[Ffmpeg-devel-irc] ffmpeg-devel.log.20141208

[burek]

[00:22] <ubitux> http://silveiraneto.net/2014/12/07/imagemagick-four-point-perspective-distortion-in-a-video/
[00:22] <ubitux> don't we have a perspective filter for that exact purpose?
[00:22] <ubitux> i see the guy is using ffmpeg at the end, i'm wondering if we can't actually do it just with it
[00:24] <rcombs> "Ps: this step is not really necessary as you could use png as input to ffmpeg." <-- then why are you showing it
[00:24] <ubitux> :D
[00:25] <ubitux> if quvi support still works we could do all of this in one ffmpeg command @_@
[00:25] <ubitux> or like with ffplay directly...
[00:25] <ubitux> maybe i'm going to try this
[00:26] <rcombs> if I actually wanted to do that I'd pull up Final Cut though
[00:34] <ubitux> alright it does the trick
[00:37] <ubitux> -vf perspective=60:90:589:147:50:415:582:418
[00:39] <rcombs> there are 2, maybe 3 good ways to do that with ffmpeg, and "export as a PNG sequence, apply transforms in imagemagick, convert to JPEG sequence for no reason at all, then convert back to H.264 with ffmpeg" is not one of them
[00:41] <ubitux> :Ddddd
[00:41] <ubitux> oups
[00:42] <ubitux> anyway, now i'm working on the timeline editing to make it relevant
[00:43] <cehoyos> ubitux: Did you have a look at the uspp patch? Is it mosltly ok?
[00:43] <ubitux> cehoyos: there are a bunch of things to do remaining
[00:43] <ubitux> are you in a hurry to apply it?
[00:44] <cehoyos> Absolutely not!
[00:44] <cehoyos> I just meant: Is she on the right way?
[00:44] <cehoyos> (I did not look)
[00:45] <ubitux> i don't see the doc, nor the remove from mp, i also see a lot of redundancy with other filters (typically the filters i ported)
[00:45] <ubitux> i need to check if the inplace code is actually properly done
[00:46] <ubitux> and if there is no leak
[00:46] <ubitux> that kind of stuff
[00:46] <ubitux> didn't find time yet
[00:46] <ubitux> as you can see i'm busy with way more important stuff
[00:46] <ubitux> like using a perspective filter on a random youtube video
[00:48] <ubitux> -vf "perspective=60:90:589:147:50:415:582:418:enable='not(between(t,0,5)+between(t,50,68))'  need to be more accurate but well
[00:49] <ubitux> oh some people game the frame ids
[00:56] <cehoyos> Thank you for the update!
[00:57] <ubitux> ffmpeg -i zelda.mp4 -vf "perspective=60:90:589:147:50:415:582:418:enable='not(between(n,0,149)+between(n,1488,2063))'" -c:a copy -c:v libx264 zelda.mkv
[00:57] <ubitux> ffmpeg iz magic
[01:06] <ubitux> i should have adjusted the coordinates relative to the height and width
[01:06] <ubitux> so the filter could work with any resolution
[01:12] <cone-37> ffmpeg.git 03Carl Eugen Hoyos 07master:7502e1a3b743: Fix xcb 32bit pix_fmts: There is no screen transparency.
[01:12] <cone-37> ffmpeg.git 03Carl Eugen Hoyos 07master:d9a4db657dce: Improve configure help text.
[01:47] <cone-37> ffmpeg.git 038E08; C:>2=8:>2 07master:71f29410e7c5: lavfi: Increase drawtext font size limits
[01:47] <cone-37> ffmpeg.git 03Luca Barbato 07master:4f5906a1d7f4: latm: Do not give a score for a single instance
[01:48] <cone-37> ffmpeg.git 03Michael Niedermayer 07master:9bbb4a98fe53: Merge commit '71f29410e7c58f2ce4aef1d4b8d8d3c064c8ec06'
[01:48] <cone-37> ffmpeg.git 03Michael Niedermayer 07master:dce10f817834: Merge commit '4f5906a1d7f4368cec43b224e8a675f54d2001d2'
[02:19] <cone-37> ffmpeg.git 03Luca Barbato 07master:4fd7e63c8722: mp3: Tweak the probe scores
[02:19] <cone-37> ffmpeg.git 03Michael Niedermayer 07master:ef79bbeddd43: Merge commit '4fd7e63c87226584e60200b82630a6057c0a6663'
[02:41] <cone-37> ffmpeg.git 03Luca Barbato 07master:6a880090cd75: doc: Use the correct @subsection tag
[02:41] <cone-37> ffmpeg.git 03Michael Niedermayer 07master:208c03a4d8bc: Merge commit '6a880090cd75b6c31b5ee1f088b6578a293099c6'
[04:57] <cone-37> ffmpeg.git 03Rens Dijkshoorn 07master:dceb8b12525c: avformat/movenc: Add 2048x1080 to the aivx case
[05:12] <cone-37> ffmpeg.git 03Benoit Fouet 07master:4fa2758c799a: avcodec/pngdec: check av_image_get_linesize return value
[10:12] <j-b> 'lo
[11:03] <anshul_mahe> any suggestion for making tag in nut format for scte-35 message cue stream
[12:14] <cone-600> ffmpeg.git 03Carl Eugen Hoyos 07master:3f24dacf810e: configure: Let the libx264rgb encoder select the libx264 encoder.
[12:24] <cone-600> ffmpeg.git 03Martin Storsjö 07master:29bc7bfba288: rtpproto: Write a warning if the input data written isn't RTP packetized
[12:24] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:5ab467043a2b: Merge commit '29bc7bfba288ff8572ed967a8752a1dbde7b724b'
[12:31] <cone-600> ffmpeg.git 03Martin Storsjö 07master:ae81576414f2: arm: fft_vfp: Add a missing "endconst" when building in thumb mode
[12:31] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:8554a10992de: Merge commit 'ae81576414f2d2083d3118fb4abe1ebc5a7a4c54'
[12:43] <cone-600> ffmpeg.git 03Martin Storsjö 07master:b280c6202b28: arm: fft_vfp: Unify the behaviour in ff_fft_calc_vfp between arm/thumb
[12:43] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:5054d2a92ba7: Merge commit 'b280c6202b28b371a8d96850194fd69d7ad5dcc0'
[15:29] <cone-600> ffmpeg.git 03Rong Yan 07master:9bd8f2cc32c2: avcodec/ppc/vp3dsp_altivec: POWER LE support to vp3_idct_add_altivec()
[15:47] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:41a52740d47a: avutil/opt: Fix AV_OPT_TYPE_BINARY case in av_opt_is_set_to_default()
[18:23] <kierank> why does urspp need avcodec?
[18:23] <kierank> does it use the qps?
[18:24] <wm4> it uses SNOW
[18:24] <wm4> holy shit
[18:25] <kierank> wh at the fu ck
[18:26] <wm4> I'm so sorry
[18:27] <ubitux> well it's a deblocking filter using wavelets
[18:28] <ubitux> i guess that's interesting?
[18:28] <Compn> anyone ever compare uspp to synth deblocking filters?  
[18:28] <Compn> just curious the amount of bitching about uspp lol
[18:28] <ubitux> it's supposed to be more efficient than most of pp filters (but slow)
[18:28] <ubitux> Compn: snow codec is supposed to be taboo
[18:34] <Compn> for some forks maybe :P
[18:34] <Compn> if its huge problem just duplicate the code 
[18:41] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:ed9be7dd470c: avcodec/x86/pngdsp: fix off by 1 error
[18:41] <cone-600> ffmpeg.git 03Michael Niedermayer 07master:82a60de3b78f: avcodec/pngdec: handle 4 more bytes with SIMD for RGB64 in paeth prediction
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:9e65065080f5: avformat/rmdec: Check codec_data_size
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:223ae2467e3a: swscale/x86/rgb2rgb_template: fix crash with tiny size and nv12 output
[19:26] <cone-600> ffmpeg.git 03wm4 07release/2.2:9a02be31225b: avformat/matroskadec: fix handling of recursive SeekHead elements
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:8b8d794800db: avcodec/h264_slice: Clear table pointers to avoid stale pointers
[19:26] <cone-600> ffmpeg.git 03Carl Eugen Hoyos 07release/2.2:cd01611d7b7a: lavc/utils: Make pix_fmt desc pointer const.
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:84fdfcab99aa: avcodec/options_table fix min of audio channels and sample rate
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:4d8b82160dbc: avcodec/utvideodec: fix assumtation that slice_height >= 1
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:d054ec868d96: avcodec/wmaprodec: Fix integer overflow in sfb_offsets initialization
[19:26] <cone-600> ffmpeg.git 03Lukasz Marek 07release/2.2:944007921667: lavu/opt: fix av_opt_get function
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:217f781adca3: avformat/hlsenc: Free context after hls_append_segment
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:de7671e4c47f: swscale/x86/rgb2rgb_template: handle the first 2 lines with C in rgb24toyv12_*()
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:5d6f8bab02ba: avcodec/mjpegdec: Fix context fields becoming inconsistent
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:a63941eec2f9: avcodec/utils: Check that the data is complete in avpriv_bprint_to_extradata()
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:d41010b89575: avcodec/flacdec: Call ff_flacdsp_init() unconditionally
[19:26] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:d7470271c7ca: avcodec/pngdec: Check IHDR/IDAT order
[19:27] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:a06432b6c315: avcodec/rawdec: Check the return code of avpicture_get_size()
[19:27] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:4fde30ba9d05: avcodec/hevc_ps: Check num_long_term_ref_pics_sps
[19:27] <cone-600> ffmpeg.git 03Michael Stypa 07release/2.2:3d9860f68af5: fix Makefile objects for pulseaudio support
[19:27] <cone-600> ffmpeg.git 03Michael Niedermayer 07release/2.2:7390c2629dc2: avcodec/motion_est: use 2x8x8 for interlaced qpel
[20:19] <llogan> jamrial: this shows 2.5: https://packages.debian.org/sid/ffmpeg
[20:24] <jamrial> ah, so Andreas updated the package
[20:24] <jamrial> last time i checked it was still 2.4
[20:24] <jamrial> ok, nevermind then
[20:28] <ubitux> what about ubuntu?
[20:29] <llogan> 2.5
[20:29] <llogan> according to http://packages.ubuntu.com/vivid/ffmpeg
[20:38] <llogan> curses, whose editing Downstream?
[20:43] <jamrial> wouldn't "generally latest" and "rolling release" apply to Debian sid like it does to archlinux?
[20:45] <llogan> jamrial: sounds good to me. i included it in:
[20:46] <cone-902> ffmpeg.git 03Carl Eugen Hoyos 07master:250e077ee9ee: riff: Support decoding png in avi (ImageJ)
[20:46] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:3cfe1d9b3e65: Merge commit '250e077ee9eec0176a6d54a78542dc792943e71a'
[21:00] <cone-902> ffmpeg.git 03Zoltan Kovacs 07master:4843a9c74f77: dpx: always write the image offset
[21:01] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:c9128c2c8318: Merge commit '4843a9c74f77f5eb8d2b1ee648fe2ac6e2a48ec3'
[21:13] <cone-902> ffmpeg.git 03Joakim Plate 07master:90411b333d6b: Allow reading of growing avi files (currently being written)
[21:13] <cone-902> ffmpeg.git 03Vittorio Giovara 07master:d0b224054f13: vf_frei0r: do not increment string if it reached the end
[21:13] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:879ea06bc557: Merge commit '90411b333d6b6bb69d8937117b9250785730dc53'
[21:13] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:6b4cbfec757a: Merge commit 'd0b224054f13bf57244694a3ff092cfef68d66f9'
[21:24] <llogan> jamrial: i wonder why none of the BPG related port-able stuff was submitted upstream
[21:25] <koda> llogan: yeah submitting upstream is hard
[21:26] <llogan> koda: perhaps he's not familair with the project
[21:27] <jamrial> heh
[21:30] <jamrial> the actual bpg parsing code on that library looks a lot like an actual lavc codec, for that matter, so he could have gone all the way and submitted a native decoder himself without much extra work
[21:31] <koda> these people forking should know theyll be assimilated
[21:32] <koda> imho in bpg the only interesting thing is monochrome
[21:32] <koda> the rest will have so little adotion noone will care
[21:40] <jamrial> it seems to have better overall compression than webp
[21:42] <kierank> koda: he's got no reason to upstream
[21:42] <ubitux> he's the upstream
[21:43] <gnafu> I don't think The Man Himself is so concerned with what other people do or don't do with his code.
[21:47] <koda> kierank: keeping all changes within the project is not a good way to do Open Source
[21:47] <koda> even worse, no git history, bleah
[21:48] <kierank> neither is becoming the kitchen sink of all multimedia
[21:48] <kierank> if he wants to work on something by himself let him do so
[21:49] <cone-902> ffmpeg.git 03Vittorio Giovara 07master:6fd91fa11909: h264: increase MAX_SLICES to 32
[21:49] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:74622ad320bb: Merge commit '6fd91fa11909f27902498648680dbb3d13f1f175'
[21:50] <koda> kierank: if only every project had this physophy&
[21:50] <kierank> libav was the same in the past
[21:50] <kierank> emailing ffmpeg contributors too
[21:50] <kierank> so don't go and whitewash history
[21:52] <koda> kierank: ??? i wasnt whitewashing anything, actually i wasnt talking bout libav at all
[21:53] <kierank> ok
[21:53] <koda> what _ever_ gave you the impression i was trolling? geez
[21:53] <kierank> 8:50 PM <koda> kierank: if only every project had this physophy& --> so waht are you referring to then?
[21:53] <koda> ffmpeg of course, about not becoming the kitchen sink ;)
[21:54] <kierank> as i will say at fosdem these many to many projects don't work
[21:54] <kierank> you can do most things
[21:54] <kierank> but certain things just won't work ever
[21:54] <koda> yep there must be a balance
[21:57] <Compn> you mean like gstreamer ?
[21:57] <Compn> as an example of a project that doesnt work, i meant
[21:57] Action: Compn runs
[21:58] <kierank> well certain things will never work in ffmpeg
[21:58] <kierank> like ts muxing
[21:59] <kierank> because you can't take arbitrary crap and make a valid ts out of it
[22:02] <kierank> bah vf_scale doesn't take field order into account
[22:05] <cone-902> ffmpeg.git 03Vittorio Giovara 07master:35384934d6e2: mov: cosmetics: reorder the list of tags
[22:05] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:82f0f0482ed6: Merge commit '35384934d6e27e0334060a23a0c83a3cb5cef198'
[22:32] <cone-902> ffmpeg.git 03Vittorio Giovara 07master:174c5fde9006: mov: parse @PRM and @PRQ metadata tags
[22:32] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:420aebf0c585: Merge commit '174c5fde90060faab67796a5eaef742630f1db6e'
[23:00] <cone-902> ffmpeg.git 03Vittorio Giovara 07master:b704b648f9ec: mov: parse XMP metadata on demand
[23:00] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:0dcebe26209f: Merge commit 'b704b648f9ecb830874627db958a37e004107d1b'
[23:08] <cbsrobot_> michaelni: I suggest to disable export_xmp by default
[23:10] <cbsrobot_> I received a file lately with not less then 300Mb of xmp metadata and I was waiting minutes just to see what codec the file was ...
[23:10] <ubitux> aren't we already annoyed at random xml metadata exported?
[23:10] <ubitux> i remember random xml dumps in the ffmpeg output
[23:11] <cbsrobot_> that could have been the xmp stuff
[23:12] <cbsrobot_> ubitux: do you agree to disable it by default ?
[23:13] <ubitux> i'm saying i'm already annoyed at what is currently exporting, so i'd say yes
[23:13] <ubitux> also, aren't some format remuxing them under similar keys?
[23:13] <ubitux> like, what happens when you remux to mkv?
[23:13] <cbsrobot_> maybe it would be useful to print at least the size of the xmp metadata 
[23:14] <cbsrobot_> because if you want to convert to a h264 for internet you really want to get rid of them
[23:15] <cbsrobot_> I guess the xmp data will be kept when converting to mkv, but I'm not sure
[23:16] <cone-902> ffmpeg.git 03Lukasz Marek 07master:72732307d9f2: lavc/options: add test for avcodec_copy_context
[23:16] <cone-902> ffmpeg.git 03Lukasz Marek 07master:4acc717a05f2: fate: add libavcodec/options-test
[23:20] <michaelni> kierank, you mean vf_scale doesnt forward the top_field_first flag or some other issue ?
[23:21] <kierank> if I were to add chroma positioning vf_scale would need to setup swscale with the correct chroma positioning. I think it could just choose the right swscale context based off top_field_first though
[23:22] <cone-902> ffmpeg.git 03Vittorio Giovara 07master:e2e07dbaab1f: mov: expand the list of parsed metadata tags
[23:22] <cone-902> ffmpeg.git 03Michael Niedermayer 07master:885e74622e3f: Merge commit 'e2e07dbaab1f17e995842b7eacec4665d44c3d14'
[23:23] <kierank> michaelni: however please don't get any ideas and add interlaced chroma to swscale
[00:00] --- Tue Dec  9 2014