[Ffmpeg-devel-irc] ffmpeg-devel.log.20150610

burek burek021 at gmail.com
Thu Jun 11 02:05:04 CEST 2015


[00:00:47 CEST] <anoop_r> how to fork ffmpeg ?
[00:02:44 CEST] <Compn> github ffmpeg > click fork
[00:03:59 CEST] <philipl> There's a libav joke in there somewhere.
[00:04:42 CEST] <iive> well, they can't help you with the fork, because they butchered it.
[00:07:50 CEST] <anoop_r> can rename ffmpeg
[00:08:00 CEST] <anoop_r> can I rename ffmpeg
[00:08:06 CEST] <llogan> yes, but convention dictates that the new name should already be associated or be use by the original. so i recommend "ffmpeg_g" or "libavcodec".
[00:09:06 CEST] <anoop_r> does handbrek use ffmpeg
[00:09:19 CEST] <philipl> Handbrake uses libav
[00:09:43 CEST] <philipl> So, hey what do you know? You're already using a fork! Bonus.
[00:10:19 CEST] <anoop_r> no i am using ffmpeg but i woud like to change it to OT.exe
[00:10:50 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07master:58142a27ea96: MAINTAINERS: add 2.7
[00:11:52 CEST] <philipl> If you literally asking if you can rename ffmpeg.exe to OT.exe, obviously you can. But I'm scared to hear what you expect that to achieve.
[00:12:17 CEST] <anoop_r> no i want to reflect the changes in help doc too
[00:12:49 CEST] <anoop_r> means OT -h should give OT fork of ffmpeg like
[00:14:11 CEST] <philipl> s/ffmpeg/OT/g
[00:15:27 CEST] <anoop_r> no ffmpeg version 2.6.git Copyright (c) 2000-2015 the FFmpeg developers
[00:15:58 CEST] <anoop_r> i want it to be like OT forked from ffmpeg
[00:16:20 CEST] <philipl> If you're looking for ways to use ffmpeg without giving any credit to the ffmpeg developers, then you definitely need to go ask on #libav-devel. They're experts on that.
[00:17:02 CEST] <anoop_r> no if i want to do that i don't have to mention OT is a for of ffmpeg
[00:17:12 CEST] <anoop_r> fork of ffmpeg
[00:17:21 CEST] <llogan> it's probably in cmdutils.c
[00:17:46 CEST] <anoop_r> there is no easy way
[00:18:10 CEST] <llogan> 1. edit file. 2. done
[00:18:24 CEST] <anoop_r> going thorug source is very hard
[00:18:32 CEST] <llogan> pay someone
[00:18:45 CEST] <anoop_r>  i not rich
[00:20:13 CEST] <durandal_1707> one fork less
[00:21:01 CEST] <anoop_r> 2272 lines
[00:21:19 CEST] <anoop_r> not a possible task for me
[00:22:56 CEST] <iive> I bet most of these are copyright notice.
[00:23:21 CEST] <iive> why not just rename the executable?
[00:23:40 CEST] <philipl> Maybe the libav guys could give you the script they used.
[00:24:25 CEST] <anoop_r> i think they didn't go for that. they may be riped it apart
[00:26:58 CEST] <anoop_r> well i am doing an acdemic project and if i rename it and present my evaluaters will say you had just copied ffmpeg 
[00:27:24 CEST] <anoop_r> even if i had compiled and added som extralibs they won't count it as a work
[00:27:51 CEST] <nevcairiel> so, instead you try to cheat them by renaming it, instead of actually working on something yourself?
[00:28:27 CEST] <anoop_r> but i not that bright
[00:28:54 CEST] <philipl> Is this conversation actually happening?
[00:29:15 CEST] <anoop_r> also it not possible to right an entire mux/demux lib in a month
[00:29:31 CEST] <anoop_r> i had made a gui for ffmpeg in c#
[00:29:33 CEST] <anoop_r> works fine
[00:29:44 CEST] <anoop_r> now i want to fork ffmpeg
[00:30:07 CEST] <anoop_r> well i not cheating them because forked open source are acceptable
[00:30:17 CEST] <anoop_r> as per their standards
[00:31:41 CEST] <philipl> If it's acceptable, then just use it. No need to rename anything.
[00:32:32 CEST] <anoop_r> i which that helped me 
[00:33:04 CEST] <anoop_r> i wish
[00:33:35 CEST] <anoop_r> some people who don't know any thing about it make great mess
[00:34:10 CEST] <anoop_r> they don't even accept that even i didn't some work
[00:36:06 CEST] <anoop_r> i mean i did some work evn it is very little
[00:39:19 CEST] <wm4> this is hilarious
[00:39:44 CEST] <Compn> its not like you ever cheated in college / high school, wm4
[00:39:46 CEST] <Compn> :P
[00:39:56 CEST] <nevcairiel> I didn't
[00:40:15 CEST] <wm4> Compn: don't apply your standards to me
[00:40:17 CEST] <Compn> lol
[00:40:29 CEST] <Compn> i am just kidding, sorry if i offended wm4
[00:40:58 CEST] <Compn> banana sticker for nevcairiel
[00:41:27 CEST] <wm4> anoop_r: what kind of school are you in which has such crappy standards?
[00:41:32 CEST] <llogan> talking about stickers, anyone want any that didn't get any?
[00:42:11 CEST] <anoop_r> well it's some country
[00:42:29 CEST] <rcombs> with poor standards for spelling as well
[00:42:36 CEST] <anoop_r> ya
[00:42:47 CEST] <anoop_r> sorry for by band english
[00:42:52 CEST] <anoop_r> bad
[00:43:22 CEST] Action: rcombs tries to come up with a definition for "band english"
[00:43:40 CEST] <Compn> anoop_r : i am sorry , we are not helping correctly. what are you trying to  do, just rename all instances of "ffmpeg" ? thats easy to do.
[00:43:41 CEST] <anoop_r> i man bad english
[00:44:22 CEST] <rcombs> Compn: pipe `git grep` to `xargs sed -i`
[00:44:24 CEST] <anoop_r> forget about above line
[00:44:27 CEST] <Compn> but this channel is for development of ffmpeg itself. we dont help people with their projects (unless they help us)
[00:44:44 CEST] <Compn> so your question maybe better suited for #ffmpeg 
[00:45:31 CEST] <anoop_r> most of the time that channel look like empty
[00:45:46 CEST] <anoop_r> even though a lots of people are there
[00:46:19 CEST] <Compn> or try asking on ffmpeg-user mailing list
[00:46:26 CEST] <Compn> thats fairly busy :)
[00:46:31 CEST] <anoop_r> ok
[00:47:06 CEST] <anoop_r> :(
[00:50:20 CEST] <Compn> anoop_r : what is your native language ?
[00:50:46 CEST] <anoop_r> its malayalam
[00:51:05 CEST] <Compn> i'm not sure why rcombs is making fun of people who cant speak english, most of the devs here are not native english speakers.
[00:51:19 CEST] <rcombs> Compn: I'd be making fun of him either way
[00:51:26 CEST] <rcombs> poor English is just an easy target
[00:51:31 CEST] <anoop_r> i don't mind 
[00:51:50 CEST] <anoop_r> because you people at least talks to me
[00:52:12 CEST] <Compn> anoop_r : we dont have many indiain developers , at least none on irc that i'm aware of
[00:52:51 CEST] <Compn> i've been trying to get more developers from china and india to contribute
[00:54:11 CEST] <iive> actually good ffmpeg front-ends are quite rare.
[00:54:56 CEST] <wm4> Compn: seriously
[00:55:06 CEST] <wm4> Compn: I don't even know why you're saying this
[00:55:41 CEST] <Compn> which part? 
[00:56:14 CEST] <wm4> the last part
[00:56:22 CEST] <Compn> because sponsoring companies read our public irc logs and look for our openness to cooperate with developers from all over the world ?
[00:56:35 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07master:913685f55208: avdevice/lavfi: do not rescale AV_NOPTS_VALUE in lavfi_read_packet()
[00:56:36 CEST] <iive> diversity is good. 
[00:57:38 CEST] <wm4> this is not about diversity, just Compn's stupid claim
[00:57:51 CEST] <anoop_r>  does native aac support for ffmpeg comes under this channel
[00:57:57 CEST] <Compn> yes
[00:58:04 CEST] <Compn> do you have a problem with aac support ?
[00:58:12 CEST] <Compn> well maybe #ffmpeg for that too
[00:58:41 CEST] <Compn> wm4 : all my claims are stupid.
[00:58:51 CEST] <anoop_r> no i have no problem with fixed bit rate
[00:58:51 CEST] <Compn> anoop_r : lets talk in #ffmpeg
[00:58:53 CEST] <iive> i would advice against cheating... it would get a lot worse if they find out you forked ffmpeg just to hide its name.
[00:59:12 CEST] <anoop_r> ffmpeg -i Air.mkv -c:v libx265 -preset ultrafast -x265-params crf=30 -c:a aac -strict experimental -q:a 10 output.mp4
[00:59:40 CEST] <anoop_r> but cbr mode not working as expected
[01:00:15 CEST] <anoop_r> atleast meadiainfo shows it as constant bitrate
[01:00:26 CEST] <wm4> Compn: I've never seen you encourage something, other than replying positively to some bad ideas (independent of who had the idea)
[01:05:04 CEST] <Compn> wm4 : so you dont like me replying positively to the opencl support ?
[01:06:18 CEST] <durandal_1707> replying to almost anything 
[01:06:20 CEST] <wm4> it's alwqays the same: bad idea gets posted -> people try to discourage the idea using good technical arguments -> Compn replies with some nonsense encouraging it
[01:11:38 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.7:f5bb7b99927d: MAINTAINERS: add 2.7
[01:11:39 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.7:6e94e77632df: avdevice/lavfi: do not rescale AV_NOPTS_VALUE in lavfi_read_packet()
[01:12:12 CEST] <Compn> nobody likes me :(
[01:14:18 CEST] <wm4> well it's not like I hate you, but I find these posts I mentioned annoying and unhelpful
[01:15:16 CEST] <Compn> which posts ? 
[01:15:32 CEST] <wm4> but regarding the opencl support in general... I guess ffmpeg in general would rather accept a feature than reject it
[01:15:39 CEST] <wm4> even if it's completely pointless
[01:16:08 CEST] <wm4> ok I made enough friends today, I'm off
[01:16:27 CEST] <jamrial> opencl is not pointless. it's just currently used for a handful of supposedly not-very-useful filters
[01:16:36 CEST] <cehoyos_> You do realize that this was the main demand of the "new team"?
[01:16:46 CEST] <Compn> uh oh, opencl troll war incoming
[01:16:53 CEST] <Compn> cehoyos_ : which demand ?
[01:17:18 CEST] <cehoyos_> Not to reject patches easily
[01:20:37 CEST] Action: Compn also goes bbl
[01:20:52 CEST] <Compn> cehoyos_ : let me know if i'm doing anything wrong or need to improve 
[01:34:01 CEST] <anoop_r> anyway to show encoding progress in % rather than frames compeleted
[01:35:13 CEST] <llogan> anoop_r: please stop asking usage questions here.
[01:35:40 CEST] <anoop_r> sorry
[01:35:59 CEST] <anoop_r> i thought ffmpeg is always build to show output like that
[01:43:39 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07n2.7:HEAD: avdevice/lavfi: do not rescale AV_NOPTS_VALUE in lavfi_read_packet()
[01:48:05 CEST] <BBB> that was a funny backlog
[01:48:10 CEST] <BBB> so sad I missed that part
[01:57:36 CEST] <philipl> BBB: I'd swear we were being trolled.
[01:57:46 CEST] <philipl> Maybe his whole goal was to get wm4 and Compn to argue.
[02:11:43 CEST] <llogan> "Automatically rotate videos based on metadata in ffmpeg" <-- when was that implemented?
[02:15:44 CEST] <llogan> oh, 1630224
[02:15:56 CEST] <jamrial> ubitux implemented that for ffplay long ago
[02:16:20 CEST] <jamrial> then libav reimplemented it, and i guess for some reason the changelog entry wasn't removed for 2.7
[02:17:02 CEST] <jamrial> or maybe i'm thinking about a different feature
[02:17:57 CEST] <llogan> i recall ubitux's ffplay autorotate. i was trying to find out if the changelog was referring to the ffmpeg cli tool
[02:29:42 CEST] <michaelni> llogan, the merge of the autorotate/avconv adds the autorotate to ffmpeg (f5b26fbc2f564831b337f1de9faaaf6753a03871) IIRC
[02:31:05 CEST] <llogan> yeah, i'm looking at it now. i wonder how it works when using a filtergraph (haven't tried yet).
[03:24:50 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:3190acae6fd2: avformat/hlsenc: Free context after hls_append_segment
[03:24:51 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:ff6ee4571ceb: swscale/x86/rgb2rgb_template: handle the first 2 lines with C in rgb24toyv12_*()
[03:24:52 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:1e18ed781f8b: avcodec/mjpegdec: Fix context fields becoming inconsistent
[03:24:53 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:42b0ef9056e7: avcodec/utils: Check that the data is complete in avpriv_bprint_to_extradata()
[03:24:54 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:71ad971244a7: avcodec/flacdec: Call ff_flacdsp_init() unconditionally
[03:24:55 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:a0f1da9baab0: avcodec/pngdec: Check IHDR/IDAT order
[03:24:56 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:eeff4bdbb868: avcodec/rawdec: Check the return code of avpicture_get_size()
[03:24:57 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:0c50e41d63f0: avcodec/motion_est: use 2x8x8 for interlaced qpel
[03:24:58 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:f43127a51602: avformat/rmdec: Check codec_data_size
[03:24:59 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:a0976c15e694: swscale/x86/rgb2rgb_template: fix crash with tiny size and nv12 output
[03:25:00 CEST] <cone-815> ffmpeg 03wm4 07release/2.0:b9ab59a82907: avformat/matroskadec: fix handling of recursive SeekHead elements
[03:25:01 CEST] <cone-815> ffmpeg 03Rob Sykes 07release/2.0:32d24c8c05a9: swresample/soxr_resample: fix error handling
[03:25:02 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:4e6706f851d3: avformat/aviobuf: Check that avio_seek() target is non negative
[03:25:03 CEST] <cone-815> ffmpeg 03wm4 07release/2.0:a54aaa822ae7: lavu/frame: fix malloc error path in av_frame_copy_props()
[03:25:04 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:47b82e51be67: configure: create the tests directory like the doc directory
[03:25:05 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:87af5b3877a3: avcodec/vmdvideo: Check len before using it in method 3
[03:25:06 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:9a15f326828c: avcodec/utvideodec: Fix handling of slice_height=0
[03:25:07 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:22f8dfafa86b: avformat/mov: check atom nesting depth
[03:25:08 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:8b130c4aab85: swscale: increase yuv2rgb table headroom
[03:25:09 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:587cd92bd729: avcodec/h264: make the first field of H264Context an AVClass
[03:25:10 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:53dc6f66430e: avcodec/indeo3: use signed variables to avoid underflow
[03:25:11 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:ae19e19678a1: avcodec/h264: Clear delayed_pic on deallocation
[03:25:12 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:1fa9b7feb66d: avcodec/h264: Check *log2_weight_denom
[03:25:13 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:932e5c374a25: avcodec/indeo3: ensure offsets are non negative
[03:25:14 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:ca39fbe14da3: avformat/mov: Fix memleaks for duplicate STCO/CO64/STSC atoms
[03:25:15 CEST] <cone-815> ffmpeg 03Stefano Sabatini 07release/2.0:43b601d36ffc: lavf/segment: remove duplicated and inconsistent cleanup code in seg_write_packet()
[03:25:16 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:5ba020d64810: avformat/cdxl: Fix integer overflow of image_size
[03:25:17 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:5426a36300f3: avformat/flvdec: do not inject dts=0 metadata packets which failed to be parsed into a new data stream
[03:25:18 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:b0369f33693d: avformat/flvdec: Increase string array size
[03:25:19 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:9ac17454a92b: avfilter/vf_sab: fix filtering tiny images
[03:25:20 CEST] <cone-815> ffmpeg 03wm4 07release/2.0:776c481eb902: avcodec/dvdsubdec: fix out of bounds accesses
[03:25:21 CEST] <cone-815> ffmpeg 03Dale Curtis 07release/2.0:da29aadeb7f2: mov: Avoid overflow with mov_metadata_raw()
[03:25:22 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:227a26cd404c: avformat/mov: fix integer overflow in mov_read_udta_string()
[03:25:23 CEST] <cone-815> ffmpeg 03Dale Curtis 07release/2.0:3e11a186facf: mov: Fix negative size calculation in mov_read_default().
[03:25:24 CEST] <cone-815> ffmpeg 03Paul B Mahol 07release/2.0:e148369df93d: lavfi: check av_strdup() return value
[03:25:25 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:81a766e57a34: avformat/matroskadec: Use av_freep() to avoid leaving stale pointers in memory
[03:25:26 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:04e28b3b257f: avformat/mov: Fix mixed declaration and statement warning
[03:25:27 CEST] <cone-815> ffmpeg 03Johan Andersson 07release/2.0:bda7aa7cb39a: cmdutils: update copyright year to 2015.
[03:25:28 CEST] <cone-815> ffmpeg 03wm4 07release/2.0:59fc55b29738: avcodec/dvdsubdec: error on bitmaps with size 0
[03:25:29 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:c2439cad5101: ffmpeg: Clear error message array at init.
[03:25:30 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:61b31c693918: avcodec/flac_parser: fix handling EOF if no headers are found
[03:25:31 CEST] <cone-815> ffmpeg 03Vittorio Giovara 07release/2.0:3d5c48937beb: swscale: check memory allocations
[03:25:32 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:2324c670763c: avformat/rmdec: Check for overflow in ff_rm_read_mdpr_codecdata()
[03:25:33 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:1a8f5c9e2dbd: avformat/utils: Fix number suffixes in tb_unreliable()
[03:25:34 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:c21604b5e72d: swresample/dither: Cleanup number suffixes
[03:25:35 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:5c36d0e1cdb8: avformat/matroskadec: Fix number suffixes
[03:25:36 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:6e4e32e7590c: avformat/smacker: Fix number suffix
[03:25:37 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:89e08520fc61: avcodec/h264_cabac: use int instead of long for mbb_xy
[03:25:38 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:904f75d3c46b: avcodec/mpegvideo_enc: Fix number suffixes in rc_buffer_size calculation
[03:25:39 CEST] <cone-815> ffmpeg 03wm4 07release/2.0:64857ecd68b7: avformat/tta: fix crash with corrupted files
[03:25:40 CEST] <cone-815> ffmpeg 03wm4 07release/2.0:2e6b1915cb32: avformat/mpc8: fix hang with fuzzed file
[03:25:41 CEST] <cone-815> ffmpeg 03wm4 07release/2.0:76d18b6e095b: avformat/mpc8: fix broken pointer math
[03:25:42 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:2acdf2983832: avformat/mpc8: Use uint64_t in *_get_v() to avoid undefined behavior
[03:25:43 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:c6ef9ca9710a: avcodec/mjpegdec: Check escape sequence validity
[03:25:44 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:fe618beccf10: avcodec/mjpegdec: Check number of components for JPEG-LS
[03:25:45 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:0d728ed7926e: swscale/utils: Limit filter shifting so as not to read from prior the array
[03:25:46 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:11afd2e2dd15: avformat/thp: Check av_get_packet() for failure not only for partial output
[03:25:47 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:a22da15b0cac: avcodec/h264_ps: More completely check the bit depths
[03:25:48 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:1081c9decc63: avcodec/h264_slice: ignore SAR changes in slices after the first
[03:25:49 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:b4e4a5cbaa57: avcodec/mjpegdec: Skip blocks which are outside the visible area
[03:25:50 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:d20c761782c6: avformat/idcin: Use 64bit for ret to avoid overflow
[03:25:51 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:052cd9e55230: avformat/gxf: Use 64bit for res to avoid overflow
[03:25:52 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:fc0d183729c2: avformat/mvdec: Use 64bit for ret to avoid overflow
[03:25:53 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:79460c552f8d: avformat/vqf: Use 64bit for ret to avoid overflow
[03:25:54 CEST] <cone-815> ffmpeg 03Anton Khirnov 07release/2.0:e201e753688c: h264: initialize H264Context.avctx in init_thread_copy
[03:25:55 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:24bb74619415: avcodec/x86/mlpdsp_init: Simplify mlp_filter_channel_x86()
[03:25:56 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:c4353d7ee203: avcodec/a64multienc: use av_frame_ref instead of copying the frame
[03:25:57 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:ffdfa80147f3: avcodec/a64multienc: don't set incorrect packet size
[03:25:58 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:1cba89a135ba: avcodec/a64multienc: fix use of uninitialized values in to_meta_with_crop
[03:25:59 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:3cca4c770ea6: swscale/utils: More carefully merge and clear coefficients outside the input
[03:26:00 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:bc0bf39e4ccf: avcodec/snowdec: Fix ref value check
[03:26:01 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:22dc1fd166eb: avcodec/zmbv: Check len before reading in decode_frame()
[03:26:02 CEST] <cone-815> ffmpeg 03James Cowgill 07release/2.0:1697813db84a: mips/acelp_filters: fix incorrect register constraint
[03:26:03 CEST] <cone-815> ffmpeg 03Dyami Caliri 07release/2.0:5ae61395afe6: Fix buffer_size argument to init_put_bits() in multiple encoders.
[03:26:04 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:15fc5263f9d4: avformat/adxdec: check avctx->channels for invalid values
[03:26:05 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:77955dcabfce: avformat/bit: check that pkt->size is 10 in write_packet
[03:26:06 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:45432879ff41: avformat/bit: only accept the g729 codec and 1 channel
[03:26:07 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:852ef62b85c0: swscale/utils: clear formatConvBuffer on allocation
[03:26:08 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:3bfaadfadaba: avformat/flvenc: check that the codec_tag fits in the available bits
[03:26:09 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:5f92a0d1e94c: avcodec/utils: Align YUV411 by as much as the other YUV variants
[03:26:10 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:ff0985347c2c: avcodec/tiff: move bpp check to after "end:"
[03:26:11 CEST] <cone-815> ffmpeg 03Vittorio Giovara 07release/2.0:e69c43035363: aic: Fix decoding files with odd dimensions
[03:26:12 CEST] <cone-815> ffmpeg 03Federico Tomassetti 07release/2.0:27432f259d50: mdec: check for out of bounds read
[03:26:13 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:63efad67ee56: avcodec/rv10: check size of s->mb_width * s->mb_height
[03:26:14 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:3d11279745a6: avformat/rm: limit packet size
[03:26:15 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:9be586e16647: ffmdec: fix infinite loop at EOF
[03:26:16 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:0f2401e3cb44: ffmdec: limit the backward seek to the last resync position
[03:26:17 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:1c1574e3777a: avcodec/options_table: remove extradata_size from the AVOptions table
[03:26:18 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:c9edf502cd8b: roqvideoenc: set enc->avctx in roq_encode_init
[03:26:19 CEST] <cone-815> ffmpeg 03Xiaohan Wang 07release/2.0:46ee330d3a1e: Fix read-after-free in matroska_read_seek().
[03:26:20 CEST] <cone-815> ffmpeg 03Martin Storsjö 07release/2.0:b13cba308253: arm: Suppress tags about used cpu arch and extensions
[03:26:21 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:ae14a444f282: avcodec/012v: Check dimensions more completely
[03:26:22 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:e5f7aeb46f55: avcodec/012v: redesign main loop
[03:26:23 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:572cfee405d9: ffmdec: make sure the time base is valid
[03:26:24 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:a26a200b3f13: avformat/mov: Use sizeof(filename) instead of a literal number
[03:26:25 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:dece653d0de9: avformat/mov: Check for string truncation in mov_open_dref()
[03:26:26 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:9c10ef55b0aa: avformat/mov: Disallow ".." in dref unless use_absolute_path is set
[03:26:27 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:81ba3b1b9166: avcodec/dnxhddec: Check that the frame is interlaced before using cur_field
[03:26:28 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:936554f1d738: avcodec/msrledec: restructure msrle_decode_pal4() based on the line number instead of the pixel pointer
[03:26:29 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:50e17cdc984e: avutil/pca: Check for av_malloc* failures
[03:26:30 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:2f5186575177: avcodec/h264_refs: Do not set reference to things which dont exist
[03:26:31 CEST] <cone-815> ffmpeg 03Timothy Gu 07release/2.0:b73e47a413ae: tests/fate.sh: report different status for different errors
[03:26:32 CEST] <cone-815> ffmpeg 03Timothy Gu 07release/2.0:feb869b6829c: tests: Fix test name for pixfmts tests(cherry picked from commit e1ee0521a698809ed216e9e5c11bd2bbb466ed04)
[03:26:33 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:a5e2f793170b: avcodec/aacdec: Fix storing state before PCE decode
[03:26:34 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:97b137a64071: msrledec: use signed pixel_ptr in msrle_decode_pal4
[03:26:35 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:743973f56f94: aasc: return correct buffer size from aasc_decode_frame
[03:26:36 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:c962405c1968: aacpsy: avoid psy_band->threshold becoming NaN
[03:26:37 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:997a997465b5: ac3: validate end in ff_ac3_bit_alloc_calc_mask
[03:26:38 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:82e24ec792fc: alsdec: ensure channel reordering is reversible
[03:26:39 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:baa58c19c48f: avcodec/alsdec: Use av_mallocz_array() for chan_data to ensure the arrays never contain random data
[03:26:40 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:a1f0c1b6fe10: alsdec: validate time diff index
[03:26:41 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:1a87782c2e55: avformat/utils: Ensure that AVFMT_FLAG_CUSTOM_IO is set before use
[03:26:42 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:1303d8a204d1: alsdec: check sample pointer range in revert_channel_correlation
[03:26:43 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:cee951b596f1: mpeg4videodec: only allow a positive length
[03:26:44 CEST] <cone-815> ffmpeg 03Vittorio Giovara 07release/2.0:cf13885622c3: lavf: Reset global flag on deinit
[03:26:45 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:c7e56b8b9cc2: alsdec: only adapt order for positive max_order
[03:26:46 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:2d204f313bd0: alac: reject rice_limit 0 if compression is used
[03:26:47 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:9f665b0d6a12: tests/fate-run: do not attempt to parse tiny_psnrs output if it failed
[03:26:48 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:57208a0999a0: apedec: set s->samples only when init_frame_decoder succeeded
[03:26:49 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:e6cc6a08d993: apedec: prevent out of array writes in decode_array_0000
[03:26:50 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:35ca5eb11f19: ffmpeg: remove incorrect network deinit
[03:26:51 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:7f2b3c3bee6f: nutdec: check for negative frame rate in decode_info_header
[03:26:52 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:0691554e68e3: nutdec: check chapter creation in decode_info_header
[03:26:53 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:9d5362a45cbd: nutdec: fix illegal count check in decode_main_header
[03:26:54 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:9ac72b1dde34: avcodec/wavpack: Check L/R values before use to avoid harmless integer overflow and undefined behavior in fate
[03:26:55 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:ac7a211b3be1: matroskadec: use uint64_t instead of int for index_scale
[03:26:56 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:d2142c3b3a98: avcodec/ffv1dec: Check chroma shift parameters
[03:26:57 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:68e1c80c67da: matroskadec: check s->streams[k] before using it
[03:26:58 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:35eabb85b2a6: avformat/matroskadec: Use tracks[k]->stream instead of s->streams[k]
[03:26:59 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:4d27d864a653: avidec: avoid infinite loop due to negative ast->sample_size
[03:27:00 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:869613728cea: diracdec: prevent overflow in data_unit_size check
[03:27:01 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:ebc768cfa1c9: diracdec: avoid overflow of bytes*8 in decode_lowdelay
[03:27:02 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:994538281645: diracdec: check if reference could not be allocated
[03:27:03 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:bf6655576915: diracdec: check that block length is valid
[03:27:04 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:998d943cce73: aacsbr: break infinite loop in sbr_hf_calc_npatches
[03:27:05 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:fdc62caf3075: avutil/dict: Use size_t for appending strings
[03:27:06 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:db0c4d2e3c70: avformat/vorbiscomment: Check entry length in ff_vorbiscomment_write()
[03:27:07 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:93a04473fc09: avutil/avstring: Use size_t in av_strlcatf()
[03:27:08 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:c7b089b1ac80: avformat/url: Use size_t for len from strlen()
[03:27:09 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:5399107682ae: avformat/subtitles: Use size_t for len
[03:27:10 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:1dad249ae66c: tools/graph2dot: use larger data types than int for array/string sizes
[03:27:11 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:4a9ce87ef334: avformat/rtpdec_xiph: Check upper bound on len in xiph_handle_packet()
[03:27:12 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:b470e21d6155: avformat/rtpenc_jpeg: Check remaining buffer size for SOS
[03:27:13 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:8ad873ed4e99: avformat/nutdec: Fix use of uinitialized value
[03:27:14 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:e553282860c8: avformat/matroskadec: Cleanup error handling for bz2 & zlib
[03:27:15 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:0bf455c0543d: avcodec/vqavideo: Check chunk size
[03:27:16 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:32f568f82c46: avcodec/jpeg2000dec: fix boolean operator
[03:27:17 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:c0726da9339f: avcodec/mjpegdec: fix len computation in ff_mjpeg_decode_dqt()
[03:27:18 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:136bd71f690f: avcodec/mpeg4audio: add some padding/alignment to MAX_PCE_SIZE
[03:27:19 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:36bc9519b6a6: avcodec/cavsdec: Check esc_code
[03:27:20 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:c532c56e7b77: avcodec/dcadec: Check nchans
[03:27:21 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:3fedff398525: avcodec/dcadec: Check subsubframes
[03:27:22 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:ffd894917fd5: ffmpeg_opt: Fix -timestamp parsing
[03:27:23 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:990c2ee3bd95: avcodec/proresdec2: Reset slice_count on deallocation
[03:27:24 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:0888e7110e7a: avcodec/shorten: Fix code depending on signed overflow behavior
[03:27:25 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:a7c8dc67e773: avcodec/shorten: Check skip_bytes()
[03:27:26 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:f58c9746c397: avcodec/shorten: More complete pred_order check
[03:27:27 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:41c81556a70c: avcodec/smvjpegdec: check avcodec_decode_video2() return code
[03:27:28 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:025b38f3a614: avcodec/sonic: More completely check sample_rate_index and channels
[03:27:29 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:a3d05bf6be2e: avcodec/dcadec: Check scale table index
[03:27:30 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:c01f5517280c: avcodec/dcadec: Check active_bands
[03:27:31 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:13b22617c185: avcodec/libtheoraenc: Check for av_malloc failure
[03:27:32 CEST] <cone-815> ffmpeg 03Andrey Utkin 07release/2.0:ced57c6ef3e5: rtpenc_jpeg: handle case of picture dimensions not dividing by 8
[03:27:33 CEST] <cone-815> ffmpeg 03Martin Storsjö 07release/2.0:fd1be2bd874c: rtsp: Make sure we don't write too many transport entries into a fixed-size array
[03:27:34 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:9dd9090b89d5: avformat/nutdec: Return error on EOF from get_str()
[03:27:35 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:00b8a9dd8d99: avformat/nutdec: Fix recovery when immedeately after seeking a failure happens
[03:27:36 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:6cb8599bab1b: avformat/nutdec: Check X in 2nd branch of index reading
[03:27:37 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:5123c8aa0bf3: nutdec: fix infinite resync loops
[03:27:38 CEST] <cone-815> ffmpeg 03Andreas Cadhalpun 07release/2.0:b6ae28c2e9cb: nutdec: stop skipping bytes at EOF
[03:27:39 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:03c80a740043: avcodec/h264_slice: More complete cleanup in h264_slice_header_init()
[03:27:40 CEST] <cone-815> ffmpeg 03Luca Barbato 07release/2.0:04bb8cc84212: h264: Make sure reinit failures mark the context as not initialized
[03:27:41 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:9db97584ca5b: swresample: Check the return value of resampler->init()
[03:27:42 CEST] <cone-815> ffmpeg 03Ganesh Ajjanagadde 07release/2.0:6d7384bd735b: swresample/dither: check memory allocation
[03:27:43 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:ff0e9060a53f: avformat/mxfenc: Accept MXF D-10 with 49.999840 Mbit/sec
[03:27:44 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:a3740b6a8644: avcodec/alsdec: Check for overread
[03:27:45 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:6fd4b2b84f60: avcodec/adpcm: Check for overreads
[03:27:46 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:1fe67c9472ad: avformat/ffmdec: Check ffio_set_buf_size() return value
[03:27:47 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:7a9f4f2816be: avcodec/jpeg2000dec: Check that coords match before applying ICT
[03:27:48 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:8320aa7dc743: avcodec/x86/h264_weight: handle weight1=128
[03:27:49 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:af79d964a73f: avdevice/lavfi: do not rescale AV_NOPTS_VALUE in lavfi_read_packet()
[03:27:50 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07release/2.0:0caff57c42ca: Update for 2.0.7
[03:43:53 CEST] <jk_> Hello  everyone
[03:46:10 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07n2.0.7:HEAD: avdevice/lavfi: do not rescale AV_NOPTS_VALUE in lavfi_read_packet()
[03:46:16 CEST] <jk_> i use  msys2  and mingw-w64  compile ffmpeg 2.5.7   configured with " ./configure --enable-gpl --enable-version3 --disable-static --enable-shared --disable-iconv --disable-bzlib --disable-zlib --disable-w32threads"  on my win7 os    ,and then make  ,but i got some errors , can some one help me?
[03:56:21 CEST] <michaelni> Luigi12_work, just made a new 2.0.7 release 
[04:40:35 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07master:213f2da46adc: avcodec/jpeg2000dec: Fix shift for RGB(A)48/64
[05:04:22 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07master:81471d824676: avcodec/jpeg2000dec: increase tile_parts to 6
[05:04:23 CEST] <cone-815> ffmpeg 03Michael Niedermayer 07master:b58a3c897559: avcodec/jpeg2000dec: Print the correct variable in the component error message
[09:41:44 CEST] <durandal_1707> michaelni: pushed tak patch?
[11:04:16 CEST] <j-b> 'morning
[11:04:22 CEST] <michaelni> durandal_1707, i thought andreas would push it but i could too
[11:51:42 CEST] <cone-059> ffmpeg 03Michael Niedermayer 07master:aef0e0f00980: avcodec/h264_slice: Use AVFrame diemensions for grayscale handling
[12:22:49 CEST] <cone-059> ffmpeg 03Andreas Cadhalpun 07master:05c57ba2f423: takdec: ensure chan2 is a valid channel index
[13:29:24 CEST] <cone-059> ffmpeg 03Shivraj Patil 07master:271195f85bbc: avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for HEVC loop filter and sao functions
[13:51:45 CEST] <Luigi12_work> michaelni: thanks!
[14:31:12 CEST] <cone-059> ffmpeg 03Shivraj Patil 07master:d6d98237ed01: avcodec/mips: MSA (MIPS-SIMD-Arch) optimizations for HEVC intra prediction functions
[15:47:03 CEST] <ubitux> VideoToolbox ’ "vt", "videotoolbox", "vtb"?
[15:58:06 CEST] <ubitux> wm4: so, no _VLD pix fmt?
[15:58:37 CEST] <ubitux> only AV_PIX_FMT_<HWACCEL>?
[15:59:55 CEST] <wm4> I don't even know what the VLC means
[15:59:58 CEST] <wm4> *VLD
[16:00:17 CEST] <ubitux> variable length decoding apparently
[16:00:20 CEST] <wm4> so I'd propose AV_PIX_FMT_VT
[16:00:39 CEST] <ubitux> > Variable-Length Decoding (VLD), more commonly known as slice-level acceleration
[16:00:46 CEST] <wm4> eh
[16:00:46 CEST] <ubitux> assuming that's what it's about
[16:00:59 CEST] <ubitux> VLD actually refering to the entropy level
[16:01:04 CEST] <ubitux> anyway ok
[16:01:19 CEST] <ubitux> so basically, same interface as lavc/vda.h
[16:01:22 CEST] <wm4> the pixel format only signals that the AVFrame points to some API data structure
[16:01:32 CEST] <wm4> (or a hardware surface ID)
[16:01:49 CEST] <wm4> it could be that it's the same
[16:01:50 CEST] <ubitux> yeah, in the plane 4 or something?
[16:01:57 CEST] <wm4> in that case, a separate pixfmt makes no sense
[16:02:08 CEST] <wm4> other than the fuck retarded way of selecting the hw accel API
[16:02:27 CEST] <wm4> (so 2 decoders can't use the same pixfmt, even if they output exactly the same thing)
[16:02:49 CEST] <wm4> nevcairiel: ping
[16:02:51 CEST] <ubitux> aren't they sharing frames with a context very specific to the hw decoder?
[16:02:54 CEST] <wm4> is this correct?
[16:03:03 CEST] <wm4> ubitux: doesn't matter
[16:03:09 CEST] <wm4> and with VDA, there's actually no context
[16:03:36 CEST] <wm4> anyway, the VDA format wraps CVPixelBufferRef
[16:03:39 CEST] <wm4> in plane 3
[16:03:43 CEST] <wm4> (the 4th)
[16:03:52 CEST] <wm4> what does VT output?
[16:06:28 CEST] <ubitux> wm4: still trying to figure out things so dunno.
[16:10:08 CEST] <ubitux> i just see all kind of confusing glue code, but it looks pretty straightforward
[16:11:00 CEST] <wm4> all I know is that the VDA API uses VT underneath
[16:11:06 CEST] <ubitux> yeah
[16:11:29 CEST] <ubitux> the last version of the patch is @ http://ffmpeg.org/pipermail/ffmpeg-devel/2012-September/131545.html
[16:11:42 CEST] <ubitux> i contacted the guy but no answer so far
[16:11:58 CEST] <ubitux> > frame->data[3] = (void*)vt_ctx->cv_buffer;
[16:12:07 CEST] <ubitux> looks pretty similar to vda 
[16:13:37 CEST] <ubitux> cv_buffer begin a CVImageBufferRef, which somehow mismatch with the CVPixelBufferRef from vda but well 
[16:13:40 CEST] <nevcairiel> wm4: ?
[16:14:20 CEST] <ubitux> > typedef CVImageBufferRef CVPixelBufferRef;
[16:14:22 CEST] <ubitux> lol ok
[16:14:41 CEST] <ubitux> so yeah well, i could probably just alias VT on VDA
[16:16:54 CEST] <wm4> nevcairiel: we have 2 hwaccel APIs (VDA and VT), which both return the same surface type
[16:16:59 CEST] <wm4> so they'd have the same pixfmt
[16:17:10 CEST] <wm4> which makes it impossible to select one of them with get_format
[16:17:50 CEST] <nevcairiel> just accept defeat and create a new pixfmt
[16:20:00 CEST] <nevcairiel> maybe someone will create a new API at some point
[16:22:55 CEST] <nevcairiel> does one of you happen to know if linking in VDA or VT will still produce a binary thats usable on all relevant OS versions? ie. maybe even those were VT isn't supported yet?
[16:24:17 CEST] <wm4> VT was always there, but until a while ago it wasn't public
[16:24:19 CEST] <wm4> whatever that means
[16:24:31 CEST] <nevcairiel> i'm mostly interested in this for $work, where I do need to provide a binary that at least runs on 10.6, even if some things don't work there (ie. we dont support video on 10.6, but of course the app itself still has to work)
[16:25:00 CEST] <nevcairiel> i should somehow get a 10.6 test system .. but those old OSX versions are hard to get running in a VM :(
[16:25:40 CEST] <Daemon404> nevcairiel, should be easy if the vm host is also os x
[16:25:48 CEST] <nevcairiel> maybe i should try that
[16:26:00 CEST] <nevcairiel> u've been trying on a windows host, and newer OSX versions work there
[16:26:02 CEST] <Daemon404> parallels etc support os x vms fine... on os x
[16:26:10 CEST] <Daemon404> its the only official way to run os x vms
[16:26:38 CEST] <wm4> Apple manages to be evil in all sorts of creative ways
[16:26:41 CEST] <nevcairiel> its about time i get hw decoding on osx working i guess
[16:27:13 CEST] <Daemon404> wm4, i can kind of understand becaus they only test on their hw... but still
[16:27:20 CEST] <nevcairiel> VDA was present in 10.6 at least, so using that should give me some kind of portable binary
[16:27:42 CEST] <Daemon404> i thought the general rule was any non-current os x is dead to everyone
[16:27:58 CEST] <wm4> it's like with windows XP
[16:28:20 CEST] <nevcairiel> Luckily I was strong on that front and we just dont support video hwaccel before vista =P
[16:28:37 CEST] <nevcairiel> (which introduced DXVA2)
[16:29:30 CEST] <wm4> oh, so DXVA2 is this "new"
[16:30:50 CEST] <nevcairiel> well it inherited a lot of things from DXVA1, but it was terrible to use back then
[16:31:15 CEST] <nevcairiel> for example, it only supported direct rendering, copy back to system memory was impossible
[16:33:04 CEST] <BtbN> not even via rendering to a framebuffer?
[16:33:36 CEST] <nevcairiel> i'm sure you could've come up with some elaborate way to somehow cheat it, but it would have been extremely convoluted
[16:34:00 CEST] <wm4> I should try to get direct rendering to work with DXVA2
[16:34:03 CEST] <wm4> but it seems pretty useless
[16:34:13 CEST] <wm4> the driver would get control over YUV conversion and probably more
[16:34:21 CEST] <nevcairiel> probably
[16:35:11 CEST] <nevcairiel> if you have a d3d renderer, you somehow need to convert the surface the decoded frame is on to a texture, which unfortunately isn't a straight 1:1 conversion
[16:35:28 CEST] <nevcairiel> if your renderer is opengl, you could try doing it with the interop .. i wonder how that works
[16:37:06 CEST] <wm4> (probably barely)
[16:37:26 CEST] <nevcairiel> if you can minimize memcpy's, ie. reduce it to two copys, one from the surface to a membuffer, and one from the membuffer back onto your renderer surface, its pretty efficient as it is
[16:37:35 CEST] <nevcairiel> I got CPU usage down to nearly nothing on NVIDIA cards
[16:37:41 CEST] <nevcairiel> its a bit higher on intel and amd, since they suck
[16:38:16 CEST] <BtbN> You can make the frames never leave the GPU with nvidia dxva and nvenc on windows.
[16:38:17 CEST] <nevcairiel> NVIDIA seems to actually use proper DMA to access the GPU memory, CPU usage seems consistently lower
[16:39:48 CEST] <nevcairiel> You can also do that when rendering to the screen, of course, but the question is if its worth it, since you may give up quite a bit of flexibility
[16:40:25 CEST] <BtbN> If you just want to reencode something, without applying any additional effects, it's the fastest way
[16:40:52 CEST] <nevcairiel> I can decode 4K with DXVA2, copy it to sys mem, upload it back to renderer, and then render it, and still have only ~3% CPU use, so any direct rendering seems kinda pointless
[16:41:10 CEST] <BtbN> It takes a lot of load from the PCIe bus
[16:41:31 CEST] <BtbN> Which actualy can have a lot of impact, especialy on games
[16:41:47 CEST] <nevcairiel> presumably i'm not watching movies and gaming at the same time =P
[16:42:33 CEST] <BtbN> But people play games and stream them
[16:42:52 CEST] <wm4> nevcairiel: and the copy is with the CPU? (i.e. SSE memcpy)
[16:42:52 CEST] <nevcairiel> but there isnt even a decoder in play here
[16:43:19 CEST] <nevcairiel> wm4: yes, some fancy specialized SSE2 copy functions
[16:43:36 CEST] <BtbN> Yeah, but for games, the raw frame is pulled from the gpu, and then re-uploaded there to encode.
[16:43:40 CEST] <nevcairiel> eh, SSE4 i mean
[16:43:44 CEST] <BtbN> Same PCIe bus load as if the GPU were decoding
[16:43:54 CEST] <nevcairiel> actually its probably higher, since its RGB
[16:44:19 CEST] <BtbN> You can, and should, convert it to YUV420 on the gpu
[16:45:00 CEST] <nevcairiel> If you can avoid that, and for your use-case you think its worth it, then great. I'm just saying for simple video playback, the flexibility that a copy to sysmem offers is IMHO worth it, at least on NVIDIA where this is super efficient
[16:46:44 CEST] <BtbN> At least OBS pushes everything to the GPU, even raw system ram sources. Depends a lot on the implementation
[16:46:53 CEST] <BtbN> Every effect and filter is done via shaders
[16:47:18 CEST] <BtbN> And because of how it's implemented, it still copys the final yuv from the gpu to system ram, just so nvenc can re-upload it...
[16:55:44 CEST] <ubitux> grumbl i'm still confused at this AV_PIX_FMT_VDA vs AV_PIX_FMT_VDA_VLD
[16:56:14 CEST] <nevcairiel> one is the old one, one the new one we inhereited from libav?
[16:58:24 CEST] <ubitux> both are present in libav
[16:58:38 CEST] <nevcairiel> oh well
[16:59:10 CEST] <ubitux> "Add AV_PIX_FMT_VDA for new-style VDA acceleration." sounds indeed like the "new" one but mmh
[16:59:23 CEST] <ubitux> i see no deprecated or alike wrt to _VLD one
[17:00:02 CEST] <wm4> huh
[17:00:27 CEST] <nevcairiel> i always get lost when i think about new and old with VDA
[17:00:40 CEST] <nevcairiel> ffmpeg has a h264_vda decoder, does it not? is that the "old" one?
[17:00:42 CEST] <wm4> it's possible that AV_PIX_FMT_VDA_VLD required you to manage the surface manually
[17:00:49 CEST] <wm4> just forget about the old API
[17:00:59 CEST] <wm4> and the "decoder" is actually a thing that reads back to system memory
[17:01:06 CEST] <wm4> so it doesn't have anything to do with this
[17:01:44 CEST] <wm4> the VDA hwaccel API used to be awful, but the "new" one is trivial
[17:02:11 CEST] <wm4> really, laughably trivial... you only need to call av_vda_default_init()
[17:03:16 CEST] <nevcairiel> apparently, there is a init2 now
[17:03:31 CEST] <nevcairiel> lets you ask for a pixel format, or so i am told
[17:03:44 CEST] <wm4> yes
[17:03:55 CEST] <wm4>     AVVDAContext *vdactx = av_vda_alloc_context();
[17:03:55 CEST] <wm4>     vdactx->cv_pix_fmt_type = kCVPixelFormatType_420YpCbCr8BiPlanarVideoRange;
[17:03:55 CEST] <wm4>     int err = av_vda_default_init2(ctx->avctx, vdactx);
[17:04:05 CEST] <wm4> and you get NV12 as 2 textures
[17:04:15 CEST] <nevcairiel> does that actually work reliably
[17:04:23 CEST] <wm4> really not sure
[17:08:28 CEST] <ubitux> j-b: how is vt broken on mobile?
[17:09:13 CEST] <ubitux> Daemon404: yes the only reason so far is that one is available on ios
[17:10:17 CEST] <Daemon404> any reason to keep vda?
[17:10:26 CEST] <ubitux> compat with <10.8
[17:10:42 CEST] <wm4> i.e. "none"
[17:10:45 CEST] Action: nevcairiel votes for keeping
[17:10:52 CEST] <ubitux> (and similar argument for ff compat)
[17:11:18 CEST] <ubitux> anyway, before wondering about keeping vda, i need to figure out how to have vt properly in ff
[17:11:24 CEST] <wm4> or what lu_zero said, transparently switching between them
[17:11:24 CEST] <nevcairiel> remove all the old vda code if you want to clean up, but keep the new vda hwaccel which is apparently easy to use :)
[17:11:28 CEST] <ubitux> but using vda as a base reference is really confusing
[17:36:24 CEST] <ubitux> yet another q: should the new VT accel pixfmt, be named with the _VLD suffix like any other accel (but still copy the -new- vda api)?
[17:37:07 CEST] <nevcairiel> the VLD suffix is more legacy than anything else, standards have long moved away from non-VLD HWaccels
[17:37:59 CEST] <nevcairiel> so .. do what you want
[17:37:59 CEST] <nevcairiel> :D
[17:38:03 CEST] <ubitux> :(
[17:38:07 CEST] <ubitux> ok.
[17:39:59 CEST] <ubitux> i don't understand why vda had/has a _vld suffix in the first place
[17:41:47 CEST] <ubitux> it seems to be able to deal with slices so..
[17:48:09 CEST] <cone-690> ffmpeg 03Igor Derzhavin 07master:203f9c8ff050: libavformat/flvdec.c: don't build index_entries for input stream if AVIOContext is not seekable
[17:48:10 CEST] <cone-690> ffmpeg 03Michael Niedermayer 07master:63276fa9c1d1: avcodec/mips: remove ;;
[18:41:40 CEST] <cone-690> ffmpeg 03Stephan Holljes 07master:de3a7ef2042f: lavf/http: Correctly terminate session with HTTP POST client.
[20:30:01 CEST] <cone-690> ffmpeg 03Janne Grunau 07master:529c05698e88: movenc: fixes a questionable valgrind uninitialized value warning
[20:30:02 CEST] <cone-690> ffmpeg 03Michael Niedermayer 07master:049161fb7c66: Merge commit '529c05698e88b057f0bea61e0d85f2b42925b5ea'
[20:49:40 CEST] <cone-690> ffmpeg 03Vittorio Giovara 07master:7061bf0925a7: mpegvideo: Move Picture-related functions to a separate file
[20:49:41 CEST] <cone-690> ffmpeg 03Michael Niedermayer 07master:88679aeb08b9: Merge commit '7061bf0925a7d6c5899bbe710b2b8cdae9005274'
[20:58:24 CEST] <cone-690> ffmpeg 03Vittorio Giovara 07master:c6aa0554b0c3: mpegvideo: Drop release_unused_pictures() function
[20:58:25 CEST] <cone-690> ffmpeg 03Michael Niedermayer 07master:d784d6b7ed25: Merge commit 'c6aa0554b0c3e31fec4580b68ea85b66966cd381'
[21:07:52 CEST] <cone-690> ffmpeg 03Vittorio Giovara 07master:d1dd0d404c08: mpegvideo: Move block permutation function where is used
[21:07:53 CEST] <cone-690> ffmpeg 03Michael Niedermayer 07master:21d8e2c0b5aa: Merge commit 'd1dd0d404c085f4fce7b8358b4aea677761c5d88'
[21:19:01 CEST] <cone-690> ffmpeg 03Vittorio Giovara 07master:c0e6b8cab874: mpegvideo: Mark one function as static
[21:19:02 CEST] <cone-690> ffmpeg 03Michael Niedermayer 07master:11f86ec20483: Merge commit 'c0e6b8cab874db97b6818007bc86507c8d213083'
[23:05:07 CEST] <cehoyos_> michaelni: libavcodec/mpegpicture.c needs a copyright notice
[23:09:28 CEST] <cone-690> ffmpeg 03Lou Logan 07master:dc52e6200df6: doc/ffmpeg: remove tips
[23:23:57 CEST] <michaelni> cehoyos_, yes, do want to fix it?
[23:51:38 CEST] <cone-690> ffmpeg 03Michael Niedermayer 07master:c465254fbf18: avformat/mpegts: Do not ignore registration descriptors when probing is enabled
[00:00:00 CEST] --- Thu Jun 11 2015


More information about the Ffmpeg-devel-irc mailing list