[Ffmpeg-devel-irc] ffmpeg-devel.log.20170711

[burek]

[00:28:09 CEST] <michaelni> durandal_1707, if you want distinct vlc for 0 occuring symbos they can be set to the lowest probability (1). If the vlc doesnt matter as they do not occur, then anything could be use das their lengt and bits
[00:52:42 CEST] <cone-641> ffmpeg 03Rostislav Pehlivanov 07master:ba67c2349c15: opus_celt: move postfilter taps table to the shared opustab.c file
[00:52:42 CEST] <cone-641> ffmpeg 03Rostislav Pehlivanov 07master:03d8fbc09cdb: opustab: comment ff_celt_tf_select
[00:52:42 CEST] <cone-641> ffmpeg 03Rostislav Pehlivanov 07master:8041b2420f11: opusenc: don't set avctx->cutoff
[00:52:42 CEST] <cone-641> ffmpeg 03Rostislav Pehlivanov 07master:e7d977b44619: opus_rc: fix encoder desyncs on very low bitrates
[00:53:10 CEST] <atomnuker> this took nearly 30 seconds to push, something's up with the server
[00:54:54 CEST] <iive> the videolan one?
[00:55:07 CEST] <atomnuker> yep
[00:56:04 CEST] <nevcairiel> even pulling is slow for me these days, been that way for a couple months now
[08:50:28 CEST] <cone-176> ffmpeg 03Muhammad Faiz 07master:0780ad9c688c: avcodec/rdft: remove sintable
[10:06:29 CEST] <scriptso> I was wondering if anyone would be able to help me understand these to Snippets... By no means do I hold or claim that I personally wrote the entirety of these scipts but bs but I did write the majority of one of them at least... You can tell by  how ugly it is | ffstreaming screen script https://pastebin.com/URLcbBgn, ffstreamit  https://pastebin.com/6mBVYK6u|
[10:08:59 CEST] <scriptso> The idea here is that I want to be able to cast from  a a boom 2 machine to a Windows machine... I did not use that client script on the Windows machine.... yeah...,  for some reason I thought VLC would handle it??    I have gone over ffmpeg entire Docs but what I'm confused on is on the client side for Windows,  exactly how do I go about setting the right settings?? I am very very new to ffmpeg
[10:32:07 CEST] <scriptso> WOWOWOWWW! not all at the same time!
[10:32:08 CEST] <scriptso> lol
[10:35:10 CEST] <wm4> scriptso: wrong channel
[10:35:27 CEST] <scriptso> ..... oh no imsosorry 
[10:35:28 CEST] <scriptso> wait
[10:35:35 CEST] <scriptso> ohh durp
[10:35:38 CEST] <scriptso> im sorry
[10:36:34 CEST] <scriptso> this is for.... the actual dev cycle of the application?
[10:36:42 CEST] <scriptso> im such a noob
[13:05:48 CEST] <J_Darnley> kierank, atomnuker: I might have just had an epiphany
[13:06:21 CEST] <J_Darnley> I think I see the deinterleave in encode_subband
[13:10:51 CEST] <J_Darnley> Yes, in the original the subbands are read out of the whole plane transform with an offset from the start of the subband based on the slice position
[13:14:19 CEST] <J_Darnley> Now comes the tricky part of re-engineering all that.
[13:16:03 CEST] <kierank> good!
[14:52:14 CEST] <J_Darnley> I hate these structures.  Too much is precalculated or hidden away in them.
[14:52:26 CEST] <J_Darnley> Subbands are based on the whole plane transform
[14:52:41 CEST] <J_Darnley> the dimensions are too large
[14:53:02 CEST] <J_Darnley> there are at 3 strides in use
[14:53:07 CEST] <J_Darnley> *at least
[14:54:22 CEST] <BBB> its usually for performance reasons
[14:54:33 CEST] <BBB> so sadly, yes, readability and understandibility goes down as performance goes up
[14:59:55 CEST] <J_Darnley> Do I really have to rewrite the quantisation as well?
[15:02:12 CEST] <J_Darnley> Is there no part of this encoder that isn't tightly coupled to every other part?
[15:03:01 CEST] <J_Darnley> The dwt is about the least coupled thing but even that takes a struct pointer
[15:10:59 CEST] <J_Darnley> As a concete example: why does a SubBand need its own stride field.  It is a "constant".  It is the same as the Plane coefficient stride so just use that!
[15:30:54 CEST] <J_Darnley> God fucking dammit!  I can't abort a git stash pop
[15:50:20 CEST] <BBB> now now now
[15:52:47 CEST] <J_Darnley> Yeah, I shouldn't swear in public channels
[15:53:58 CEST] <uau> what do you mean by "aborting" it?
[15:54:13 CEST] <J_Darnley> Like marge --abort
[15:54:19 CEST] <J_Darnley> cherry-pick --abort
[15:54:22 CEST] <J_Darnley> rebase --abort
[15:54:47 CEST] <iive> J_Darnley:  don't worry, it's perfectly normal :)
[16:05:44 CEST] <J_Darnley> atomnuker: In dirac are the chroma slices with yuv420p (or any other subsampling) smaller than the luma slices?
[16:06:50 CEST] <atomnuker> yes
[17:24:37 CEST] <kierank> J_Darnley: make: *** No rule to make target `libavcodec/x86/simple_idct.c', needed by `libavcodec/x86/simple_idct.o'. Stop.
[17:24:40 CEST] <kierank> why do i get that
[17:24:49 CEST] <nevcairiel> make clean
[17:24:54 CEST] <kierank> I did
[17:25:08 CEST] <kierank> let's try again
[17:25:14 CEST] <nevcairiel> you get that because simple_idct.c doesnt exist anymore, its now simple_idct.asm =p
[17:25:31 CEST] <nevcairiel> (at worst, make distclean)
[17:45:48 CEST] <ubitux> kierank: rm libavcodec/x86/simple_idct.d
[17:45:58 CEST] <kierank> builds after a make clean
[17:45:58 CEST] <kierank> thanks
[18:17:20 CEST] <durandal_1707> peloverde: i still get artifacts with that hunk applied
[18:17:44 CEST] <peloverde> which stream?
[18:18:20 CEST] <durandal_1707> aac_isobmff_i60.m4a
[18:27:18 CEST] <cone-029> ffmpeg 03wm4 07master:02d248d5828d: videotoolbox: fix crash when decoding interlaced video with new API
[18:27:57 CEST] <durandal_1707> golomb uses special bitreader, it have own header, how to convert it to optional cached variant?
[18:50:13 CEST] <peloverde> durandal_1707: I hear it, probably another windowing issue, let me take a closer look
[18:50:28 CEST] <tdjones> I'm getting this odd reverb with the vorbis encoder on short frames. I set the encoder to use all transient frames to exaggerate it: https://0x0.st/GSo.ogg . It's not evident to me why this would occur
[18:54:02 CEST] <atomnuker> tdjones: seems its only on the left channel
[18:57:08 CEST] <tdjones> atomnuker: I didn't even notice that, it's what I get for using crappy headphones
[18:57:16 CEST] <tdjones> atomnuker: Thank you
[19:24:54 CEST] <peloverde> durandal_1707: This should work better https://github.com/aconverse/FFmpeg/commits/aac960
[19:25:40 CEST] <peloverde> I wasn't turning off the default SBR parameters properly
[21:03:40 CEST] <durandal_1707> some devs had serious lobotomy
[21:26:19 CEST] <amey> Hello all developers. I am using ffmpeg API to build an audio fingerprinting system for GSoC. In first phase it requires pre-processing. I have used ffmpeg filters to resample the audio. It is running fine with some video but it is dumping some error with others. I think the problem is with some file format or codec. Apart from that also if you find any mistakes tell me. 
[21:26:19 CEST] <amey> Link to project repository: https://github.com/Amey-Jain/audiofingerprinting
[21:26:19 CEST] <amey> Error output: https://pastebin.com/sBVz4ZXR
[21:56:47 CEST] <J_Darnley> What value should an encoder return when the supplied options are wrong?  AVERROR(EINVAL)?
[21:57:08 CEST] <nevcairiel> EINVAL for configuration error, yes
[21:57:25 CEST] <J_Darnley> I thought so.
[21:57:34 CEST] <nevcairiel> AVERROR_INVALIDDATA for bitstream errors, although perhaps not applicable to encoders
[21:57:48 CEST] <nevcairiel> (those two often get mixed up)
[21:57:59 CEST] <J_Darnley> nah, this would be command line options for ffmpeg
[21:59:23 CEST] <jamrial> J_Darnley: for invalid option instead of invalid argument for a valid option, there seems to be AVERROR_OPTION_NOT_FOUND
[21:59:54 CEST] <nevcairiel> that would appear to only be needed in AVOption code
[22:00:05 CEST] <wm4> (does anything really try to make sense out of these error codes)
[22:00:19 CEST] <nevcairiel> they get converted to strings for CLI output sometimes
[22:00:36 CEST] <wm4> yes, which is even more wronger and awkward
[22:00:39 CEST] <nevcairiel> but sticking to some sense of convention is always a good thing
[22:01:00 CEST] <nevcairiel> I just print their numeric values in my logs and then always wonder wtf this one is
[22:01:25 CEST] <wm4> yeah, and I start trying to look up raw errno values on various platforms/OSes
[22:01:33 CEST] <wm4> (on Linux they can be CPU dependent)
[22:01:45 CEST] <J_Darnley> Does anyone really need anything more than a boolean?
[22:01:49 CEST] <wm4> I always found it crazy that ffmpeg "wraps" errnos
[22:02:28 CEST] <wm4> and that it tries to handle platforms that use negative errnos, which AFAIK don't really exist
[22:02:29 CEST] <jamrial> INVALIDDATA, PATCHWELCOME and EOF are used everywhere and they are not wrapped errno values
[22:02:53 CEST] <J_Darnley> No but then those are Invented Here codes
[22:02:59 CEST] <jamrial> the others are much less common, like BUG, UNKNOWN and $moduletype_NOT_FOUND
[22:04:10 CEST] <wm4> supposedly negative errnos can happen on BeOS?
[22:04:24 CEST] <wm4> 8fa36ae09dddb1b639b4df5d505c0dbcf4e916e4
[22:04:44 CEST] <wm4> also calling "-ESOMETHING" over "AVERROR(ESOMETHING)" ugly
[22:16:20 CEST] <rcombs> the worst thing is when people return -1 from a function that is expected to return an AVERROR
[22:16:33 CEST] <rcombs> (-1 is AVERROR(EACCES))
[22:38:05 CEST] <cone-119> ffmpeg 03Rostislav Pehlivanov 07master:aef5f9ab05c3: mdct15: remove redundant scale argument to imdct_half
[22:38:05 CEST] <cone-119> ffmpeg 03Rostislav Pehlivanov 07master:594cd1f38a7b: opus_celt: normalize using mdct scale
[00:00:00 CEST] --- Wed Jul 12 2017