[Ffmpeg-devel-irc] ffmpeg.log.20120608

[burek]

[00:17] <juanmabc> ffmpeg formed on vlc plugins, or something
[00:20] <JEEB> teratorn, VLC is a player project that bases itself on the libraries that ffmpeg provides
[00:20] <JEEB> (as well as others, but ffmpeg's libraries provide parsing of many formats as well as decoding functions -- as well as encoding in some cases)
[00:20] <teratorn> JEEB: yeah, but I heard about vlc releasing an a/v library
[00:21] <teratorn> was wondering what that was all about, and if it relies on or wraps ffmpeg
[00:21] <JEEB> I don't exactly know, but I would guess it would be something that enables one to use ffmpeg and friends in a VLC'ish kind of way without the necessity of going and looking at all of the gritty details
[00:22] <JEEB> just like ffmpegsource (ffms[2]) is a more high-level API to get frame-exact sourcing of video and audio streams using ffmpeg
[00:32] <heimlich> hi guys, got a small problem.. i am trying to  make a quicktime movie from an image sequence... and it works perfectly when my image sequence is .0001, .0002, etc via .%04d however, if my image sequence doesnt start at 1, but starts at like 316-400, then it tells me it can't find the file/directory... is there a way to tell it that the first frame is 316 ?
[00:34] <heimlich> i know i can rename the files.. but i was wondering if there is a way to do that without that
[00:40] <saste> heimlich: http://ffmpeg.org/faq.html#toc-How-do-I-encode-single-pictures-into-movies_003f
[00:40] <saste> short answer, no way, although there is an open ticket for that feature iirc
[00:42] <heimlich> yeah.. i saw on the howto the rename thing.. which is what i ended up doing.. but.. good that there's a ticket for it already.. thanks
[00:48] <saste> heimlich: http://ffmpeg.org/trac/ffmpeg/ticket/247
[00:48] <saste> upvote
[01:36] <juanmabc> !random
[01:51] <LordCrc> hi, im trying to encode with x264, and in my output info i see "10000k tbn, 30 tbc"... is that value for tbn an issue?
[02:11] <llogan> why can't something like ticket 247 be a SoC project? it would be less complicated and "grandiose" but probably more useful
[02:18] <Anusien> I have some x264 .mp4 video files I am trying to get audio-only versions via ffmpeg. I'm usingthis command: ffmpeg -i $1 -ab 320k -ac 2 -ar 44100  $1.mp3. The resulting files won't seek properly on an iphone (but mp3s converted from non x264 .mp4s will). Is there something I can do to fix this?
[02:22] <Anusien> The problem only occurred when the video source switched from XVID to x264, so I'm wondering if I need any extra configuration
[02:48] <burzum>         (int) 10 => '[mp3 @ 0x38d260]Could not find codec parameters (Audio: mp3, 0 channels, s16)',
[02:48] <burzum>         (int) 11 => 'C:\\xampp\\htdocs\\music\\tmp\\processing\\4fd14a8f-015c-4b8e-b0bb-3cfce1a32268-sample: could not f
[02:48] <burzum> ind codec parameters',
[02:48] <burzum> how can i fix that?
[02:48] <burzum> the file has no file extension, im aware of that
[02:48] <burzum> im already using -f mp3
[02:48] <burzum> but why am i still getting this error?
[02:50] <burzum> ah fuck its still using the wrong ffmpeg bin
[03:27] <LordCrc> so apparently avi files can have variable framerate content now... good times
[04:59] <LordCrc> what does "Que input is backward in time" mean?
[06:14] <CoJaBo> There was a really nice sequence of options, I *think* using ffmpeg, to convert a video to a series of PNGs or JPEGs or something (decimating the framerate in the process); but I can't seem to find it- any ideas?
[06:28] <CoJaBo> All I'm getting is the reverse process -_-'
[09:53] <Mista-D> any way to have a first pass generate "universal" passlog file that works on many resolutions?  "[libx264 @ 0x1d19da80] MB-tree doesn't support different resolution than 1st pass (640x352 vs 320x176)"
[10:03] <Valo> whats the diff between f_size and s_size
[10:04] <Valo> in vstats
[10:20] <burzum> can anyone please explain me why the result is different for the flac and wav file? i need to parse the meta data and it breaks because it differs for some reason: http://bin.cakephp.org/view/437998687
[10:23] <funyun> hi. can someone tell me what ffmpeg command i would use to turn this file http://pastebin.com/PbV4ErmU into this file http://pastebin.com/hayMrWkR ?
[10:29] <funyun> anyone?
[10:43] <funyun> anyone at all?
[11:28] <funyun> hi. can someone tell me what ffmpeg command i would use to turn this file http://pastebin.com/PbV4ErmU into this file http://pastebin.com/hayMrWkR ?
[11:31] <burek> instead of using mediainfo, use ffmpeg -i file
[12:02] <rainmaker1> Hi, what is default profile the ffmpeg uses? baseline or main?
[12:08] <funyun> burek: original = http://pastebin.com/6dN0ybPK encode = http://pastebin.com/RyNPga8t
[12:09] <burek> oh, your ffmpeg is too old
[12:09] <burek> can you update it?
[12:10] <funyun> burek: ya. but on a mac it's a long process so i planned on doing it tomorrow. i just wanted to know what syntax would work so i could try it tomorrow
[12:11] <burek> ok, do you want original to look like encoding or vice versa
[12:13] <funyun> burek: yes. then encode loads fine into final cut pro but the original doesn't so i need the original to have the same settings as the encode
[12:13] <burek> yes what?
[12:13] <funyun> i need to original to look like the encode
[12:15] <funyun> the*
[12:17] <burek> ffmpeg -i original.mp4 -vcodec libx264 -acodec aac -strict experimental -ab 128k -ar 48k -ac 2 output.mp4
[12:19] <funyun> burek: thanks. so will that automatically make the codec profile Main at L3.2?
[12:19] <burek> well, you can control that using -profile
[12:20] <burek> also, -preset -tune -crf and -r
[12:20] <burek> type x264 --help for more info
[12:20] <burek> and for -r it's a bitrate, so you can limit the bitrate like -r 1300k
[12:20] <funyun> so add -profile 3.2?
[12:21] <burek> x264 --help
[12:21] <funyun> alright thanks again
[12:23] <burek> :beer: :)
[12:24] <ahuillet> someone should create "beerpal"
[13:03] <smjms> how do I know when ffmpeg is using VA-API acceleration?
[14:45] <leandrosansilva> Hello to all. Can I create a new input protocol for FFMpeg outside of it's source tree?
[14:46] <leandrosansilva> I need to use ffmpeg to read some files from memory (I really must use memory) and I saw I need to implement a new protocol
[14:48] <ahuillet> why don't you call libavcodec with some code of yours?
[14:55] <leandrosansilva> ahuillet, can you exaplain more your idea?
[14:55] <ahuillet> leandrosansilva : I'm not sure I understand what to do, but if you want to pass some data "from memory" to "FFmpeg", you don't need to use the commandline tool called ffmpeg
[14:56] <leandrosansilva> ahuillet, sorry. I didn't explain my problem very well :-)
[14:56] <leandrosansilva> I'm not using ffmpeg executable
[14:56] <leandrosansilva> I'm using libav API
[14:57] <leandrosansilva> and I want to reuse the libavformat code to load files located in memory
[14:58] <leandrosansilva> reading libavformat source code I can see I need to implement a new URLProtocol
[14:58] <ahuillet> what's "file located in memory" to begin with?
[14:58] <ahuillet> *what's a
[14:58] <leandrosansilva> and I need to know if I can do it outside of ffmpeg source tree
[14:59] <leandrosansilva> in fact it's not a file in memory
[14:59] <ahuillet> I had guessed as much :)
[14:59] <leandrosansilva> is a buffer in memory which contais the exactly bytes from a file
[14:59] <ahuillet> please explain to me how that is different to writing a program that reads a file and feeds the data to avcodec
[15:00] <ahuillet> because you seem to describe precisely what ffmpeg (and any other program reading a file) does
[15:00] <leandrosansilva> the problem is I don't have a filesystem to access
[15:01] <leandrosansilva> my data is in memory
[15:01] <ahuillet> so what?
[15:01] <ahuillet> I'm reading a file, the data usually goes to memory, then I have a pointer to "a buffer in memory which contains [...]" have I not?
[15:01] <ahuillet> I'm trying to understand here if I'm stupid, if you're unclear, or if you're simply confused about what you're trying to achieve.
[15:02] <leandrosansilva> my mistake again
[15:02] <leandrosansilva> let me explain again :-)
[15:02] <leandrosansilva> initially I have a sequence of bytes in memory (acessible from a uint8_t*)
[15:03] <leandrosansilva> these sequence of bytes are the same content that a mjpeg file
[15:03] <ahuillet> ok, I can get that with mmap() on my file
[15:04] <leandrosansilva> and I can't write this buffer in filesystem to open it after with file:// protocol
[15:05] <ahuillet> ok, so you have a series of bytes in memory
[15:05] <ahuillet> which is the place where you typically store data to be treated on a modern computer ;)
[15:05] <ahuillet> (and even on a non-modern computer FWIW)
[15:06] <leandrosansilva> (as you can see English isn't my native language, so forgive me if I'm not being clear)
[15:06] <leandrosansilva> in fact the videos are stored in a kind of database
[15:06] <ahuillet> I think I understand what you're trying to tell me very well
[15:07] <ahuillet> you have a pointer to a buffer containing your data
[15:07] <ahuillet> that's essentially what you're telling me, correct?
[15:07] <leandrosansilva> yes
[15:07] <leandrosansilva> they're in a "database" and to access them I load them into the memory
[15:07] <ahuillet> ok, so you can create an AVPacket
[15:07] <ahuillet> and set its "data" member to your pointer
[15:07] <ahuillet> and pass that to avcodec
[15:08] <leandrosansilva> but does AVPacket work with many frames?
[15:09] <ahuillet> I think it's one at a time, but I decode one frame at a time anyway so I can't really tell you
[15:09] <leandrosansilva> I usually use AVPacket with every frame
[15:09] <ahuillet> can't you do that?
[15:09] <leandrosansilva> I think I can
[15:10] <ahuillet> sounds like it to me, especially in motionjpeg :)
[15:10] <ahuillet> this is an example of what I use: http://pastebin.com/UawNVanZ
[15:11] <tonsofpcs> if anyone ever wants to work on ancillary data in MXF (SMPTE 436M), let me know and I can likely get you sample files.
[15:13] <leandrosansilva> ahuillet, and where did you get the AVCodecContext used to encode the video? How did you fill it?
[15:14] <ahuillet> 	ctx->codec = avcodec_find_decoder(CODEC_ID_H264
[15:14] <ahuillet> 	ctx->codec_ctx = avcodec_alloc_context3(ctx->codec);
[15:14] <ahuillet> + some stuff
[15:15] <ahuillet> pretty much your basic libavcodec example
[15:15] <leandrosansilva> nice! thx a lot!
[15:15] <leandrosansilva> I'll try it
[15:16] <ahuillet> grab an example somewhere, what you're trying to do really doesn't sound that hard.
[15:17] <leandrosansilva> yes, I've already used these functions
[15:18] <ahuillet> let me show you what I have
[15:43] <hackish> I have a lot of research video footage that needs to be translated from a poorly chosen divx format to h264. There are about 1000h of video, mostly static with 2 insects moving around. Unfortunately it is taking eons to transcode. The video is 364x364 @ 30fps and the files are generally 7-8h at about 3-4gb each. Trouble is it takes about 1-2h of recode each.
[15:43] <hackish> Given that most of the frame is static and the only importantance in the encoded result is sharp edge detail is there a good setting I can change to increase the recoding speed?
[15:44] <ahuillet> 250h of encoding?
[15:44] <ahuillet> have you thought of buying some CPU resources at a "cloud" provider?
[15:44] <ahuillet> it doesn't sound that it would be very expensive
[15:44] <ahuillet> s/that/like/
[15:45] <hackish> well first step would be to find good settings. here is what I came up with for the x264 presets: coder=0 bf=0 flags2=-wpred-dct8x8 wpredp=0 me_method=hex me_range=8 subq=5
[15:45] <ahuillet> I can't help you here sorry - #x264 and x264 --fullhelp may have more info
[15:45] <hackish> The files are auto-processed with an opencv program I wrote and split into 2 files. 1 that contains frames where the insects are in close proximity and the other when they are not.
[15:47] <hackish> I will try there. Another question though... Do you know if running on a machine with CUDA support would have a big impact on transcoding performance?
[15:48] <hackish> I bought a GTX 570 because nvida promised great things with their PureVideo drivers but so far it's shown to be pretty useless.
[15:48] <ahuillet> under Linux it sure won't help
[15:48] <ahuillet> because they don't implement HW accelerated encoding there AFAIK
[15:49] <hackish> Right. This is a windoze machine.
[15:49] <ahuillet> as for Cuda encoding, the problem is it has horrible quality
[15:49] <ahuillet> and it's not even that fast (about 2x as compared to software encoding according to what I read)
[15:49] <ahuillet> http://www.hardware.fr/marc/h264nvenc/imgcompivy.htm?inception720a if you want to see for yourself
[15:49] <hackish> The opencv program runs on it. I have not been successful at all on coaxing opencv to encode in h264 so I'm stuck dumping the output in DIVX then trying to transcode.
[15:50] <hackish> wow that is pretty bad
[15:51] <hackish> in my case the insects move pretty slow so I'd hope one could get reasonable results without needing too much of the expensive motion search stuff.
[16:15] <nimesh_accenture> can anyone please tell me how can i play raw RGB565
[16:15] <nimesh_accenture> i can only get hold of players that play rgb444
[16:16] <ahuillet> do you have an example clip?
[16:18] <nimesh_accenture> ahuillet, yes
[16:19] <ahuillet> one that you could send me? :)
[16:19] <ahuillet> I don't think I ever got mplayer to play raw RGB correctly..
[16:20] <nimesh_accenture> ahuillet, dcc didnt work
[16:20] <ahuillet> it won't
[16:20] <nimesh_accenture> ahuillet, so how do i send it?
[16:20] <ahuillet> well, I don't think I can help you, sorry
[16:21] <ahuillet> I thought that my Xv output (nvidia geforce 9) could do raw RGB but it can't
[16:21] <ahuillet> anyway your solution is to hack up a player
[16:21] <nimesh_accenture> ahuillet, np thx for trying
[16:21] <ahuillet> not *that* hard
[16:22] <ahuillet> grab whatever "SDL video player example" (one that doesn't use YUV overlays) on the net and hack it up, or the same without SDL if you feel like it
[16:23] <nimesh_accenture> this has already frustrated me enough...and i'm in no mood to do coding today
[16:24] <ahuillet> if you get ImageMagick to treat your raw images, you could convert each of the images to JPEG
[16:24] <ahuillet> but last time I tried I wasted two hours figuring out the parameters and ended up using gimp. :)
[20:29] <aro> is there a way to get ffmpeg to work with 5.4?
[20:30] <smjms> 5.4?
[20:31] <aro> php
[20:34] <aro> anyone know?
[20:34] <aro> i get an error when i try to run it
[20:35] <iive> aro: in case you talk about ffmpeg-php, it is completely separate project that uses ffmpeg.
[20:35] <aro> ok
[20:36] <smjms> have you asked in #php?
[20:39] <markmx> hi, how I can edit mediainfo in videofile?
[20:39] <llogan> markmx: you mean metadata?
[20:40] <markmx> yep
[20:40] <markmx> title, genre etc
[20:41] <llogan> ffmpeg -i input -c:copy -metadata title="Your Title" -metadata author="Author" output
[20:41] <ubitux> -c copy
[20:41] <llogan> oh, yeah. i should talk on phone and irc at same time
[20:41] <llogan> n't
[20:41] <llogan> see
[20:42] <markmx> hm... it's creates new file with new metadata? recodded?
[20:42] <markmx> or just copied?
[20:42] <smjms> I think just copied
[20:42] <llogan> does not re-encode but creates a new file.
[20:49] <markmx> it's starts to recoding =(
[20:49] <markmx> or...
[20:49] <markmx> just copycoding
[20:52] <markmx> copycoding - copy file frame-by-frame :) seeml like this... got file with new metadata and same size
[20:52] <markmx> thanks
[21:17] <APoulos> Hey burek are you online?
[21:18] <APoulos> burek, I need some help with a conversion.
[21:21] <APoulos> burek - http://pastebin.com/wxu8TE36 - what would I need to modify in this command to make a straight copy conversion but modify the file size to get it down a little bit?
[21:21] <APoulos> burek, just pm me when you get the chance.
[21:28] <llogan> APoulos: see the -fs or -t options
[21:28] <APoulos> llogan, I need an example using my command
[21:28] <APoulos> llogan, if you could, kindly
[21:28] <llogan> of course they will truncate your video upon reaching the declared file size or time.
[21:28] <Mavrik_> so you want to copy the file... and make it smaller? :)
[21:29] <llogan> ffmpeg -i input.mkv -t 10 -acodec copy -vcodec copy -scodec copy out.mp4
[21:30] <llogan> just like most other output options
[21:40] <APoulos> Mavrik_, like mkv to mp4 with the same codecs, just lower the file size
[21:53] <max_L> HI everyone
[21:53] <max_L> I would like to compare a video received after streaming it through a lossy network, to the original video sent
[21:54] <max_L> because of the packets lost, I don't have the same number of frame in the two videos
[21:54] <max_L> is it possible to add dummy frames in the received video to be able to compare it frame by frame to the original video ?
[21:57] <llogan> APoulos: if you want to lower the file size then you will either have to re-encode, copy a shorter duration, or exclude streams form the copy.
[21:57] <llogan> *from
[22:10] <max_L_> hi again, sorry for accenditaly leaving, I'm still there :)
[22:37] <APoulos> llogan, okay so If I choose to rencode can I keep the video quality and lower the audio quality?
[22:38] <APoulos> llogan, if I remember correctly theres a syntax "-q(number here)" that I can use instead?
[22:46] <iive> APoulos: if you want to keep the video quality you should better just "copy" the video stream.
[22:47] <APoulos> iive, right but Im trying to lower the file size, how could I lower the audio quality down?
[22:48] <iive> by making it sound worse? give it smaller bitrate? imho, most audio codec already use less than 128kbps, so it is not good idea to cut that. and video is around 1mbps.
[22:53] <phraktal77> http://pastebin.com/0LkfVj56
[22:53] <phraktal77> ffmpeg -i song.mp3 -f mp3 http://localhost:8090/feed1.ffm
[22:53] <phraktal77> trying to stream mp3 ..
[22:54] <phraktal77> quits right away
[22:54] <phraktal77> what am I doing wrong there ?
[22:56] <relaxed> phraktal77: try, ffmpeg -re -i song.mp3 -f mp3 http://localhost:8090/feed1.ffm
[22:57] <phraktal77> hmm .. I think it's ffserver's issue.. if I run the same command and output to a test.mp3, it works fine
[23:12] <max_L_> any idea of how I could find out about the missing frame, the best for me would be to insert a black frame (something that can be spotted easily)
[23:15] <APoulos> okay, so I still need help getting this file down about a GB
[23:15] <relaxed> APoulos: why?
[23:16] <APoulos> relaxed, because 4.7GB is too large for me to stream
[23:16] <llogan> you could always re-encode it and target a specific output file size
[23:16] <relaxed> It would be easier to cut it into segments
[23:16] <llogan> ^ or that
[23:17] <APoulos> llogan, thats what I was asking you about, how can I do that?
[23:17] <APoulos> llogan, how would I modify my original command to fit that?
[23:17] <relaxed> mkvextract can output chucks into certain sizes.
[23:17] <llogan> bitrate = file size/duration
[23:17] <relaxed> Then streamcopy to mp4 if that's what you need to stream.
[23:18] <llogan> example: You want an output of ~200 MB for a 50 minute duration video
[23:18] <llogan> no...let me change that just so i can copy and paste...
[23:18] <APoulos> relaxed, im on ubuntu
[23:19] <APoulos> llogan, here's the file im dealing with: http://pastebin.com/6ca6WXFR
[23:19] <llogan> how big do you want it?
[23:20] <APoulos> llogan, well, I want to keep some of the quality, but I want it down about to about a 2.7GB file
[23:20] <APoulos> llogan, here's my original command http://pastebin.com/5PpJEMbc
[23:20] <APoulos> how would I alter that?
[23:20] <APoulos> to fit the quality/filesize
[23:20] <llogan> you didn't answer my quesiton
[23:22] <APoulos> llogan, I did actually, if you scroll up  I said "I want it down to about 2.7GB"
[23:24] <llogan> x264 core 57 svn-702
[23:24] <llogan> so old.
[23:25] <LordCrc> hi, what does "Que input is backward in time" mean and does it matter that the aac encoded spits it out constantly while encoding?
[23:26] <APoulos> llogan, i didnt encode it buddy. so if you'd be so kind to just to focus on how to help me?
[23:26] <llogan> i didn't say you encoded it, buddy
[23:27] <llogan> maybe someone else can help you if you're going to be that impatient
[23:27] <llogan> later
[23:27] <APoulos> llogan, impatient? I just didnt appreciate the way you criticized the file rather than helped me out.
[23:29] <APoulos> llogan, you know what, whatever man, I'll ask someone else who isnt going to be picky over something so small as an outdated codec.
[23:59] <JacobF> ran into trouble trying to build latest git
[23:59] <JacobF> http://pastie.org/4052947
[23:59] <JacobF> I'm on linux
[00:00] --- Sat Jun  9 2012