems to work is that you put several -map x:y args in the command.
[20:16] <Hfuy> So, if I state: -map 0:0 -map 0:1 I get the same order as I had in the input.
[20:17] <Hfuy> I'm not sure if you can simply state more than one -i and then use -map 0:1 -map 1:1 in the command to get the outputs you want.
[20:17] <Hfuy> I'm guessing here. I don't know. I haven't tried it.
[20:17] <Hfuy> TSM: I shoot quite a bit for news, believe it or not.
[20:18] <Hfuy> Big companies may have 4K tvs to sell, but most of my work is still DVCAM.
[20:18] <TSM> true i mean how much is in real 4:3, ie not anemorphic
[20:19] <Hfuy> Not much here in London.
[20:19] <Hfuy> In the US, where widescreen standard def was never so much of a thing, quite a bit I suspect.#
[20:19] <TSM> not since we are now all 16:9 on freeview etc
[20:19] <Dresk|Laptop> Does anyone know what parameters affect pass 1 on video streams?  For example, I am making 4 video files from 1 source, and they all vary in resolution and bitrate - do I need to do a pass 1 for EVERY one because my chosen resolution and bitrate, or is the pass a general thing that can be used across all formats?
[20:19] <TSM> yup ive seen quite a bit on US cheepo channels
[20:20] <Hfuy> TSM I think they just expect you to go to HD if you want wide
[20:20] <Hfuy> Does anyone know how you set prores profiles?
[20:20] <x__> Hey does anyone know how I can compress my .mov files from 1.9 to about 700-800 mb?
[20:20] <TSM> I have to consider all countries, work in UK but lots of the video content comes from US videographers
[20:20] <Hfuy> TSM: are these professional news guys or what
[20:21] <x__> I have one command that I used to get 1.9 to 1.2 but not show how to condense it a bit more: avconv -i 12\ Dates\ of\ Christmas\ \[2011\].mov -acodec libfaac -b:a 128k -vcodec mpeg4 -b:v 1200k -flags +aic+mv4 final.mp4
[20:21] <TSM> no way, just papz that get hold of a cam and hound celebs :p
[20:21] <Hfuy> Oh puh-leease.
[20:21] <Hfuy> x__: you might try reducing the 1200k.
[20:22] <x__> Hfuy: To what? not sure what that actually means?
[20:22] <x__> Hfuy: Should I put it to 800k ?
[20:23] <Hfuy> I'm not sure, either, but I suspect "-b:v 1200k" may mean "bitrate for video should be 1200 kilobits per second."
[20:23] <Hfuy> Try it. See what you get.
[20:23] <TSM> b:v = video, b:a=audio
[20:23] <Hfuy> Controlling the bitrate of h.264 is a bit approximate anyway.
[20:24] <TSM> anyone tried the opencl stuff of ffmpeg 2 yet?
[20:25] <x__> Hfuy: I am checking it out now
[20:26] <Hfuy> How do I get it to drop a stream entirely?
[20:28] <x__> Hfuy: I checked in libav's documentation and it said to set the video bitrate of a file to 64kbits/s you use: avconv -i input.avi -b 64k output.avi
[20:29] <Hfuy> well there yo go
[20:31] <x__> Hfuy: I will let you know how it goes. Thanks for the help.
[20:31] <Hfuy> Good grief, I'm actually dispensing advice on #ffmpeg
[20:31] Action: Hfuy ducks the flying pigs
[20:32] <Hfuy> Does anyone know how to have ffmpeg drop an output stream entirely?
[20:39] <Hfuy> Hn. Alternatively, is there some way I can get it to translate the video content from yuv422p10le to yuv422p10 so I can use an alternate codec?
[22:23] <thisislawatts> Hey there, I am struggling with the -filter_complex "blend=all_mode='multiply'" outputting only green videos, would anyone like to point out the obvious error I am making?
[22:27] <durandal11707> using yuv pixel format
[22:30] <thisislawatts> durandal11707: what format would you recommend?
[22:30] <durandal11707> gbrp
[22:31] <durandal11707> format=gbrp for each input,blend...
[22:33] <durandal11707> just: format=gbrp,blend=all_mode=multiply
[22:38] <thisislawatts> Hm, now it's outputting just black. Are there any other formats you'd recommend?
[22:38] <durandal11707> its black when one of images are mainly black
[22:39] <durandal11707> it does A * B / 255 for each pixel in each image
[22:40] <durandal11707> you can create own ones with expressions... there are some examples in documentation
[22:40] <thisislawatts> ah, so that's how the expressions work, that's great I was a little perplexed by the documentation.
[22:41] <durandal11707> you can use modes, but they are just limited - no fancy stuff like animations and dynamic opacity
[22:42] <durandal11707> it all depends on what you really want to archieve
[22:43] <thisislawatts> in truth what I am trying to achieve is a very basic, just a colour burn -> http://www.thefirefliestour.com/wp-content/themes/fireflies/img/riders-archive.jpg
[22:43] <thisislawatts> Well at least I am assuming it's very basic :)
[22:44] <durandal11707> well there is burn mode
[22:47] <thisislawatts> Yeah I have tried with that also and it's still outputting solid black
[22:49] <durandal11707> what images you use?
[22:52] <thisislawatts> These are them -> http://thisis.la/blendmode.png
[22:54] <durandal11707> you mean you use same images for both inputs ?
[22:55] <thisislawatts> At the moment yeah, just using that red box to test things out?
[22:57] <durandal11707> i don't get it what is exact command you use?
[22:57] <thisislawatts> ffmpeg -i sample.mov -i bw.mov -filter_complex "format=gbrp,blend=all_mode=burn" test$(date +"%s")\.mov
[23:01] <durandal11707> and where is sample.mov and bw.mov?
[23:01] <thisislawatts> in the same directory as where I am executing the command
[23:03] <durandal11707> swap inputs
[23:08] <burek> can "subtitles" filter ( http://www.ffmpeg.org/ffmpeg-all.html#subtitles ) take an input from the input stream (subtitle stream) or it strictly takes its input from files only?
[23:08] <durandal11707> iirc from files only
[23:09] <burek> ok, thanks :)
[23:09] <brontosaurusrex> how would i use the segment thingy when encoding to prores (I'd just like multiple files as output say 1 minute long) ?
[23:09] <brontosaurusrex> wanted container is mov, codec prores
[23:10] <thisislawatts> I have tried swapping the inputs to no avail, I think I will have to take another look in the morning, I would prefer to generate this coloured overlay from within ffmpeg
[23:11] <ubitux> burek: you can pass it a mkv
[23:11] <ubitux> burek: it will pick the first subtitles stream iirc
[23:11] <ubitux> but if it's ass you won't have fonts etc
[23:12] <durandal11707> thisislawatts: pehaps you need linear burn and not color burn
[23:12] <durandal11707> the burn mode is actually color burn
[23:12] <burek> ubitux, the input is already mkv: http://ffmpeg.gusari.org/viewtopic.php?f=11&t=1119
[23:12] <burek> the guy just wants
[23:12] <burek> to do it all in one step
[23:13] <ubitux> redirect him to the ticket requesting that feature
[23:13] <burek> ok
[23:13] <ubitux> i can find it for you
[23:13] <burek> thank you :)
[23:14] <ubitux> http://trac.ffmpeg.org/ticket/1305
[23:14] <burek> +1
[23:24] <brontosaurusrex> this is what i have so far, nothing; http://paste.debian.net/plain/60251
[23:27] <durandal11707> thisislawatts: linear burn is: blend=all_expr=A+B-255
[23:48] <thisislawatts> durandal11707: Thank you kindly, I will let you know how I get on :D
[00:00] --- Tue Oct 22 2013


More information about the Ffmpeg-devel-irc mailing list