[Ffmpeg-devel-irc] ffmpeg-devel.log.20140528

burek burek021 at gmail.com
Thu May 29 02:05:02 CEST 2014

[00:10] <UtUser> How am I supposed to upload this 80 MB file?
[00:11] <UtUser> If I host it on a web server, can one of you download it, test it, and re-open the ticket?
[00:12] <J_Darnley> There are lorts of websites that offer this service.  FFmpeg also has an ftp server that you could use.
[00:13] <UtUser> I can host it temporarily
[00:13] <UtUser> but I want you to handle the trac ticket, okay?
[00:14] <UtUser> if the FTP serer is public, I suppose it would be easiest for me to upload it there
[00:14] <J_Darnley> http://ffmpeg.org/bugreports.html
[00:15] <UtUser> you handle the bug report
[00:15] <UtUser> I'm not going to explain all the little details
[00:16] <UtUser> it would be better if you could confirm this problem yourself before we go and reopen the ticket
[00:16] <J_Darnley> Me?  Personally?  I don't even have an account on trac
[00:16] <UtUser> are you familiar with ffmpeg's source code?
[00:17] <J_Darnley> A little but not with utvideo
[00:17] <UtUser> I was talking to the utvideo guy on #ffmpeg just a few days ago
[00:17] <UtUser> that was before I knew about the problem
[00:17] <llogan> you were talking to him today
[00:17] <UtUser> really?
[00:18] <llogan> refer to 12:14:52 -8 UTC
[00:18] <UtUser> hey Daemon404
[00:19] <UtUser> how am I supposed to do that?
[00:21] <UtUser> help me out here
[00:22] <UtUser> !log
[00:22] <UtUser> !viewlog
[00:22] <J_Darnley> What are you looking for?
[00:23] <UtUser> some asshole told me to check the log
[00:23] <UtUser> I am not familiar with the ffmpeg developers
[00:23] <J_Darnley> So scroll up
[00:23] <UtUser> my log doesn't go that far back
[00:24] <UtUser> only to 13:10
[00:24] <wm4> people don't like pointing out the obvious, so sometimes they say the same thing, but slightly less obvious
[00:24] <UtUser> I do not understand
[00:24] <UtUser> I am in GMT-8
[00:24] <UtUser> UTC=GMT AFAIK
[00:25] <J_Darnley> http://pastebin.com/Q49urRqe
[00:26] <cone-554> ffmpeg.git 03Diego Biurrun 07master:512f3ffe9b4b: dsputil: Split off HuffYUV encoding bits into their own context
[00:26] <cone-554> ffmpeg.git 03Michael Niedermayer 07master:48a691630833: Merge commit '512f3ffe9b4bb86767c2b1176554407c75fe1a5c'
[00:27] <UtUser> yeah, so @Daemon404
[00:27] <UtUser> I assume he's not sleeping
[00:27] <J_Darnley> I can't say.
[00:28] <UtUser> he's a great guy
[00:28] <UtUser> he ported the ut video code all himself
[00:28] <J_Darnley> But as you've you've pinged him a few times he will reply when he sees it.
[00:28] <UtUser> hopefully
[00:28] <UtUser> so I've got this huge video file for him to test
[00:29] <UtUser> what do you recommend?
[00:30] <J_Darnley> The usual line is: cut it down to the smallest section that will show your problem.  Are you asking how to do that?
[00:30] <wm4> UtUser: you said you use mpv to play it? did you try another player?
[00:30] <UtUser> that might be a good start, but I've done my best so far
[00:30] <UtUser> wn4: I should test it in VLC
[00:30] <J_Darnley> On an unrelated note, does anyone know who open billling system dot com are?  I've got an email from them about "business" and FFmpeg but it smells like spam.
[00:32] <UtUser> VLC does not squeal about errors
[00:32] <UtUser> like I said, it looks like an FFMPEG decoder issue
[00:33] <UtUser> must have been this way for 3 months since my build of MPV could be that old
[00:36] <cone-554> ffmpeg.git 03Andrew Kelley 07master:1d3eb0b5735f: oggenc: Move ogg_write_pages up above ogg_write_header
[00:36] <cone-554> ffmpeg.git 03Michael Niedermayer 07master:5313d60dc0f2: Merge commit '1d3eb0b5735f6dd72d2bf648455cf06ac7c24bc1'
[00:37] <UtUser> J_Darnley, I'm afraid I've never heard of that site but I would reply if ffmpeg was specifically mentioned unless you have a @ffmpeg.org email address
[00:38] <UtUser> how do I trim the file down to the nearest frame?
[00:39] <UtUser> no approximation, just the exact frame number
[00:39] <UtUser> the file is 1920x1080 24 fps, currently 2 seconds long and 78 MB
[00:39] <UtUser> rgb24 color space
[00:40] <cbsrobot> what frame do you need ?
[00:40] <UtUser> actually, I'm not sure
[00:40] <UtUser> the problem occurs on several frames
[00:40] <UtUser> and the error message has subtle variations, so I'm thinking all of the frames are important
[00:41] <UtUser> but I can trim off, say, the last 5 frames if necessary
[00:41] <UtUser> the file is exactly 48 frames long
[00:42] <J_Darnley> What file format?
[00:42] <cbsrobot> ffmpeg -i infile -t 1.791 -c copy outfile
[00:42] <UtUser> decimals, of course!
[00:42] <UtUser> J: it's MKV utvideo + wavpack
[00:46] <J_Darnley> If you can't get ffmpeg to do it, I think mkvmerge has splitting options (it has been a while since I've used it)
[00:47] <UtUser> cool beans
[00:47] <UtUser> ffmpeg does copy the streams pretty well though
[00:47] <UtUser> do you think it's safe to cut off a couple of redundant error messages at the end?
[00:48] <J_Darnley> No, just post everything printed by ffmpeg
[00:48] <UtUser> I mean the video file itself
[00:48] <UtUser> by error messages I mean bugged frames
[00:48] <J_Darnley> oh my bad
[00:49] <UtUser> sorry, I have a strange way of speaking
[00:49] <UtUser> I'm going to assume those frames are garbage unless you like to see repeated error messages...
[00:49] <J_Darnley> If you can reproduce your problem with just "1 error message" then yes, chop of anything not required
[00:49] <UtUser> technically I guess those frames must be different
[00:49] <UtUser> ok
[00:49] <J_Darnley> *off
[00:51] <UtUser> hell, the first frame should do it I guess
[00:52] <UtUser> it says too long VLC then it says cannot build huffman codes on a new line for each error
[01:06] <UtUser> I can't get it smaller than 20 MB
[01:06] <UtUser> well, really 19.6 MB
[01:06] <Daemon404> thats tiny; it's fine
[01:07] <UtUser> great
[01:07] <UtUser> and welcome back
[01:07] <UtUser> what do I do now?
[01:07] <Daemon404> also i didnt write te internal decoder, and i only mentored the internal encoder
[01:07] <Daemon404> but i do know the format
[01:07] <Daemon404> upload it somewhere
[01:07] <UtUser> ok
[01:07] <Daemon404> we have a samples ftp, but i dont know how to access it
[01:07] <Daemon404> any old link will do
[01:07] <Daemon404> ill just dl it
[01:08] <Daemon404> and mirror it or w/e
[01:08] <UtUser> OK, to keep it semi-permenant
[01:08] Action: Daemon404 brb; walking home
[01:22] <UtUser> uploading to 6 different sites now
[01:27] <Daemon404> O_o
[01:27] <UtUser> I'm doing some of your work for you
[01:28] <UtUser> it already finished on 3 sites
[01:28] <UtUser> http://www42.zippyshare.com/v/71324301/file.html
[01:28] <UtUser> http://uppit.com/ivu0tovxvtch/brokenframes.mkv
[01:28] <UtUser> http://hugefiles.net/k3xvup9eodjt
[01:30] <UtUser> my neighbor's music is so loud I can't think
[01:31] <UtUser> I
[01:31] <UtUser> 'm trying to remember the exact console output
[01:31] <UtUser> because it might red flag my ISP that the same upstream data sequence is being repeated so many times (probably not but w/e)
[01:31] <ubitux> what's broken?
[01:32] <ubitux> the frames look fine here
[01:32] <UtUser> try and transcode them in a newer ffmpeg
[01:32] <ubitux> there is a "Too long VLC (31) in init_vlc" error though
[01:32] <UtUser> exactly
[01:32] <UtUser> and an error that follows it
[01:32] <UtUser> can't remember what it was
[01:33] <ubitux> but the output looks fine anyway
[01:33] <ubitux> or is it not?
[01:33] <UtUser> it probably isn't
[01:33] <UtUser> see, utvideo is supposed to be a 100% lossless archival format
[01:33] <UtUser> and even other than that
[01:34] <UtUser> ffmpeg-based players like MPV will have trouble
[01:34] <UtUser> and according to some guy, the encoder is perfect
[01:34] <UtUser> so it must be the decoder
[01:34] <UtUser> maybe
[01:34] <ubitux> i tried ffplay and mpv and got an output that looks correct
[01:34] <ubitux> that's why i was asking
[01:34] <UtUser> because so many frames in the video are the same
[01:34] <UtUser> I think there are some non-broken frames in there
[01:35] <UtUser> so if it has trouble with a broken frame it should still show the non-broken frame
[01:35] <UtUser> I didn't cut it very exactly, see 
[01:35] <UtUser> that's why it's 20 MB
[01:36] <UtUser> http://www.sockshare.com/file/EBEC6306126D92F0
[01:37] <UtUser> both bitshare and filenuke are showing NaN%, so I assume they won't work
[01:37] <ubitux> (why are you uploading it in so much places?)
[01:37] <UtUser> IDK, posterity
[01:39] <UtUser> calling @Daemon404
[01:39] <Daemon404> i can grab it but i am not poking it tonight
[01:39] <Daemon404> nearly 1am
[01:39] <UtUser> oh boy
[01:40] <UtUser> well, I stayed up until 4:30 last night
[01:40] <UtUser> watching TV
[01:40] <cone-554> ffmpeg.git 03Andrew Kelley 07master:9767d7c092c8: oggenc: Flush after writing headers
[01:40] <cone-554> ffmpeg.git 03Michael Niedermayer 07master:b5657516b2b2: Merge commit '9767d7c092c890ecc5953452e8a951fd902dd67b'
[01:40] <cone-554> ffmpeg.git 03Michael Niedermayer 07master:919c320f7226: avformat/oggenc: fix EOS flag
[01:42] <UtUser> don't forget the ticket, guys
[01:42] <UtUser> http://trac.ffmpeg.org/ticket/2661
[01:44] <ubitux> this ticket is closed
[01:45] <Daemon404> g 52
[01:45] <Daemon404> fffs
[01:46] <UtUser> I know it's closed
[01:46] <UtUser> it should be re-opened
[01:46] <UtUser> g 52? no idea
[01:47] <Daemon404> typo
[01:47] <Daemon404> /g 52 switches channels
[01:47] <UtUser> k
[01:48] <UtUser> I think that ticket needs to be re-examined
[01:48] <UtUser> my point is that it was most likely broken again since then
[01:49] <UtUser> I've had that happen in my own projects where I thought a bug was fixed but then months later I'd re-discover it and realize it was unfixable by me
[03:49] <cone-554> ffmpeg.git 03Thierry Foucu 07master:12875df324e5: Timecode: Support 48fps
[03:49] <cone-554> ffmpeg.git 03Lazaros Koromilas 07master:bbc10a1aa537: libavformat/utils: Fix segfault on m4a cover artwork parsing
[04:28] <cone-554> ffmpeg.git 03Lukasz Marek 07master:8518da23f59d: ffmpeg: add -devices param
[12:20] <cone-584> ffmpeg.git 03Anton Khirnov 07master:f13ffb6636fd: flacdec: do not overwrite a channel layout set by the caller
[12:20] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:ef13967e601b: Merge commit 'f13ffb6636fdecb5e3e0ddcff48f096e7b3db362'
[12:26] <cone-584> ffmpeg.git 03Anton Khirnov 07master:d6b9ce99ea38: flac demuxer: parse the WAVEFORMATEXTENSIBLE_CHANNEL_MASK tag
[12:26] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:03ffaed3f019: Merge commit 'd6b9ce99ea384fb676561461768b8316725a4ccd'
[12:32] <cone-584> ffmpeg.git 03Anton Khirnov 07master:54ed488b1af5: flac muxer: write WAVEFORMATEXTENSIBLE_CHANNEL_MASK tag for multichannel files
[12:33] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:61301ca7860f: Merge commit '54ed488b1af583df6c9d2a73b4a44f16b7e4f82c'
[12:38] <cone-584> ffmpeg.git 03Anton Khirnov 07master:5fdaf312c554: flac: make avpriv_flac_parse_block_header() inline
[12:38] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:40beec6a431d: Merge commit '5fdaf312c5541b77b6364db8b49d6abb416a25c0'
[12:47] <cone-584> ffmpeg.git 03Anton Khirnov 07master:6df478bf891b: matroskadec: split parsing tracks into a separate function
[12:47] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:69de229bb881: Merge commit '6df478bf891b9fc5626e4a0b993899f310ba0a1c'
[12:56] <cone-584> ffmpeg.git 03Anton Khirnov 07master:4efdadc8ec50: matroskadec: export just the STREAMINFO block as FLAC extradata
[12:56] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:1c0ce2d4e0e7: Merge commit '4efdadc8ec50332c812e8a95e8c67f5a260e7cb0'
[13:08] <cone-584> ffmpeg.git 03Anton Khirnov 07master:23f741f79327: matroskadec: parse the channel layout mask for FLAC
[13:08] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:579e2b2874ae: Merge commit '23f741f79327e31be7b2a75ebb2e02111e06e52f'
[13:16] <cone-584> ffmpeg.git 03Anton Khirnov 07master:efcde917af40: vorbiscomment: simplify API by using av_dict_count()
[13:16] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:dff4aa6afd40: Merge commit 'efcde917af407a6031ecff68edd51fce7b83d104'
[13:52] <cone-584> ffmpeg.git 03Anton Khirnov 07master:106b62f4ba60: matroskaenc: write the channel mask for FLAC
[13:52] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:61917a104e0c: Merge commit '106b62f4ba600f24415eaded5e020aeceb23fd59'
[14:53] <cone-584> ffmpeg.git 03Diego Biurrun 07master:593886b5dd2b: ppc: util_altivec: Drop unused macros
[14:53] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:232073161494: Merge commit '593886b5dd2b583657aa3fd80bb173f702c9710d'
[15:13] <cone-584> ffmpeg.git 03Diego Biurrun 07master:b5aa48551300: ppc: Move vec_unaligned_load macro to util_altivec
[15:13] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:175aeab0b401: Merge commit 'b5aa48551300eed678aaea86ced7086758598a35'
[15:28] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:cf3a038f50e4: avformat/matroskaenc: fix ident string
[15:53] <Daemon404> michaelni, do you want anotehr fun jpeg sample?
[15:53] <Daemon404> looks like a variant of cmyk
[15:53] <michaelni> sure
[15:53] <Daemon404> pmed it to you
[15:54] Action: Daemon404 is remidned to dl that utvideo sample too
[16:39] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:ec33f59fed82: avcodec/mjpegdec: Support pix_fmt_id== 0x42111100
[16:40] <michaelni> Daemon404, fixed
[16:40] <Daemon404> nifty.
[16:42] <nevcairiel> more magic numbers!
[16:44] <Daemon404> theyre a mask of the subsamplign and stuff
[16:44] <nevcairiel> i know
[16:44] <Daemon404> its just implemented in a really, really weird wya
[16:44] <nevcairiel> no less magic tho
[16:44] <nevcairiel> should just implement all 32-bit combinations of it and be done with it
[17:09] <michaelni> llogan, theres a reply for you: https://github.com/FFmpeg/FFmpeg/pull/72
[17:10] <michaelni> comments from everyoe else welcome too
[18:46] <cone-584> ffmpeg.git 03Christophe Gisquet 07master:0810608e2345: x86: hevc_mc: better register allocation
[19:28] <cone-584> ffmpeg.git 03Lazaros Koromilas 07master:0f87b42b44f8: libavformat/utils: Warning about invalid cover art instead of an error
[19:50] <kurosu> michaelni, btw about very old patches, I still have a wmalosslesspro patch that soon should conflict with ape changes
[19:52] <cone-584> ffmpeg.git 03Vignesh Venkatasubramanian 07master:c14045c81867: lavf: Remove experimental flag for Opus in WebM
[19:53] <michaelni> kurosu, which wmalossless patch where ? and which ape changes where ?
[19:53] <kurosu> I'll send an updated changes
[19:53] <kurosu> one using the scalprod_and_madd from ape
[19:54] <kurosu> and I think libav is planning on rewriting it
[19:54] <kurosu> *it -> apedsp
[19:54] <kurosu> the change is not bad but takes shortcuts because it's only used in ape
[19:54] <kurosu> maybe I should wait for the change to be applied, actually
[20:08] <llogan> michaelni: i replied to the github PR.
[20:17] <jamrial> So who's gonna claim the comment #300 on the aac ticket?
[21:16] <UtUser> Daemon404: What about the Ut Video decoder?
[22:27] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:7dba055bb0d3: oggenc: Fix the EOS flag
[22:27] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:b3f9052177f2: Merge commit '7dba055bb0d31b2c79d9c3a66abb849a71417d3c'
[22:53] <cone-584> ffmpeg.git 03Konrad Iturbe 07master:c1485afe4975: README: add markdown syntax
[22:53] <cone-584> ffmpeg.git 03Konrad Iturbe 07master:50f4d0bb5052: LICENCE: add markdown syntax
[22:53] <cone-584> ffmpeg.git 03Konrad Iturbe 07master:c3ee6e3153c3: INSTALL: add markdown syntax
[23:02] <Dave77> are there any official examples in how to get ffmpeg to work with QT project?
[23:06] <wm4> quicktime or the gui lib
[23:24] <cone-584> ffmpeg.git 03Michael Niedermayer 07master:d228271813ab: tests/fate/opus: change test to use the official reference samples
[23:44] <cone-584> ffmpeg.git 03James Almer 07master:561bfc85eb26: x86/dsputilenc: implement SSE2 versions of pix_{sum16, norm1}
[23:47] <kurosu> michaelni, do you remember why you saturate the result of pix_sum16 to 2^16 (while I'd expect the result <=256*255) ?
[00:00] --- Thu May 29 2014

More information about the Ffmpeg-devel-irc mailing list