[Ffmpeg-devel-irc] ffmpeg-devel.log.20141104

burek burek021 at gmail.com
Wed Nov 5 02:05:03 CET 2014


[00:42] <cone-701> ffmpeg.git 03Michael Niedermayer 07master:09711545f5d2: avformat/rtmpproto: Do not fail when the length cannot be determined for live streams
[00:59] <kevmitch> are github pull requests encouraged/discouraged? 
[01:01] <Compn> encouraged
[01:01] <Compn> i think
[01:01] <Compn> put it another way, we like pull requests
[01:01] <kevmitch> good to hear. patches/email are so cumbersome
[01:01] <llogan> kevmitch: patches sent to ffmpeg-devel mailing list is the best way, mostly because some pull requests have been forgotten on occassion, but we've gotten better i think.
[01:02] <Compn> yes, if you want more review, patches to list are good, and can send with git send email
[01:02] <Compn> which is easier ... i'm not sure how easy than git pull request though
[01:02] <llogan> only a few people look at the pull requests, AFAIK.
[01:02] <kevmitch> yeah, but i have  to confdigure my MTA first
[01:03] <kevmitch> which is another error prone pain
[01:03] <llogan> you could send as an attachment made from git format-patch.
[01:03] <kevmitch> yeah, that's what I've been doing
[01:03] <kevmitch> gmail is not designed for it though
[01:04] <llogan> gmail is not designed for *
[01:04] <kevmitch> lol
[01:04] <llogan> if you *must* use pull request we can live with it
[01:07] <kevmitch> I'll just figure out my MTA and use git-email. I kind of got the impression that the turnaround might be longer on github.
[01:13] <J_Darnley> If you can't get git send-email to work we also tolerate attached pacthes from git format-patch
[01:28] <Compn> kevmitch / llogan : we could make our own email server for people who use gmail but want to send patches
[01:29] <Compn> i mean... free email server for devels and git-send-email :P
[01:29] <Compn> gmail people could still then reply on list
[01:29] <Compn> just git would use our servers
[01:42] <kevmitch> Compn: it should be possible to git-send email through gmail, so I don't know if that'd be necesssary
[01:43] <kevmitch> although it would make it easy to provide a very general set of instructions for setting up git send-email I guess
[01:45] <jamrial> of course it's possible. just configure their smtp server on git
[01:49] <J_Darnley> The only problem I saw with using gmail was that the password appears to be stored in plain text.
[01:50] <llogan> then your password should be password = "Enter password here"
[01:51] <llogan> nobody would suspect it
[01:51] <J_Darnley> :)
[01:52] <J_Darnley> I just abuse my ISP's mail server which apparently lets me send mails without a password and with lying about the from address.
[01:54] <llogan> is there a wtv file in fate suite samples? i'm downloading but it's slow here in 2001 network tech land. i accidentally deleted my samples dir.
[03:54] <Compn> llogan : theres some wtv samples on samples repo
[03:54] <Compn> i'm not sure but i think the fate samples are http browseable somewhere
[10:09] <j-b> 'morning
[10:28] <ubitux> what would be the old way of doing ffmpeg -i in.flv -af aresample=first_pts=0 -y out.wav?
[10:29] <ubitux> did we have such option in the past?
[10:29] <ubitux> (in order to pad with silence at the beginning)
[10:36] <pross> does libavfilter support the concept of 'subtitle' filtering, for example: to strip or add coloring
[11:28] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:3dae05f4f7a8: avformat/mpegts: also print PMT version in av_dlog()
[12:49] <cbsrobot> BuxiNess_: https://github.com/CodecCentral/roger
[12:52] <cbsrobot> The author said he tries to have an alpha version this spring
[12:53] <BuxiNess_> cbsrobot, many thks
[14:02] <ubitux> pross: no
[14:29] <cone-311> ffmpeg.git 03Benoit Fouet 07master:dc351e138190: id3v2: prefer TDRC for date over TDRL.
[14:44] <Compn> pross: wouldnt converting ass to srt automatically take out a lot of style ? also mplayer -dumpsrtsub :P
[14:44] <Compn> not even sure that works anymore
[18:14] <pross> Compn: thanks. nah i want to manipulate the text and add my own colors
[18:37] <ChrisHan_> Hi guys
[18:39] <ChrisHan_> areany of the devs online this evening
[18:41] <iive> nope
[18:42] <iive> it is supper time in Europe
[18:42] <ChrisHan_> I'm in europe too lol
[18:42] <ChrisHan_> But okay, I'll try later
[18:48] <iive> ChrisHan_: just ask your question, so people can see it and answer if/when they can.
[18:49] <ChrisHan_> A new version of x265 has been released today that significantly speeds up encoding. I was wondering when this will be included into FFMpeg.
[18:50] <ChrisHan_> Then, if I hopefully get an answer to this question, I will then ask the Handbrake devs when they will include the new FFMpeg version in their SVN builds on their IRC.
[18:50] <iive> ffmpeg uses it as external library. if there are no api changes, it should just work.
[18:51] <ChrisHan_> So if I compiled FFMPEG with the new x265 build it should just sort of slot in and build?
[18:51] <iive> x265 does releases? or they just merged some speed up changes?
[18:51] <Daemon404> they do releases yes
[18:51] <jamrial> they do a release every 4 to 6 weeks
[18:51] <ChrisHan_> it was on their Facebook page today they have released 1.4
[18:51] <Daemon404> the api has been stable
[18:51] <Daemon404> you can build ffmpeg with it
[18:54] <ChrisHan_> Great! thanks
[18:54] <j45> ChrisHan_: I've already tested x265 1.4 in HandBrake.  There is a review request currently under review for it.  It is *not* significantly faster.
[18:54] <ChrisHan_> OH, I was under the impression it was, my fault.
[18:54] <Daemon404> j45, rd mode 4 and 5 are significantly more parallel now
[18:54] <Daemon404> when you use pmode
[18:54] <Daemon404> but thats only going to help if you have more than 10-16 cores
[18:55] <ChrisHan_> I'm encoding at 0.7FPS atm on Handbrake SVN
[18:55] <ChrisHan_> was hoping for at least 1FPS lol
[18:55] <iive> is that HD?
[18:55] <ChrisHan_> 1080p
[18:55] <j45> ok.  I have 8 cores on my best machine and I've only tested a few presets
[18:55] <Daemon404> j45, its mostly useful for placebo in data centers
[18:55] <Daemon404> not normal people using handbrake
[18:55] <j45> ChrisHan_: https://reviews.handbrake.fr/r/773/
[18:56] <ChrisHan_> Thanks j45
[18:56] <ChrisHan_> Avatar is 3 hours long and looks shitty on x264 at 3000kbps. Hence x265
[19:00] <jamrial> h264 3000kbps is kinda low even for 1280x720
[19:00] <ChrisHan_> 6000kbps is fine for normal films. I'm squeezing them onto a 4.7GB DVD
[19:01] <jamrial> also, unless you have a haswell CPU you wont see any real speed gain from this new x265 version
[19:01] <ChrisHan_> Dual core ivy bridge in laptop
[19:01] <jamrial> last time i checked, most development was on avx2 simd
[19:01] <ChrisHan_> It's got a GeForce 660M but they don't have CUDA or OpenCL in it yet
[19:02] <jamrial> then you're not going to see much difference compared to 1.3. like the guy in that handbrake review
[19:02] <ChrisHan_> How gutting. Guess my encode can wait a week lol
[19:03] <j45> i was testing on a quad haswell laptop. but haven't really compared the slower presets much yet.
[19:03] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:e4f8a973aa8b: swresample: Fix swr_drop_output so it does not flush the buffers
[19:03] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:97da68172aa7: avfilter/af_aresample: split flushing code out
[19:03] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:09024fe681e1: avfilter/af_aresample: Limit data per inserted packet
[19:03] <ChrisHan_> I CBA splashing out on a bluray writer so trying to cram into regualr DVDs
[19:04] <ChrisHan_> anyway thanks for the help
[19:43] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:786594184a17: avformat/mpegts: fix iteration count in add_pid_to_pmt()
[21:00] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:66b9e60af0b8: ffmpeg_opt: store canvas size in decoder context
[21:08] <cone-311> ffmpeg.git 03Vittorio Giovara 07master:351d0f8b7a6e: get_bits: remove unused assignment
[21:09] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:a28313b09e73: Merge commit '351d0f8b7a6ecce411ae75fb3511573c34317218'
[21:17] <cone-311> ffmpeg.git 03Vittorio Giovara 07master:240b22afe14e: motion_est: remove dead code
[21:17] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:567ea2d64c63: Merge commit '240b22afe14ef477da1b439b9ed7bca6cc7d6c26'
[21:37] <cone-311> ffmpeg.git 03Vittorio Giovara 07master:c442190a6bfd: error_resilience: initialize prev_* variables
[21:38] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:e1bcbca9987e: Merge commit 'c442190a6bfd8036f6c32b78e1e96ff3b830f8f0'
[21:52] <cone-311> ffmpeg.git 03Vittorio Giovara 07master:f52b8717617e: celp_filters: don't use filter lenght as loop bound
[21:52] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:d9103e848dae: Merge commit 'f52b8717617e94da90a45afdfff23e94f9ecf35c'
[22:00] <cone-311> ffmpeg.git 03Vittorio Giovara 07master:9f6f407463ff: aacsbr: treat 1-d arrays as such
[22:01] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:ce63cb4ff217: Merge commit '9f6f407463ff8b7681befd04b6655bb7c6d9b3e1'
[22:31] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:930ffd46e1e7: aacsbr: change order of operation to prevent out of array read
[22:31] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:df82125acba5: Merge commit '930ffd46e1e742674aa7cc1c2450020c63b5015b'
[22:46] <cone-311> ffmpeg.git 03Vittorio Giovara 07master:77ab341c0c6c: aacdec: add default case in channel layout
[22:46] <cone-311> ffmpeg.git 03Michael Niedermayer 07master:ae4bb6c4884f: Merge commit '77ab341c0c6cdf2bd437bb48d429e797d1e60da2'
[23:14] <wm4> can anyone tell whether the bayer pixdesc formats are correct
[23:14] <wm4> the 8 bit formats all say that the depth is R=2, G=4, B=2
[23:15] <wm4> so that amounts to 8 bit per pixel?
[23:15] <wm4> instead of 8 bit per component? (with components skipped)
[23:17] <wm4> e.g. AV_PIX_FMT_BAYER_BGGR8 is documented as having "BGBG" components (for odd lines), with 8 bit per sample, so that means B actually has 4 bit per pixels, not 2?
[23:18] <wm4> hm I guess it becomes 2 bit by average
[23:18] <wm4> but that really seems... surprising, in context of pixdesc as generic description
[23:27] <wm4> (though I bet these bayer formats completely break drawutils)
[00:00] --- Wed Nov  5 2014


More information about the Ffmpeg-devel-irc mailing list