[Ffmpeg-devel-irc] ffmpeg-devel.log.20150827

burek burek021 at gmail.com
Fri Aug 28 02:05:02 CEST 2015


[00:14:53 CEST] <cone-346> ffmpeg 03Paul B Mahol 07master:0190c372ef79: avfilter/framesync: allocate FFFrameSyncIn internally
[00:14:53 CEST] <cone-346> ffmpeg 03Paul B Mahol 07master:319440e54f47: avfilter: add hstack & vstack filter
[01:48:10 CEST] <michaelni> kierank, did you see "Marina Zhurakhi (1.6K) [Outreachy Announce] [COORDINATION] internships completion feedback from mentors due Friday, August 28"
[03:01:21 CEST] <cone-346> ffmpeg 03Carl Eugen Hoyos 07master:7f9656f10df5: lavc/dnxhdenc: Fix ibias default.
[03:31:50 CEST] <cone-346> ffmpeg 03Carl Eugen Hoyos 07master:33908f08377d: lavf/mov: Support unusual alac files without frma and alac atoms.
[03:47:12 CEST] <cone-346> ffmpeg 03Timothy Gu 07master:da0e76955a08: ffmpeg_opt: Add -hwaccels option that lists all supported hwaccels
[04:39:52 CEST] <cone-346> ffmpeg 03Michael Niedermayer 07master:81a8701eb52d: avformat/oggenc: Check segments_count for headers too
[09:20:29 CEST] <cone-386> ffmpeg 03Vittorio Giovara 07master:fdd021884d5c: dnxhddata: Keep a single CID in the table names
[09:20:29 CEST] <cone-386> ffmpeg 03Vittorio Giovara 07master:d3ae4c65942d: dnxhddata: List the reused tables in a comment
[09:20:29 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:bd074bde02aa: Merge commit 'fdd021884d5c06fb9ad65cb0040bb5717a7b084b'
[09:20:29 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:3f491224fe7f: Merge commit 'd3ae4c65942d67a294fd56eabbbdcce6756fab5f'
[09:29:42 CEST] <cone-386> ffmpeg 03Vittorio Giovara 07master:5e129ed655bf: dnxhddata: Group together DC-related tables
[09:29:43 CEST] <cone-386> ffmpeg 03Vittorio Giovara 07master:403ea4ac7289: dnxhddata: Merge a few duplicated DC tables
[09:29:44 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:3b0af76c1b43: Merge commit '5e129ed655bff5b6d90355c0b713d7aaba3898ec'
[09:29:45 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:305f46175429: Merge commit '403ea4ac7289ac39229452b6b5e2f8ebcc00f2a1'
[09:34:39 CEST] <cone-386> ffmpeg 03Vittorio Giovara 07master:efbfb1ad1112: dnxhddata: Group together RUN-related tables
[09:34:40 CEST] <cone-386> ffmpeg 03Vittorio Giovara 07master:a4615572b576: dnxhddata: Merge a few duplicated RUN tables
[09:34:41 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:f19a2838678e: Merge commit 'efbfb1ad1112cea79bef51fd9f30c0c94735abfc'
[09:34:42 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:f30132f9c406: Merge commit 'a4615572b576d3ef7ee2f11529d935e61bf4ebb8'
[09:40:21 CEST] <cone-386> ffmpeg 03Vittorio Giovara 07master:d68705c9756e: dnxhddata: Add tables for missing DNx100 profiles
[09:40:22 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:427598a2f423: Merge commit 'd68705c9756e6558c8e28d90b4c364f25ba72083'
[09:49:55 CEST] <cone-386> ffmpeg 03Luca Barbato 07master:2157df425bd9: hlsenc: Support outputting specific versions
[09:49:56 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:540f2800980e: Merge commit '2157df425bd909854fd4afcec4aa3311d8a3b31b'
[09:51:29 CEST] <cone-386> ffmpeg 03Luca Barbato 07master:413d4e54a9bf: nvenc: Properly free the fifos
[09:51:30 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:ebff705c2c55: Merge commit '413d4e54a9bffe2d0afdc6d8a80f516e5df6a421'
[09:53:24 CEST] <cone-386> ffmpeg 03Luca Barbato 07master:e176639bcbf4: webm: Explicitly select libvpx, libopus and libvorbis encoders
[09:53:25 CEST] <cone-386> ffmpeg 03Hendrik Leppkes 07master:f5258a7d16d0: Merge commit 'e176639bcbf4b580edb462a6b0650e53cd5e3c04'
[10:44:14 CEST] <wm4> j-b: looks like I could make it to VDD on Saturday
[10:44:28 CEST] <wm4> I still need to get a flight though
[10:57:43 CEST] <wm4> j-b: let me know if I can be confirmed as participating, and I need a room as well
[11:42:50 CEST] <anshul_> While decoding Closed caption from mov, Closed Caption gets some atom header from mov demuxer
[11:44:19 CEST] <anshul_> where are all atom registered and opinion where could be the problem
[11:44:34 CEST] <j-b> wm4: of course, you are confirmed.
[11:44:39 CEST] <j-b> wm4: please mail me all the info.
[11:54:14 CEST] <wm4> j-b: what info do you need? anyway, I still need to find a flight
[11:55:20 CEST] <j-b> You need to register online. And then mail me the nights where you must have a hotel.
[11:58:04 CEST] <wm4> I should be there on saturday and sunday, right?
[11:58:11 CEST] <j-b> that'd be better, yes
[12:12:45 CEST] <cone-386> ffmpeg 03Ganesh Ajjanagadde 07master:f174bfea86b3: configure: remove unused apply() function
[12:33:10 CEST] <wm4> j-b: and when on saturday does it start / when does it end on sunday?
[13:10:16 CEST] <cone-386> ffmpeg 03Michael Niedermayer 07master:6701c92fa426: avcodec/libopusenc: Fix infinite loop on flushing after 0 input
[13:44:42 CEST] <j-b> wm4: 9h30 - around 16h00
[13:44:51 CEST] <j-b> wm4: but Sunday is more relaxed for schedule.
[13:49:24 CEST] <wm4> ok, saturday 9.30 AM to sunday 4 PM
[13:54:37 CEST] <j-b> more or less
[15:14:46 CEST] <BBB> wm4: youre coming?
[15:14:47 CEST] <BBB> \o/
[15:30:34 CEST] <wm4> maybe
[16:04:48 CEST] <BtbN> philipl, nevcairiel, as gb is aparently gone for now, i'm not expecting a review of the hevc patch anytime soon. Any objections against just pushing it?
[16:05:10 CEST] <BtbN> Or should i wait for gb, as he's listed as vaapi maintainer?
[16:31:24 CEST] <philipl> i would push it. its useful enough for the real world
[16:32:31 CEST] <philipl> and he considers the gst plugin shippable with the same conformance failures.
[16:36:01 CEST] <BtbN> I'm at a point now where i don't know what else to change or improve. It works for all samples i tested except for the SLIST ones, which are also broken in gst.
[16:36:34 CEST] <wm4> with some luck it's a hardware bug?
[16:36:45 CEST] <BtbN> Or just a bug in libva
[16:36:50 CEST] <BtbN> That's not too uncommon
[16:38:09 CEST] <BtbN> The ScalingLists ffmpeg generates are very diffrent from the ones gst produces though
[16:38:16 CEST] <BtbN> but they produce the same artifacts
[16:40:16 CEST] <nevcairiel> wm4: not hardware, it works through other APIs, but libva maybe
[16:40:53 CEST] <BtbN> You mean it works through QSV?
[16:40:56 CEST] <BtbN> Or what other API?
[16:41:38 CEST] <nevcairiel> dxva
[16:41:39 CEST] <BtbN> Oh, just noticed. Does it need a Changelog entry?
[16:41:49 CEST] <BtbN> Ah, on windows. Sure. Didn't think of that.
[16:42:02 CEST] <nevcairiel> dont we have a qsv decoder too?
[16:42:06 CEST] <nevcairiel> i suppose you could try that
[16:42:23 CEST] <BtbN> I still would have to register at Intel to download the SDK.
[16:42:44 CEST] <BtbN> Did you test in on a Cherryview GPU, or on one where it's decoded via shaders?
[16:43:32 CEST] <nevcairiel> on skylake
[16:43:47 CEST] <nevcairiel> so hardware decoding
[16:43:52 CEST] <cone-386> ffmpeg 03Carl Eugen Hoyos 07master:0382c546cc93: lavc: Describe eia_608 as text subtitles.
[16:43:54 CEST] <nevcairiel> of course it could be a different hardware than in those atoms
[16:44:01 CEST] <nevcairiel> dont have one of those handy
[16:44:05 CEST] <BtbN> No, same GPU iirc
[16:44:29 CEST] <nevcairiel> last i heard, those atoms used the broadwell gpu, but with an improved media processor
[16:44:37 CEST] <nevcairiel> so something in between broadwell and skylake
[16:44:48 CEST] <nevcairiel> In any case, I would also just push it
[16:45:00 CEST] <nevcairiel> my dxva patch had totally broken scaling lists initially =p
[16:45:18 CEST] <BtbN> It does the exact same thing dxva and vdpau do for them
[16:45:28 CEST] <BtbN> I can't imagine it's that wrong.
[16:45:38 CEST] <nevcairiel> unless it wants something else
[16:45:45 CEST] <nevcairiel> but that would be weird
[16:45:57 CEST] <BtbN> It just refers to the HEVC spec in that struct
[16:46:02 CEST] <BtbN> As does DXVA
[16:46:08 CEST] <BtbN> so i'd guess they expect the same stuff
[16:46:42 CEST] <BtbN> What about the Changelog entry? Does it need one? Can't see one for other hwaccel related changes.
[16:46:54 CEST] <nevcairiel> just add one, makes everyone happy
[16:47:05 CEST] <nevcairiel> maybe a minor bump too
[16:54:15 CEST] <BtbN> god damn it kodi. "I was inactive for an hour, shutdown the system"
[16:55:59 CEST] <philipl> nevcairiel: can you run vainfo on uour skylake box?
[16:56:11 CEST] <philipl> i want to know if it claims vp9 support
[16:56:14 CEST] <nevcairiel> its not mine really, i just had access to one for a bit
[16:56:25 CEST] <nevcairiel> (and it runs windows :D)
[16:56:47 CEST] <wm4> my desktop is a skylake box, but no idea how to make it use intel graphics (using nvidia, better drivers)
[16:56:53 CEST] <BtbN> https://bpaste.net/show/e9143aaa83f8 that's on my NUC
[16:56:59 CEST] <BtbN> wm4, you don't have to make it use intel drivers
[16:57:09 CEST] <philipl> wm4, just run vainfo
[16:57:12 CEST] <wm4> BtbN: vainfo for one dumps only the nvidia info
[16:57:20 CEST] <philipl> ah wrll
[16:57:31 CEST] <BtbN> Does the intel card show up in /dev/dri?
[16:57:41 CEST] <nevcairiel> but they do offer vp9 decoding through qsv .. and presumably they have a dxva2 device id for vp9, but microsoft didnt specify a standard for vp9, so intel uses some proprietary format, and of course they dont disclose that
[16:57:44 CEST] <wm4> there's a card0, nothing else
[16:57:56 CEST] <philipl> you need x connected to the driver for vainfo to work
[16:58:02 CEST] <wm4> philipl: yep
[16:58:02 CEST] <BtbN> you don't
[16:58:10 CEST] <BtbN> It works without X
[16:58:15 CEST] <wm4> using DRM?
[16:58:16 CEST] <philipl> didnt work headless for me
[16:58:17 CEST] <BtbN> yes
[16:58:26 CEST] <BtbN> works on my server, which has no X installed
[16:58:30 CEST] <wm4> then I probably need to force vainfo to load the correct driver
[16:58:41 CEST] <nevcairiel> you may need to turn the GPU on in the BIOS
[16:58:59 CEST] <nevcairiel> something like multi-monitor support with dedicated GPU or something
[16:59:07 CEST] <nevcairiel> otherwise it just turns it off when a dedicated GPU is installed
[16:59:16 CEST] <nevcairiel> (may depend on motherboard)
[16:59:20 CEST] <wm4> running vainfo on the terminal complains it can't connect to X
[16:59:32 CEST] <wm4> (I mean, on a console where DISPLAY is not set)
[16:59:42 CEST] <BtbN> It does so for me, too
[16:59:45 CEST] <BtbN> but then lists the info anyway
[16:59:49 CEST] <BtbN> because of drm fallback
[16:59:49 CEST] <nevcairiel> heh
[16:59:58 CEST] <cone-386> ffmpeg 03Ganesh Ajjanagadde 07master:bfe525e6328e: configure: warn if GCC 4.2 is being used
[17:00:16 CEST] <wm4> I've actually run vainfo on intel drivers once
[17:00:20 CEST] <wm4> maybe I can find it again
[17:00:40 CEST] <BtbN> It's some environment var to force the intel driver
[17:00:58 CEST] <nevcairiel> I wish i knew someone related to windows intel driver/media thingys, then I could pester them to release the vp9 dxva2 structures they are using
[17:01:33 CEST] <wm4> found it: http://sprunge.us/ZTLG
[17:01:38 CEST] <wm4> it can do VP8 trollol
[17:02:10 CEST] <philipl> oh yeah, no one addwd vp9 to vainfo
[17:02:11 CEST] <nevcairiel> someone also asked me about (M)JPEG decoding through DXVA
[17:02:23 CEST] <nevcairiel> but again, only intel proprietary format, no official docs :(
[17:02:23 CEST] <philipl> so that proves nithing
[17:02:25 CEST] <philipl> gah
[17:02:34 CEST] <wm4> apparently web browsers can make good use of jpeg hw decoding
[17:02:34 CEST] <BtbN> Does ffmpeg support that through libva?
[17:02:44 CEST] <philipl> no.
[17:02:52 CEST] <BtbN> That shouldn't be too hard to add
[17:02:53 CEST] <philipl> btbn: add mjpeg!
[17:03:05 CEST] <BtbN> jpeg is fairly simple, compared to hevc
[17:03:12 CEST] <philipl> just a bit
[17:03:56 CEST] <philipl> my cunning plan is to get support for all vaapi codecs to make nvidia add them to vdpau to keep up
[17:04:11 CEST] <nevcairiel> does nvidias hardware do those
[17:04:16 CEST] <nevcairiel> otherwise they seem unlikely to be interested
[17:04:40 CEST] <philipl> they do
[17:05:32 CEST] <nevcairiel> its a shame that Microsoft is unlikely to specify them officially through dxva
[17:05:51 CEST] <wm4> btw. many vdpau cards can do many of those DIVX profiles...
[17:06:44 CEST] <cone-386> ffmpeg 03Timo Rothenpieler 07master:1dd854e10fe1: vaapi: Add hevc hwaccel support
[17:06:50 CEST] <nevcairiel> \o/
[17:07:10 CEST] <nevcairiel> show them kodi devs that are apparently working on this too!
[17:07:28 CEST] <philipl> heh. i mentioned kodi tl thwm.
[17:07:28 CEST] <wm4> those kodi devs should show up on this channel more often
[17:07:33 CEST] <wm4> instead of doing private changes
[17:07:53 CEST] <BtbN> nevcairiel, i was working on this because team kodi gave me that NUC, asking me to add vaapi hevc hwaccel to ffmpeg.
[17:08:00 CEST] <philipl> they were pleasant enough working with me.
[17:08:04 CEST] <nevcairiel> maybe that guy was just confused then
[17:08:11 CEST] <philipl> im pretty sure they arw not doing anuthing.
[17:08:30 CEST] <philipl> that guy thinks btbn is a kodi guu
[17:08:33 CEST] <philipl> guy
[17:08:39 CEST] <BtbN> hm?
[17:08:45 CEST] <philipl> fucking ipad
[17:08:49 CEST] <BtbN> which guy?
[17:08:55 CEST] <nevcairiel> one guy used that as an excuse why his tickets arent being answered, apparently the kodi devs are busy with vaapi hevc and cant
[17:09:07 CEST] <philipl> right
[17:09:09 CEST] <BtbN> the hell
[17:09:26 CEST] <wm4> he mentioned someone named FernetMenta?
[17:09:31 CEST] <wm4> or something
[17:09:44 CEST] <BtbN> Fernet and fritsch are the primary people pushing vaapi in kodi.
[17:09:45 CEST] <philipl> thats their main linux guy
[17:10:06 CEST] <BtbN> They did some amazing work. kodi plays 4K60 on my NUC.
[17:10:13 CEST] <BtbN> hevc
[17:10:30 CEST] <nevcairiel> direct rendering can do magix
[17:10:36 CEST] <philipl> with the egl insanity
[17:10:38 CEST] <nevcairiel> only took linux a couple years
[17:10:50 CEST] <BtbN> well, only took intel open source devs a couple years.
[17:10:52 CEST] <philipl> only took intel
[17:10:55 CEST] <philipl> yeah
[17:10:56 CEST] <BtbN> Their GLX implementation was butchered
[17:11:15 CEST] <BtbN> It was faster to copy to system ram and re-opload to an OpenGL texture than to use their GLX interop
[17:12:16 CEST] <wm4> nevcairiel: vdpau GL interop always worked
[17:12:32 CEST] <BtbN> yes, they have actual GL extensions handling it
[17:13:02 CEST] <wm4> I'm curious about the new EGL interop... I'll try it as soon as my drivers are probably new enough
[17:13:11 CEST] <BtbN> you need mesa-11 for it
[17:13:32 CEST] <wm4> I'm using debian, so maybe in a few years
[17:13:40 CEST] <BtbN> you can propably just add the kodi repository
[17:13:51 CEST] <BtbN> They have all the stuff needed there, for various deb based distros
[17:13:57 CEST] <nevcairiel> thats really my main gripe with linux, i need to support like 3 different API combinations to get video things going :D
[17:14:09 CEST] <philipl> BtbN: so you want to add vp8? I've got an 'in theory' patch locally to add hwaccel hooks for that and vp9.
[17:14:29 CEST] <BtbN> Yeah, mjpeg and vp8 would definitely be nice to have.
[17:14:30 CEST] <philipl> otherwise I'll do it when I finally get a skylake machine
[17:15:01 CEST] <BtbN> Can't do vp9 though, my nuc doesn't seem to support it.
[17:15:17 CEST] <nevcairiel> might be a driver or libva thing
[17:15:30 CEST] <nevcairiel> or its really only skylake
[17:15:34 CEST] <philipl> only skylake
[17:15:43 CEST] <philipl> That's what intel have said publicly.
[17:15:45 CEST] <nevcairiel> the atom nucs are somewhere in between on compat
[17:15:50 CEST] <nevcairiel> broadwell doesnt do hevc
[17:15:56 CEST] <nevcairiel> but the atom uses the broadwell gpu
[17:15:58 CEST] <nevcairiel> yet does hevc
[17:16:02 CEST] <philipl> braswell
[17:16:02 CEST] <wm4> wait, skylake has vp9?
[17:16:17 CEST] <philipl> skylake has hybrid vp9. uses shaders for part of it
[17:16:48 CEST] <nevcairiel> codec support: http://abload.de/img/5a4q2x.png
[17:16:51 CEST] <BtbN> If you have the 3000$ Intel Media SDK, it comes with a libva driver that does hevc on ivy bridge or better.
[17:17:21 CEST] <philipl> wm4. You need to patch vainfo to try and detect vp9, then we'll know for sure.
[17:17:50 CEST] <BtbN> Wait, it doesn't check for that?
[17:17:54 CEST] <wm4> just tell me what I should try, and I'll do it
[17:18:43 CEST] <BtbN> vainfo already seems to check for VAProfileVP9Profile0
[17:18:57 CEST] <philipl> Hmm. In a released version?
[17:19:06 CEST] <BtbN> just looked at git master
[17:19:54 CEST] <philipl> 14 daysAdd VP9 profile to vainfoHEADmaster
[17:19:59 CEST] <philipl> Yeah, it's the last change made in git.
[17:20:06 CEST] <BtbN> It seems to be rather generic. If the driver reports a profile vainfo doesn't know, it would print <unknown profile>
[17:20:07 CEST] <philipl> wm4: so check it out and build it, then try again
[17:20:19 CEST] <philipl> I WILL KEEP HOPING
[17:28:02 CEST] <cone-386> ffmpeg 03lummax 07master:0c800b27611c: avcodec: Assert on codec->encode2 in encode_audio2
[17:28:37 CEST] <wm4> debian must be doing something stupid
[17:28:56 CEST] <wm4> e.g. the vdpau driver is in /usr/lib/x86_64-linux-gnu/dri/nvidia_drv_video.so
[17:29:05 CEST] <wm4> while vainfo expects /usr/lib/dri/...
[17:30:05 CEST] <iive> that's their multiarch setup
[17:30:15 CEST] <wm4> and it's retarded
[17:30:39 CEST] <wm4> no purpose and actually causes tons of problems
[17:31:13 CEST] <iive> imho, it is better that lib64 and lib32a
[17:31:20 CEST] <wm4> ok, LIBVA_DRIVERS_PATH
[17:31:57 CEST] <j-b> BBB: ping
[17:32:04 CEST] <BBB> j-b: pong
[17:38:21 CEST] <wm4> meh, the va tools helper code has code for selecting a display type per command line, but vainfo doesn't use it
[17:39:00 CEST] <wm4> ok, via DRM it apparently just doesn't work
[17:39:19 CEST] <wm4> libva info: va_getDriverName() returns -1
[17:39:19 CEST] <wm4> libva error: va_getDriverName() failed with unknown libva error,driver_name=(null)
[17:39:19 CEST] <wm4> vaInitialize failed with error code -1 (unknown libva error),exit
[17:55:33 CEST] <philipl> wm4: intel don't want us to know.
[18:03:09 CEST] <wm4> either I really have to enable it in the BIOS, or I have to load/instal some obscure driver?
[18:05:24 CEST] <philipl> it has to be enabled in the bios for sure.
[18:06:04 CEST] <wm4> shouldn't Linux be able to use the hw anyway?
[18:06:25 CEST] <wm4> I don't think the BIOS allows enabling both
[18:06:44 CEST] <wm4> so I'd have to reboot Linux, which I don't want to
[18:06:50 CEST] <wm4> (lots of open windows etc.)
[18:08:01 CEST] <philipl> you can enable both but it'll be messy. 
[18:08:16 CEST] <wm4> messy?
[18:08:18 CEST] <philipl> don't worry, ill continue to live in ignorance.
[18:08:44 CEST] <wm4> well I can do it the next time I want to reboot, and maybe there are some other skylake users around
[18:08:46 CEST] <philipl> will x come up right etc
[18:08:51 CEST] <wm4> or maybe I could boot a live CD
[18:09:16 CEST] <philipl> live cd would be best but need to move monitor
[18:09:47 CEST] <wm4> the main problem with the live cd is that I'd need to compile vainfo
[18:10:07 CEST] <philipl> true
[18:10:29 CEST] <philipl> might also need to build up to date va driver too
[18:27:45 CEST] <cone-386> ffmpeg 03Stefano Sabatini 07master:57cd2f7777a3: lavf/mpegenc: provide better feedback in case of invalid media type
[20:01:53 CEST] <philipl> BtbN: __gb__ didn't take the bait on the TODO items. Guess he wants someone else to work it out :-)
[20:02:58 CEST] <philipl> http://www.phoronix.com/scan.php?page=news_item&px=Nouveau-VAAPI-Samsung-Patches
[20:03:01 CEST] <philipl> THe circle is complete!
[20:31:36 CEST] <wm4> philipl: the fuck
[20:33:13 CEST] <wm4> why would they be interested in this? makes no sense to me
[20:40:03 CEST] <wm4> ubitux: there's no way yet to distinguish subs with 0 and unknown length yet, correct?
[21:48:41 CEST] <philipl> wm4: boggles my mind. Why vaapi when vdpau works. Why Samsung. So many questions.
[21:50:10 CEST] <Compn> samsung sponsored some stuff for ffmpeg , maybe opw? 
[21:50:17 CEST] <Compn> so they are our friends 
[21:50:18 CEST] <Compn> :P
[21:54:32 CEST] <michaelni> philipl, if you have questions related to samsung, then ask reynaldo
[21:56:21 CEST] <Daemon404> IME people working for bigcorps rarely know why another division has done something
[22:00:51 CEST] <cone-386> ffmpeg 03Paul B Mahol 07master:15f4b3db58a6: avfilter: add framerate video filter
[22:03:10 CEST] <michaelni> true but they should know who to contact or ask at least if there is interrest
[22:05:24 CEST] <llogan> durandal_1707: looks like an interesting one
[22:06:53 CEST] <durandal_1707> its buggy...
[22:33:23 CEST] <philipl> Just seems highly masochistic. nouveau has working vdpau - so why would anyone want to fight with vaapi.
[22:33:38 CEST] <Daemon404> philipl, because a manager told you to
[22:33:46 CEST] <Daemon404> looks good at trade shows
[22:33:47 CEST] <Daemon404> who knows.
[22:33:56 CEST] <philipl> I guess if your applications are all based on gstreamer you might, as there's no vdpau worth a damn in gstreamer.
[22:34:16 CEST] <philipl> but I'd rather write a vdpau plugin for gstreamer than add vaapi to a driver.
[22:35:39 CEST] <nevcairiel> Does media capability and OpenGL rendering even work good enough in nouveau to care?
[22:36:22 CEST] <nevcairiel> I haven't used it for a year or so now, but it used to be quite lacking
[22:36:56 CEST] <Daemon404> well also
[22:37:01 CEST] <philipl> My perceptions are mostly historical, when it was unusable. I believe it can be used in more situations now, but why is the eternal question.
[22:37:07 CEST] <Daemon404> i know 0 people using nouveau, not including MUH FREEDUMS types
[22:37:19 CEST] <philipl> quite.
[22:37:40 CEST] <nevcairiel> Presumably nvidia is even working on KMS now for the binary blob
[22:38:07 CEST] <philipl> Possibly KMS. They were trying to push that EGL Streams thing as an alternative for wayland.
[22:38:18 CEST] <philipl> Last I heard they'd given up and were working on KMS
[22:38:34 CEST] <wm4> nevcairiel: nouveau can destroy your hardware physically
[22:38:46 CEST] <wm4> (power management is hard)
[22:38:47 CEST] <philipl> but it was free before it died.
[22:39:01 CEST] <nevcairiel> If you listen to the internet, so can the windows driver on certain laptops
[22:39:30 CEST] <nevcairiel> Technically it destroys the screen there, same diff
[23:04:52 CEST] <cone-386> ffmpeg 03Ganesh Ajjanagadde 07master:9ba511d5d3d7: avfilter/af_amerge: use the name 's' for the pointer to the private context
[00:00:00 CEST] --- Fri Aug 28 2015


More information about the Ffmpeg-devel-irc mailing list