[Ffmpeg-devel-irc] ffmpeg-devel.log.20130517

burek burek021 at gmail.com
Sat May 18 02:05:02 CEST 2013


[00:11] <ubitux> trac is having a hard time tonight
[00:11] <ubitux> it is extremely slow..
[00:19] <Daemon404> ubitux, it's always been quite slow for me
[00:19] <Daemon404> michael could never reproduce tho
[00:19] <Daemon404> iirc
[00:20] <michaelni> ubitux, are you using trac.ffmpeg.org or ffmpeg.org/trac ? and how to reproduce ?
[00:21] <ubitux> it's ok now
[00:22] <ubitux> seems i was on ffmpeg.org/trac
[00:22] <ubitux> it was just for a few minutes
[00:22] <ubitux> it's fast now :)
[00:32] <ubitux> burek: could you fix the README.html in the {32,64}bit directories of your static builds?
[00:32] <ubitux> it seems to confuse users
[00:32] <ubitux> (and i'm not sure empty readmes make much sense)
[00:36] <cone-801> ffmpeg.git 03Michael Niedermayer 07master:ab074ab9fef8: avcodec/pthread: fix ffv1 with 2 threads
[00:45] <cone-801> ffmpeg.git 03James Almer 07master:54372848f327: tools/ffhash: Use O_BINARY when available
[00:47] <cone-801> ffmpeg.git 03Stefano Sabatini 07master:0aa013fa8a5f: lavfi: factorize drawgrid and drawbox code
[00:47] <cone-801> ffmpeg.git 03Stefano Sabatini 07master:cb0b0266e5c9: lavfi/drawbox: restore verbal form for drawbox options descriptions
[03:33] <cone-801> ffmpeg.git 03Carl Eugen Hoyos 07master:17989dcf540c: configure: Autodetect VA-API
[03:52] <highgod> Hi, I want to ask a question about audio, I found that ffmpeg out put a frame after decoding audio, and I want to save the content of it, what format should I save, wav or something? how can I play it, thanks.
[07:36] <burek> ubitux, try to View Source on those README.html files and you'll get the answer why do I keep them :)
[07:40] <burek> but I'll try to find a better way to do that css styling
[07:40] <burek> not to confuse people
[09:33] <burek> at http://www.ffmpeg.org/ffmpeg-filters.html#fps
[09:33] <burek> there is a 2nd example: fps=fps=film:round=near
[09:34] <burek> "film" is never listed anywhere in this document, so it might be a good idea to insert the list of such abbreviations?
[09:56] <saste> burek, ffmpeg-utils(1)
[09:58] <burek> saste, you mean man ffmpeg-utils
[09:58] <burek> or something else?
[10:00] <burek> oh, you mean http://www.ffmpeg.org/ffmpeg-utils.html ok
[10:00] <burek> thanks
[10:01] <burek> would it be convenient that we add just a link to the http://www.ffmpeg.org/ffmpeg-utils.html#Video-rate
[10:01] <burek> not to duplicate the content
[10:04] <highgod> Hi, all ,I want to ask a question, I saw that the audio is decoded in a avframe, is it a pcm format?
[10:18] <cone-86> ffmpeg.git 03Jeremy Hinegardner 07master:28e6b7b9b4e0: Turn off usage of base_data_offset in tfhd for PIFF files
[10:26] <burek> when i think about it, it's better not to add any link to abbreviations for fps, since the structure of the documentation can be changed at any time and the link might break..
[10:26] <BuxiNess> Morning
[10:27] <burek> anyway, some reference to those abbreviations is needed in the fps filter docs, if we're going to use such an example, so that people can see the list of all possible abbrev.
[10:34] <cone-86> ffmpeg.git 03Sebastian Sandberg 07master:1fb013a57c6d: vc1dec: Add support for interlaced B-frames
[10:34] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:c453da33a3de: Merge commit '1fb013a57c6d98d3f425506eb25f32b2cf7ddc6f'
[10:37] <burek> one more bug, http://www.ffmpeg.org/ffmpeg.html describes -loop_input and -loop_output, referencing -loop (because first two are deprecated), but there is no entry for -loop ?
[10:39] <michaelni> morning BuxiNess 
[10:41] <BuxiNess> morning
[10:46] <michaelni> BuxiNess, as you are maintainer of jpeg2000 you should have op here
[10:48] <BuxiNess> michaelni, ok
[10:49] <cone-86> ffmpeg.git 03Jeremy Hinegardner 07master:40bbbce7e8ad: movenc: Don't add base_data_offset in tfhd for PIFF files
[10:49] <cone-86> ffmpeg.git 03Kostya Shishkov 07master:1232a1647ab2: Apple Intermediate Codec decoder
[10:49] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:5bc892cb1722: Merge commit '1232a1647ab27e024a3baf4d01d40c8d08d6ced9'
[10:59] <cone-86> ffmpeg.git 03Anton Khirnov 07master:b06848f4decc: vf_lut: make config_props work properly when called multiple times.
[10:59] <cone-86> ffmpeg.git 03Anton Khirnov 07master:87c31cfd7a2c: vf_gradfun: make config_props work properly when called multiple times.
[10:59] <cone-86> ffmpeg.git 03Anton Khirnov 07master:d371c3c2e283: vf_frei0r: make config_props work properly when called multiple times.
[10:59] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:d5e5e73c52d3: Merge commit 'd371c3c2e2830d9783465ecfe1ab7d93351083b7'
[11:11] <cone-86> ffmpeg.git 03Anton Khirnov 07master:7727be79d1ee: vf_fps: move initializing pts from config_props to init.
[11:11] <cone-86> ffmpeg.git 03Anton Khirnov 07master:90d9a2a04c3f: vf_drawtext: do not reset the frame number in config_input.
[11:11] <cone-86> ffmpeg.git 03Anton Khirnov 07master:4c205f42c86c: vf_drawbox: make config_props work properly when called multiple times.
[11:11] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:c7078f4907dc: Merge commit '4c205f42c86ccefa093c59434669af34ad14a52b'
[11:40] <cone-86> ffmpeg.git 03Anton Khirnov 07master:d5c66d9c5612: vf_setdar: make config_props work properly when called multiple times.
[11:40] <cone-86> ffmpeg.git 03Anton Khirnov 07master:6592cd22a230: vf_crop: make config_props work properly when called multiple times.
[11:40] <cone-86> ffmpeg.git 03Anton Khirnov 07master:3fb29588a27a: vf_drawtext: don't leak the expressions.
[11:40] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:71fcb6072ea6: Merge commit '3fb29588a27a711132106b924e27b53789a58dcb'
[11:47] <cone-86> ffmpeg.git 03Anton Khirnov 07master:3ba35a346cd2: vf_hqdn3d: make config_props work properly when called multiple times.
[11:47] <cone-86> ffmpeg.git 03Anton Khirnov 07master:16a645adeb75: vf_pixdesctest: make config_props work properly when called multiple times.
[11:47] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:1e9a050bc020: Merge commit '16a645adeb758207346a4bbf66766f02734c461e'
[11:53] <ubitux> https://www.youtube.com/watch?v=K6JshvblIcM
[11:55] <cone-86> ffmpeg.git 03Anton Khirnov 07master:30e256430eb8: lavc/pthread: remove obsolete checks
[11:55] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:36d349034e19: Merge commit '30e256430eb88c6d4c382581b89bca171d79fbc0'
[12:01] <cone-86> ffmpeg.git 03Anton Khirnov 07master:db8e336e5fc1: pthread: reindent
[12:01] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:1a0057dc881c: Merge commit 'db8e336e5fc18c41cb4e9bab1960349e9800d546'
[12:08] <burek> ubitux +1
[12:18] <cone-86> ffmpeg.git 03Jindrich Makovicka 07master:259af1b92370: mpegvideo: allocate sufficiently large scratch buffer for interlaced vid
[12:18] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:1ec578c010e7: Merge remote-tracking branch 'qatar/master'
[12:49] <ubitux> i wonder if those config_props make that much sense in all cases
[12:49] <ubitux> from what i can tell, a lot of filters could just have a flag to say they are "uninit" compliant
[12:49] <ubitux> which means uninit() could be called before config_props()
[12:50] <ubitux> well, let's just wait to see what anton will come up with
[12:58] Action: durandal_1707 *powerpoint
[13:13] <michaelni> smarter, how is your decoder going ? can something already be merged ?
[13:14] <durandal_1707> michaelni: what's about alpha in rgba64 output from non-alpha source, is that initialization really needed?
[13:14] <smarter> michaelni: not yet, but it's starting to get into shape
[13:15] <smarter> michaelni: beside various cleanups, I hope to be able to merge the parallel decoding stuff that is being worked on in https://github.com/OpenHEVC/openHEVC/tree/hm10.0
[13:17] <michaelni> smarter, ok
[13:18] <michaelni> also dont hesitate to ping me if theres something to merge into ffmpeg or something to review
[13:18] <michaelni> or if you have questions about any code from h264*.c
[13:19] <smarter> sure, thanks
[13:19] <michaelni> durandal_1707, the alpha in rgba64 that is output should be initilaized to "opaque"
[13:19] <smarter> I looked into reusing some of h264_parser.c but it seems pretty specific to h264
[13:20] <durandal_1707> michaelni: yes, but i dont see similar code for 8bit case
[13:21] <michaelni> smarter, yes, the startcode search code could be reused i guess unless its already done
[13:21] <michaelni> durandal_1707, do we output proper opaque alpha for 8bit ?
[13:22] <smarter> I think ff_h264_find_frame_end looks for things like end of slices that are different in h264 and hevc
[13:26] <michaelni> it parses the first mb index and nal type, id have to cross check h264 & 5 to know how much that differs
[13:29] <michaelni> also it contains code to parse the length values that are used in h264 in mp4 but not in raw h264
[13:32] <durandal_1707> michaelni: you incorrectly merge doc/APIchanges
[13:33] <cone-86> ffmpeg.git 03Paul B Mahol 07master:3a751eab7860: lavu/opt: add AV_OPT_TYPE_COLOR
[13:33] <cone-86> ffmpeg.git 03Paul B Mahol 07master:05b7560d39ab: lavfi/color: switch to AV_OPT_TYPE_COLOR
[13:33] <cone-86> ffmpeg.git 03Paul B Mahol 07master:83e892233c87: lavfi/life: switch to AV_OPT_TYPE_COLOR
[13:33] <cone-86> ffmpeg.git 03Paul B Mahol 07master:1d2ecf9609a7: lavfi/drawtext: switch to AV_OPT_TYPE_COLOR
[13:33] <cone-86> ffmpeg.git 03Paul B Mahol 07master:85e8a1169cfb: lavfi/pad: switch to AV_OPT_TYPE_COLOR
[13:41] <michaelni> durandal_1707, please elaborate, which commit did i mess up in which way ?
[13:43] <durandal_1707> the .0 changes from libav are in wrong place
[13:44] <michaelni> you mean the whole is not in chronological order ?
[13:49] <michaelni> I admit i did not pay much attention to the order, ill try to pay more attention to it from now
[13:49] <michaelni> about the existing entries, someone should fix their order
[13:50] <michaelni> i dont volunteer though
[13:55] <Compn> Apple Intermediate Codec decoder
[13:55] <Compn> gasp
[13:56] <Compn> :)
[13:56] <Compn> another codec down!
[14:59] <durandal_1707> ubitux: is #1907 related to subtitles?
[15:02] <ubitux> no idea
[15:02] <ubitux> possibly
[15:02] <ubitux> i have no idea how to reproduce though
[15:03] <durandal_1707> seek
[15:05] <ubitux> i don't see any sample
[15:06] <ubitux> is this detected by valgrind btw?
[15:09] <durandal_1707> address sanitizer
[15:09] <ubitux> yeah but i don't have it, so i'm wondering if i can reproduce with valgrind
[15:20] <cone-86> ffmpeg.git 03Vignesh Venkatasubramanian 07master:8ffcc826b3f3: Adding support for muxing VP8 Alpha files
[15:20] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:2fb193b1e742: vformat/matroskaenc: factorize alpha code in mkv_write_block()
[15:20] <cone-86> ffmpeg.git 03Vignesh Venkatasubramanian 07master:98c292a7a8f2: Adding support for encoding VP8 Alpha
[15:20] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:ebe8c7fe5201: vc1dec: restructure interpmvp code
[15:34] <Compn> pew pew pew
[15:34] Action: Compn shoots at cehoyos
[16:20] <durandal_1707> ubitux: can i push to files you are maintainer(libmodplug) without specific approval?
[16:21] <ubitux> i guess :)
[16:21] <ubitux> you can just show me the commit here if you want a quick review
[16:21] <ubitux> you want to adjust probing?
[16:22] <ubitux> if you wish to remove 'mod' as an extension i don't think that's a good idea
[16:22] <ubitux> since that's a very common extension
[16:22] <ubitux> for midi-like files
[16:25] <durandal_1707> ubitux: i did just s/int sz/int64 sz/ 
[16:25] <durandal_1707> for >2 gb mod files :)))))
[16:26] <ubitux> not a good idea
[16:26] <ubitux> because there is a cache
[16:26] <ubitux> there is a design issue in libmodplug
[16:26] <ubitux> which forces you to store the whole file in memory iirc
[16:27] <durandal_1707> avio_size returns int64, which in code gets into int
[16:27] <ubitux> ah, well ok
[16:27] <ubitux> feel free to fix that
[16:32] <durandal_1707> Compn: software to show at people?
[16:33] <durandal_1707> *shot lol
[16:34] <Compn> durandal_1707 : haha , well you know ... dont want to work on software that can be used against you
[16:34] <Compn> :P
[16:34] <Compn> skynet, etc
[16:35] <durandal_1707> we really need such filter
[16:35] <Compn> what? facial recognition ?
[16:35] <durandal_1707> recognition of interesting stuff....
[16:35] <durandal_1707> *detection....
[16:35] <Compn> ah
[16:43] <durandal_1707> hmm so file is probed as mod(score 50) instead mpeg(score 52) just because libmodplug have higher priority than mpeg for some reason
[16:43] <ubitux> 'o' is before 'p'? :p
[16:44] <durandal_1707> but i thougt that all are probed and the one with highest score wins
[17:00] <Compn> 2gb mod files? what?
[17:39] <durandal_1707> no more comments on ftp patch?
[18:13] <cone-86> ffmpeg.git 03Paul B Mahol 07master:d26f637b020e: libmodplug: fix sz type
[18:44] <durandal_1707> anyone have java installed?
[18:47] <durandal_1707> ubitux: what's status of vignette filter
[18:47] <ubitux> i need to add dithering
[18:48] <ubitux> and eventually add other formulas since michaelni seems to be expecting something different
[18:49] <durandal_1707> sure that "vignette" name would be correct term in such case?
[18:49] <ubitux> i could rename to lens
[18:49] <ubitux> but that's a limited use of "lens"
[18:57] <cone-86> ffmpeg.git 03Paul B Mahol 07master:47ac344970f1: id3v2enc: chapter support
[19:35] <j-b> BBB: do you know people working on libyuv?
[19:35] <JEEB> did you notice the mingw compilation failure?
[19:39] <BBB-work> j-b, yes
[19:39] <BBB-work> j-b, any specific message I should convey?
[19:41] <j-b> BBB-work: is it good or as bad as it looks like ? :D
[19:43] <BBB-work> j-b, ?
[19:43] <BBB-work> j-b, you'll have to be more specific
[19:43] <BBB-work> I can interpre that in many ways
[19:43] <j-b> BBB-work: nothing specific, I fell on it, and wondered how good it was to integrate in VLC or not
[19:44] <j-b> BBB-work: (I was looking for fast i420->nv12 conversion)
[19:45] <nevcairiel> i420 -> nv12 is so simple that even thinking about getting a lib for that is just insane :P
[19:46] <j-b> nevcairiel: read better what I said.
[19:47] <nevcairiel> its also simpel to make it fast
[19:47] <j-b> with pump* yes
[19:47] <nevcairiel> find someone with just a tiny bit of SIMD experience on your target platforms, really not much shuffling required
[19:47] <j-b> but I was looking for a simple good example to show to my students
[19:48] <j-b> and I fell on libyuv
[19:51] <j-b> because even if "simple" there are not many good examples on the net.
[19:52] <nevcairiel> this is true
[19:52] <j-b> and more examples of the opposite
[19:52] <j-b> swscale code scares me
[19:52] <gnafu> Hehe.
[19:59] <BBB-work> j-b, it's fast. quality-wise it doesn't always do the very-best (read: slower) thing, but it's fast and chrome uses it
[20:01] <Compn> hardware acceleration in a web browser :\
[20:01] <Compn> j-b : you should show them swscale. either they learn how not to do it, or one of them understands it and can come help us with it :D
[20:02] <j-b> BBB-work: ok
[20:02] <j-b> :)
[20:02] <Compn> people creating video filters to do colorspace conversions instead of adding it to swscale :\
[20:02] <Compn> (re vf xyz )
[20:05] <BBB-work> it's not like xyz is sweet
[20:48] <nevcairiel> Compn: swscale can convert from xyz to rgb now
[21:11] <ubitux> so windows build are stalled?
[21:11] <ubitux> (zeranoe)
[21:11] <ubitux> because of the windows bug?
[21:12] <michaelni> didnt i read the bug was fixed ...
[21:13] <ubitux> no idea
[21:13] <ubitux> the builds on zeranoe are not up-to-date
[21:13] <ubitux> and it happens that a user need a feature recently added
[21:14] <nevcairiel> the bug in mse was indeed fixed
[21:14] <nevcairiel> i could produce a static build without external libs if anyone needs it desperately
[21:15] <michaelni> ubitux, did you or someone else contact zeranoe and inform him that the bug has been fixed ? if not someone shoukd
[21:15] <michaelni> shouLd
[21:16] <ubitux> i don't plan to do
[21:17] <ubitux> heh seems the guy get it working anyway
[21:17] <ubitux> derp
[21:17] <ubitux> the build is up-to-date
[21:18] <ubitux> just a brain fart, forget all i said.
[21:19] <cone-86> ffmpeg.git 03Reimar Döffinger 07master:24c65eb29f05: md5: avoid unnecessary copying.
[21:21] <TheRyuu> 21:14 <+nevcairiel> the bug in mse was indeed fixed <---what did they break this time?
[21:21] <nevcairiel> it crashed when scanning swscale
[21:21] <ubitux> after xyz addition
[21:21] <TheRyuu> el oh el
[21:21] <JEEB> yeh
[21:21] <JEEB> it was a funny thing
[21:22] <JEEB> I think I even once got worried what on earth happened when MSE just told me it had gotten disarmed
[21:22] <TheRyuu> I don't think they 'fixed' the whole concurrent write thing yet though
[21:22] <ubitux> if (TheApplicationName == "swscale") return 0;
[21:22] <TheRyuu> although that's not really a problem if it's a gcc build
[21:23] <TheRyuu> mine doesn't like the example program when compiling zlib
[21:23] <TheRyuu> herp derp antivirus
[21:25] <ubitux> so the md5 optim just added is supposed to make fate faster?
[21:26] <ubitux> or the switch to crc/murmur3 was done instead already?
[21:30] <michaelni> its all improving in small steps, faster md5 is good for more than fate
[21:30] <michaelni> fate wasnt switched yet to something else
[21:32] <ubitux> ok
[21:36] <Compn> nevcairiel : i know . but i'm saying the person who wrote the filter couldnt figure out how to add xyz to swscale
[21:36] <Compn> and we should fix that
[21:36] <Compn> if possible
[22:16] <durandal_1707> "bugs only encountered by drones" ...
[22:17] <cone-86> ffmpeg.git 03Michael Niedermayer 07master:6c593f1b671b: avidec: dont randomly skip packets for offseting the index
[23:03] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/0.10:1c655c1419af: avidec: dont randomly skip packets for offseting the index
[23:03] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/0.11:f7dbe180a02f: avidec: dont randomly skip packets for offseting the index
[23:03] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/0.9:b22cc0c15de7: avidec: dont randomly skip packets for offseting the index
[23:03] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/1.0:663b22325891: avidec: dont randomly skip packets for offseting the index
[23:03] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/1.1:f544553c2993: avidec: dont randomly skip packets for offseting the index
[23:03] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/1.2:fd60aeb55b12: avidec: dont randomly skip packets for offseting the index
[23:03] <cone-86> ffmpeg.git 03Carl Eugen Hoyos 07release/1.2:af589dd5e9f8: configure: Autodetect VA-API
[23:28] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/1.0:14634f5b698d: ffmpeg: free threads on error conditions.
[23:28] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/1.1:85277ff9364c: ffmpeg: free threads on error conditions.
[23:28] <cone-86> ffmpeg.git 03Michael Niedermayer 07release/1.2:ffd28de38863: ffmpeg: free threads on error conditions.
[00:00] --- Sat May 18 2013


More information about the Ffmpeg-devel-irc mailing list