[Ffmpeg-devel-irc] ffmpeg-devel.log.20141114

[burek]

[00:41] <ubitux> random news: $100 SPI donation incoming (right i know that's not that much) from a marketing company who was happy to get help to concatenate files at format level
[00:46] <J_Darnley> neat
[00:58] <Rexor> hi. reasking a question from #ffmpeg. are there any plans to support tiled OpenEXR files ?
[01:12] <J_Darnley> I have no idea what that is but from the name I reckon it is already supported except in a rather inefficient manner.
[01:13] <rcombs> ubitux: "use the concat demuxer kthx"
[01:13] <rcombs> "that'll be $100"
[03:09] <cone-809> ffmpeg.git 03Peter Ross 07master:b186b7131e16: avfilter/vf_lut: gammaval709()
[06:20] <rcombs> is it just me, or is VideoLAN git being a bit unreliable?
[08:41] <cehoyos> j-b: Hi, is it known that the git server refuses connections?
[09:46] <j-b> cehoyos: no
[09:47] <cehoyos> The following does not work here:
[09:47] <cehoyos> git clone git://git.videolan.org/ffmpeg.git test
[09:48] <cehoyos> Same with: git clone git://source.ffmpeg.org/ffmpeg.git test
[09:53] <cehoyos> j-b: Can you reproduce?
[09:55] <j-b> not at all
[09:55] <cehoyos> The error message shown is "fatal: read error: Connection reset by peer"
[09:56] <cehoyos> Reproducible from servers located in other parts of the world.
[09:58] <cehoyos>  j-b: Does cloning work for you?
[09:58] <j-b> yes
[09:59] <cehoyos> Can you try from another server: I tried from a Debian test server and from gcc buildfarm and from ffmpeg.org
[10:02] <cehoyos> Same for "git clone git://git.videolan.org/vlc.git"
[10:04] <saste> cehoyos, same problem here
[10:04] <cehoyos> Grazie!
[10:05] <j-b> cehoyos: try.
[10:06] <cehoyos> It seems to work now, thank you!
[10:06] <cehoyos> Yes, works now!
[10:07] <cehoyos> Why did it work for you?
[10:48] <ubitux> rcombs: i actually wrote a 15 lines shell script, but yeah that's the idea
[11:16] <xxthink> is it possible to set the encoding speed?
[11:16] <av500> speed?
[11:16] <xxthink> -re option does not release the cpu
[11:17] <xxthink> I want to set the encoding speed to 25fps
[11:17] <xxthink> the input is a file on my disk
[11:18] <xxthink> I think if the speed exceeds 25fps, the cpu should be release to do other work
[11:19] <xxthink> av500, do you think so?  
[11:20] <xxthink> I'm debugging melt to set the preview speed.
[11:28] <ubitux> maybe we could introduce a -vf speed that does some sleep() :-°
[11:28] <pomaranc> hi, is there any way to get the current sar in a write_packet muxer function?
[12:25] <saste> pomaranc, don't think so
[15:48] <cone-979> ffmpeg.git 03Benoit Fouet 07master:98abb98cb1c7: avcodec/pngdec: rename decode_frame to decode_frame_png
[16:15] <cone-979> ffmpeg.git 03Benoit Fouet 07master:1523d1484d2e: avcodec/pngdec: create a function to decode IHDR chunk.
[16:15] <cone-979> ffmpeg.git 03Benoit Fouet 07master:3f1eaf590ca2: avcodec/pngdec: create a function to decode pHYs chunk.
[16:39] <gsuberland> hullo. got a question that's not specifically about ffmpeg, but requires some knowledge of video file formats, so I figure it's the best place to ask.
[16:40] <gsuberland> I'm trying to create a "datamoshing" effect (glitches) from video samples. If I were to take, for example, a pair of identically encoded, identical-ish length, identical resolution videos with different content, then xor the two together and preserve the header, am I likely to find that it's renderable?
[16:44] <rcombs> probably not
[16:44] <J_Darnley> Probably not.  Different content will lead to differently sized coded frames.
[16:44] <gsuberland> interesting
[16:44] <gsuberland> is that the case for pretty much all formats?
[16:45] <gsuberland> (barring uncompressed AVI)
[16:45] <J_Darnley> Most common "consumer" formats, yes.
[16:45] <gsuberland> bugger.
[16:45] <rcombs> maybe for pro codecs like ProRes, but I wouldn't bet on it
[16:46] <gsuberland> my normal trick is just to find/replace various sequences in a hex editor to generate glitches, then screencap them back off of VLC
[16:46] <gsuberland> 99% of the time it renders just fine and has the usual blocking glitches
[16:50] <beastd> gsuberland: hmm in that case, you could go to the same length with ffmpeg. decode the 2 videos -> combine them to your creativity in the filter graph -> encode the resulting video
[16:50] <cehoyos> daemon404: I can't help but think this is our own doing <- I don't think so, why didn't you ask for the usecase / the sample?
[16:51] <Daemon404> because we enabled everyone to encode prores
[16:51] <Daemon404> what do you think all the horrible chinese things use to encode?
[16:51] <gsuberland> beastd - I'm not really familiar with that terminology.
[16:51] <Daemon404> they hack up ffmpeg.
[16:51] <Daemon404> also i have the sample.
[16:51] <Daemon404> i work for the source
[16:51] <cehoyos> Could you provide it?
[16:51] <Daemon404> not without asking the user first
[16:51] <cehoyos> (I believe I asked for such a sample a few times.)
[16:52] <Daemon404> i will ask the community team to ask.
[16:52] <cehoyos> Unfortunately the patch means that such samples will be created which I thought you would consider hackish.
[16:52] <cehoyos> Thank you.
[16:52] <Daemon404> oh
[16:52] <Daemon404> hmm
[16:52] <cehoyos> Could you provide ffmpeg -i output?
[16:52] <Daemon404> yeah i never though of that
[16:52] <Daemon404> that's bad
[16:53] <Daemon404> it's a shame the muxing and demuxing is tied like that...
[16:53] <gsuberland> ah well, nm. at least I know it's a bad idea to mess with data so heavily ^^
[16:53] <gsuberland> thanks for the info
[16:53] <beastd> gsuberland: well i talk about using ffmpeg to do what you like, but we should talk in #ffmpeg
[16:53] <cehoyos> Just run "ffmpeg -i", I am very curious.
[16:53] <beastd> oh well, too late...
[16:56] <cone-979> ffmpeg.git 03Benoit Fouet 07master:b35fa041521c: avcodec/pngdec: create a function to decode IDAT chunk.
[16:56] <cone-979> ffmpeg.git 03Benoit Fouet 07master:4f313a50ee78: avcodec/pngdec: create a function to decode PLTE chunk.
[16:57] <Daemon404> cehoyos, i need to wait for vittorio to un-timeout
[16:57] <Daemon404> the user deleted the file, he has it somehwere.
[16:58] <cehoyos> Thank you.
[16:59] <cehoyos> Any chance that you forward such issues to FFmpeg in the future? Anything we can do to make this happen?
[17:01] <cehoyos> Did anybody read the HAP readme? It claims GPU support but I only see CPU encoding and decoding...
[17:02] <Daemon404> i usually give things to michaelni 
[17:02] <Daemon404> most of the weird jpeg fixes have been from my file
[17:02] <Daemon404> s
[17:03] <cehoyos> He has the prores file?
[17:03] <Daemon404> no
[17:04] <Daemon404> not in this instance
[17:04] <Daemon404> just most instances in the past.
[17:04] <Daemon404> i dont forward files without asking the user
[17:04] <Daemon404> unless it is under a proper creative commons license
[17:04] <Daemon404> in which case i can.
[17:06] <cehoyos> Of course not. For many issues the "ffmpeg -i" input might already be sufficient.
[17:07] <Daemon404> most of the issues are more subtle
[17:07] <Daemon404> or output useless stuff
[17:07] <Daemon404> like "pixel format '-1'"
[17:07] <Daemon404> which is my favourite useless message
[17:10] <cone-979> ffmpeg.git 03Benoit Fouet 07master:6499e63f7b9e: avcodec/pngdec: create a function to decode tRNS chunk.
[17:10] <cone-979> ffmpeg.git 03Benoit Fouet 07master:c25b6ae8a2b9: avcodec/pngdec: fix some indentation/whitespaces
[17:19] <rcombs> voting +1 for accepting string args in places that currently take confusing int (enum) args in vf_scale and other ffmpeg.c swscale CLI stuff
[17:19] <cone-979> ffmpeg.git 03Benoit Fouet 07master:24ca2ffad826: avcodec/pngdec: use else if instead of if for small bpp handling.
[17:19] <cone-979> ffmpeg.git 03Benoit Fouet 07master:8cab24df0780: avcodec/pngdec: create a function to handle small (<=4) bits per pixel values.
[17:33] <cone-979> ffmpeg.git 03Benoit Fouet 07master:00df32f6a9c4: avcodec/pngdec: split frame decoding in its own function.
[17:36] <cone-979> ffmpeg.git 03Carl Eugen Hoyos 07master:cde0ad5ea780: tests/Makefile: Fix path for creation of ffprobe-test.nut.
[17:36] <ubitux> benoit-: why not "png_decode_frame" instead of "decode_frame_png" to be consistent with the other function in the file?
[17:36] <rcombs> ^ that also seemed weird to me, looking at the commit message
[17:37] <rcombs> without knowing what other functions are in the file, but just for consistency with a lot of other lav* stuff
[17:37] <wm4> in any case, nice cleanup... why are functions in libav* always so damn large
[17:38] <rcombs> wm4: same reason ass_render_event is so damn large
[17:38] <rcombs> one day, maybe I'll know what that reason is
[17:39] <wm4> rcombs: I have a PR that splits it up
[17:39] <wm4> it was ignored
[17:40] <ubitux> benoit-: also too bad the return code are not forwarded
[17:40] <ubitux> it's most of the time invalid data anyway, but enomem could happen
[18:45] <Luigi12_work> if any ffmpeg developers are here, I was wondering if you could cut a new 2.0.x release with the CVE-2014-527[12] fixes. We need to do a security update for Mageia 4 and it would be much better to have a new release.
[18:48] <Luigi12_work> I noticed that no patches have been committed to the 2.0 branch since 2.0.5, so I guess it would just be adding those CVE fixes
[18:49] <Luigi12_work> I see there's also a CVE-2014-854[1-9], not sure if any of those are relevant to the 2.0 branch
[18:59] <michaelni> Luigi12_work, ok, ill try to make a new 2.0.x
[19:01] <cone-979> ffmpeg.git 03Thilo Borgmann 07master:204533423945: lavd/avfoundation: Introduce device alias 'none' to allow the user to record only audio or video.
[19:01] <cone-979> ffmpeg.git 03Thilo Borgmann 07master:d525e662e4f9: Changelog: Mention AVFoundation screen capturing.
[19:01] <cone-979> ffmpeg.git 03Thilo Borgmann 07master:e6e614963046: doc/indevs: Rework and update documentation of AVFoundation device.
[19:04] <Luigi12_work> michaelni: thanks
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:dd72df9845d0: avformat/utils: do not wait for packets from discarded streams for genpts
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:c2af6b500bcf: avformat: add av_stream_get_parser() to access avformat AVParser
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:2428b02bb4be: ffmpeg: Use av_stream_get_parser() to avoid ABI issues
[19:29] <cone-979> ffmpeg.git 03Anshul Maheswhwari 07release/2.0:3d10235b8352: v4l2enc: adding AVClass
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:8d48223a329a: avcodec/dvdsub_parser: never return 0 when the input isnt 0
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:c7fac44ef83e: avcodec/dvdsub_parser: Check buf_size before reading 32bit packet size
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:65f581472c48: avcodec/dvdsub_parser: print message if packet is smaller than the packet size field
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:bcb10b99f4e8: ffmpeg_opt: Use av_guess_codec() instead of AVOutputFormat->*codec
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:04973b02c3aa: avformat/tee: flip assigment direction
[19:29] <cone-979> ffmpeg.git 03Anton Khirnov 07release/2.0:c56d6f3552a0: cdgraphics: do not return 0 from the decode function
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:656f930160db: avcodec/iff: check pixfmt for rgb8 / rgbn
[19:29] <cone-979> ffmpeg.git 03Christophe Gisquet 07release/2.0:b469fce85d8a: proresenc_kostya: remove unneeded parameters
[19:29] <cone-979> ffmpeg.git 03Christophe Gisquet 07release/2.0:57a6cd8ab1c2: proresenc_kostya: report buffer overflow
[19:29] <cone-979> ffmpeg.git 03Christophe Gisquet 07release/2.0:caf08defa69f: proresenc_kostya: properly account for alpha
[19:29] <cone-979> ffmpeg.git 03Christophe Gisquet 07release/2.0:2958b8b86e36: wavpack: report if there is no bits left
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:9fc7de8d8030: avcodec: fix aac/ac3 parser bitstream buffer size
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:d4253b3a5b74: avcodec/utils: add GBRP16 to avcodec_align_dimensions2()
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:d76c9b5665a8: avcodec/snow: check coeffs for validity
[19:29] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:d7d29f0d433a: avformat/swfdec: Use side data to communicate w/h changes to the decoder
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:2fdb02693bbd: avformat/swfdec: Do not change the pixel format
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:6e70816f7191: avcodec/mpegvideo: Use "goto fail" for all error paths in ff_mpv_common_frame_size_change()
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:2b415192fd9a: avcodec/mpegvideo: check that the context is initialized in ff_mpv_common_frame_size_change()
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:beb83f0a4070: avcodec/mpegvideo: Set err on failure in ff_mpv_common_frame_size_change()
[19:30] <cone-979> ffmpeg.git 03Katerina Barone-Adesi 07release/2.0:47fe68eec821: apetag: Fix APE tag size check
[19:30] <cone-979> ffmpeg.git 03James Almer 07release/2.0:992ce9777c7e: x86/dsputil: add emms to ff_scalarproduct_int16_mmxext()
[19:30] <cone-979> ffmpeg.git 03Gianluigi Tiesi 07release/2.0:f238de199053: avcodec/libilbc: support for latest git of libilbc
[19:30] <cone-979> ffmpeg.git 03lvqcl 07release/2.0:2eda0e705a28: avutil/x86/cpu: fix cpuid sub-leaf selection
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:52fd0cda2cdd: avcodec/ac3enc_template: fix out of array read
[19:30] <cone-979> ffmpeg.git 03Reimar Döffinger 07release/2.0:8622618839c6: configure: add noexecstack to linker options if supported.
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:3f9a148022ef: avcodec/jpeglsdec: Check run value more completely in ls_decode_line()
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:49d69844f500: avcodec/mjpegdec: check bits per pixel for changes similar to dimensions
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:ac82e318bb81: avcodec/utils: Add case for jv to avcodec_align_dimensions2()
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:16775c7aaa07: avcodec/mmvideo: Bounds check 2nd line of HHV Intra blocks
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:abbfc4d87ec4: avcodec/tiff: more completely check bpp/bppcount
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:ae81d9a7da10: avcodec/pngdec: Check bits per pixel before setting monoblack pixel format
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:db48767d7c16: avcodec/pngdec: Calculate MPNG bytewidth more defensively
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:97fdbd12f909: avcodec/cinepak: fix integer underflow
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:be7105dff61e: avcodec/gifdec: factorize interleave end handling out
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:d36bb362a673: avcodec/qpeg: fix off by 1 error in MV bounds check
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:b4e0acfa043c: avcodec/smc: fix off by 1 error
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:ff59edb6dcd7: avcodec/svq3: Dont memcpy AVFrame
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:128b0510e159: avformat/mpegts: Check desc_len / get8() return code
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:0e7173826216: avcodec/h264: Check mode before considering mixed mode intra prediction
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:322470e60633: swresample/swresample: fix sample drop loop end condition
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:d7906baa8597: postproc/postprocess: fix quant store for fq mode
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:9b1673531cf9: postproc: fix qp count
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:70402f6ee786: avcodec/diracdec: Use 64bit in calculation of codeblock coordinates
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:c0be9d726440: avcodec/diracdec: Tighter checks on CODEBLOCKS_X/Y
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:2cf83677b314: avcodec/dirac_arith: fix integer overflow
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:808b0ccc03dc: avcodec/dxa: check dimensions
[19:30] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:85cac770bdb0: avcodec/dnxhddec: treat pix_fmt like width/height
[19:31] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:bde9e859b3c7: avcodec/utils: Align dimensions by at least their chroma sub-sampling factors.
[19:31] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:e2865d931618: avcodec/g2meet: check tile dimensions to avoid integer overflow
[19:31] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:8efb06c8735f: avcodec/cook: check that the subpacket sizes fit in block_align
[19:31] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:694c3dab363f: avcodec/svq1dec: zero terminate embedded message before printing
[19:31] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:0140f11c3bad: avcodec/h264_slice: Clear table pointers to avoid stale pointers
[19:31] <cone-979> ffmpeg.git 03Carl Eugen Hoyos 07release/2.0:2389309d4836: lavc/utils: Make pix_fmt desc pointer const.
[19:31] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:05e5d785fa6a: avcodec/options_table fix min of audio channels and sample rate
[19:31] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:f8675743c40a: avcodec/utvideodec: fix assumtation that slice_height >= 1
[19:31] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:0baeb59307e6: avcodec/wmaprodec: Fix integer overflow in sfb_offsets initialization
[19:31] <J_Darnley> ugh, bloody release spam
[20:01] <cone-979> ffmpeg.git 03Michael Niedermayer 07release/2.0:3d91569c5e39: update for 2.0.6
[20:36] <BadClown> J_Darnley: about those EXRs... tiled EXR are generated by most rendering software, and they are not properly handled by ffmpeg
[20:37] <BadClown> i can provide example files if anyone is interested
[20:39] <J_Darnley> Then I suggest you create a new issue on the bug tracker.
[20:43] <BadClown> ok. I will. thx
[20:47] <llogan> some oddball effects: http://effectv.sourceforge.net/colstreak.html
[20:47] <llogan> maybe some could be ported as filters
[20:48] <wm4> yeah, oddball effects is totally what we need </nag>
[20:53] <llogan> at least they're in C
[23:39] <cone-979> ffmpeg.git 03Thilo Borgmann 07fatal: ambiguous argument 'refs/tags/n2.0.6': unknown revision or path not in the working tree.
[23:39] <cone-979> Use '--' to separate paths from revisions
[23:39] <cone-979> refs/tags/n2.0.6:HEAD: doc/indevs: Rework and update documentation of AVFoundation device.
[23:56] <Luigi12_work> michaelni: thanks again for the 2.0.6 release.  I'm guessing it contains fixes for CVE-2014-527[12] and CVE-2014-854[1-8]
[23:58] <michaelni> yes, see http://ffmpeg.org/security.html
[23:59] <Luigi12_work> michaelni: ahh I had to shift-reload so it didn't reload from cache :o)
[00:00] --- Sat Nov 15 2014