[Ffmpeg-devel-irc] ffmpeg-devel.log.20150128

[burek]

[00:01] <michaelni> llogan, added for ffmpeg-trac
[01:18] <llogan> michaelni: thanks
[01:27] <cone-693> ffmpeg.git 03Michael Niedermayer 07master:840570d27b71: avfilter/vf_mcdeint: avoid uninitilaized fields in AVPacket
[05:41] <cone-454> ffmpeg.git 03Carl Eugen Hoyos 07master:449b8cf04b7e: avformat/mxfdec: Fix cleanup in mxf_free_metadataset()
[06:09] <cone-454> ffmpeg.git 03Mark Reid 07master:a181169868fe: tests/fate: added mxf tests for essencegroups and missing index
[06:28] <cone-454> ffmpeg.git 03Timo Rothenpieler 07master:fb34c580bb24: avcodec/nvenc: Handle non-square pixel aspect ratios
[10:03] <cone-110> ffmpeg.git 03Maneesh Gupta 07master:1600f85cbc45: doc/fftools-common-opts: update/extend documentation for -opencl_bench option
[11:40] <durandal_1707> is anybody here with latest mplayer binary?
[11:53] <wm4> durandal_1707: just build it
[11:53] <durandal_1707> i tried and failed
[11:53] <wm4> not many distros package an uptodate mplayer
[11:53] <wm4> failed how?
[11:54] <durandal_1707> in linkage, missing symbols
[11:54] <wm4> just remember to check it out with svn
[11:54] <wm4> if it's a build regression, try with an earlier ffmpeg, or poke michaelni or reimar
[11:54] <durandal_1707> nope, this is some local nonsense
[11:56] <durandal_1707> iive: do you have mpeg2 that is soft-telecined?
[11:58] <kierank> iirc it's possible to make with x262
[11:58] <kierank> can't remember though if that was implemented
[11:58] <kierank> tbh I don't even understand what the soft pulldown filter is even trying to achieve
[11:59] <nevcairiel> i thought its meant to convert to hard-telecine
[11:59] <nevcairiel> but i may be wrong
[11:59] <kierank> so it's a pulldown filter
[11:59] <durandal_1707> yes it changes soft-telecined material into hard-telecined 
[11:59] <kierank> soft pulldown means the pulldown is just flagged
[12:00] <kierank> then the filter name is bizzare
[12:00] <kierank> why not just call it pulldown
[12:01] <kierank> and also pts needs to change and afaik pts doesn't
[12:01] <wm4> mplayer2 had changes to correct the pts
[12:02] <wm4> in vf_softpulldown
[12:02] <kierank> the filter should therefore fail on anything not 24000/1001
[12:15] <nevcairiel> wm4: i wonder how this osx guy wants to build a multimedia api without ffmpeg behind it ... its not like he has any alternative multimedia frameworks to pick from :p
[12:15] <durandal_1707> gstreamer
[12:15] <j-b> VDA :)
[12:15] <nevcairiel> without the ffmpeg plugins, gstreamers support is minimal
[12:16] <j-b> webm+ogg!
[12:19] <wm4> nevcairiel: didn't he say he wants to use ffmpeg, just with minimizing contact with it?
[12:21] <rcombs> nevcairiel: QuickTime!
[12:22] <durandal_1707> why there is no wrapper around x262 in lavc?
[12:23] <BtbN> 262?
[12:24] <av500> x261 ftw!
[12:28] <durandal_1707> x262 have --pulldown flag
[12:28] <ubitux> wtf @ #2078
[12:28] <ubitux> people though they are on github or what
[12:31] <BtbN> It's... a dog oO
[12:36] <durandal_1707> x262 --mpeg2 --pulldown 22 -o /tmp/o.mpeg ~/matrix.mpg: simply crashes
[12:44] <wm4> ubitux: lol... it's always funny if some poor project gets bombed with retards in an issue because it got linked all over the internet
[12:44] <wm4> social coding etc.
[12:55] <kierank> wm4: lol
[12:55] <kierank> durandal_1707: might not be implemented then
[13:00] <compn> durandal_1707 : ask dalias if he has any samples for soft pulldown
[13:00] <compn> he worked on those filters iirc
[13:00] <kierank> why do you need samples
[13:01] <kierank> unless you want to ivtc and then pulldown back and compare?
[13:01] <durandal_1707> to test filter with actual file that have this flags
[13:01] <compn> kierank : to see if durandal_1707 ported it correctly i guess :P
[13:01] <durandal_1707> i believe that mp=softpulldown default behavior is wrong
[13:02] <kierank> oh i see
[13:04] <ubitux> wm4: thanks for the "social coding" words, that's what i was looking for
[13:05] <BtbN> Well, it gets fun when someone complains that your code using master/slave terminology is offensive.
[13:05] <kierank> FFmpeg-gate
[13:05] <compn> oh man that master/slave debate, classic
[13:06] <BtbN> There actualy is a new, politicaly correct, form of master/slave, but i forgot what it was.
[13:06] <nevcairiel> it sounds stupid though, tech terms should need to be "politically correct"
[13:07] <nevcairiel> should not*
[13:08] <BtbN> It _is_ stupid.
[13:09] <ubitux> slave mode of mplayer should be renamed apartheid mode to avoid any confusion
[13:09] <compn> trigger warning!
[13:09] <compn> :P
[13:32] <iive> durandal_1707: i think i have, but right now i'm busy, ask me again in few hours
[13:38] <JoshX> Hello, I made a little addiotion to libavformat/utils.c
[13:38] <JoshX> so that the filename pattern could also use %t to add Ymd_HMSfff 
[13:39] <JoshX> useful for segmentation of files with start timestamp in the filename
[13:39] <JoshX> now, i'm fairly new to github aqnd have never 'really' contributed or tried to get a simple add to a big opensource project
[13:40] <JoshX> but my guess is, this could be quite useful for more people so it might be nice to merge it in a way
[13:40] <JoshX> questing really is, 'how' :-)
[13:46] <kierank> JoshX: use git format-patch and send a patch to ffmpeg-devel at ffmpeg.org
[13:50] <JoshX> ok lets see if i can get that to work :)
[13:53] <cone-110> ffmpeg.git 03Michael Niedermayer 07master:ccaa5dcb3135: avcodec/h264_parser: Rename close()
[13:54] <cone-110> ffmpeg.git 03Michael Niedermayer 07master:0d37ca150c5e: avutil/aes: Rename crypt()
[13:55] <compn> JoshX : or even a diff -u of the two files
[13:55] <compn> if you know the command line diff program 
[13:57] <wm4> compn: no, shut the fuck up
[13:58] <wm4> git format-patch is more convenient for everyone, don't go around and suggest lesser, ahrder methods
[13:58] <wm4> *harder
[13:59] <arwa_> Hey, When I am using fspp for a paricular image, I am getting a green patch in the bottom.
[14:00] <arwa_> Can anyone tell me what can be the possible reason?
[14:00] <durandal11707> what is size of image?
[14:03] <JoshX> kierank, compn, no found the howto :-) branched it, changed it, committed it and created the patch :)
[14:03] <JoshX> have to learn git anyway
[14:04] <arwa_> It is 480X260
[14:04] <wm4> and once you know it, writing patches will be much easier
[14:04] <JoshX> wm4: agreed, that's why i did it the 'right' way :)
[14:05] <JoshX> ah it's held on the list since i'm not a member :)
[14:07] <JoshX> i understand.. I'm trying to make a dvr-like sollution to save a h264 mp4 stream from an ip camera to 900 second chunks.. i don't really want to transcode anything since the format and bitrate of the camera are good for now.. so I try to use 'copy' as vcodec and -an (no audio) but i get a lot of warnings about non-monotonous dts frames..
[14:08] <compn> JoshX : accepted your mail from the mod queue
[14:08] <compn> should be posted now.
[14:08] <wm4> JoshX: subscribing would make things easier
[14:08] <JoshX> cool thanks :)
[14:08] <JoshX> yes.. just mail with subscribe as subject to the list?\
[14:09] <durandal11707> arwa_: can't reproduce with ffplay, does it depends on specific image?
[14:09] <JoshX> ah found the subscribe page, and subscribed
[14:09] <wm4> JoshX: don't know if this works; also see http://www.ffmpeg.org/mailman/listinfo/ffmpeg-devel
[14:10] <JoshX> wm4: google beat you to it ;)
[14:10] <arwa_> Yes, because, I tried it with different images, but for one specific image I was getting this.
[14:10] <compn> wm4 : btw there is vf_overlay in mplayer fork i forgot about ;D
[14:10] Action: compn runs and hides
[14:10] <wm4> compn: which fork?
[14:11] <compn> humm
[14:12] <compn> no idea
[14:12] <compn> http://permalink.gmane.org/gmane.comp.video.mplayer.devel/55122
[14:13] <compn> another vf patch , http://onebithq.com/root/mplayer/videomixer
[14:13] <compn> shared mem filter :P
[14:13] <wm4> that looks like a rejected patch
[14:13] <wm4> and also useless
[14:13] <compn> yep
[14:14] <wm4> it mixes memalign and av_freep so it's buggy too
[14:15] <wm4> compn: what was the conclusion of this thread?
[14:18] <JoshX> oh my patch fails... :-(( going to fix it
[14:23] <wm4> JoshX: it also has a buffer overflow and undefined behavior
[14:23] <wm4> replying on the list...
[14:25] <wm4> done
[14:25] <ubitux> arwa_: can you share the image?
[14:26] <arwa_> Okay.
[14:28] <JoshX> wm4 ok.. i'm not a real c developer, but it worked for me.. now the build fails, trying to figure out what i've done different
[14:28] <JoshX> can i pm you?
[14:28] <arwa_> http://imgur.com/KwJhGny
[14:29] <wm4> sure...
[14:33] <ubitux> arwa_: asserts for me here
[14:33] <ubitux> Assertion frame->height == link->h failed at libavfilter/avfilter.c:1170
[14:34] <arwa_> Okay, I am not getting any error!
[14:34] <ubitux> that's normal, i have a higher assert level
[14:34] <arwa_> And why is there an assertion failure?
[14:34] <ubitux> basically the frame your sending doesn't match the configuration of your outlink
[14:34] <arwa_> So, what should I do?
[14:36] <ubitux> arwa_: you need to fix the output frame dimension
[14:36] <ubitux> arwa_: fspp is requesting a width and height 8 aligned buffer
[14:36] <ubitux> but that doesn't match the real dimension, is just to avoid overread/overwrite
[14:37] <ubitux> so after getting the output frame, you need to reset them to the expected dimension
[14:37] <ubitux> look how it's done in vf_pp
[14:37] <ubitux> (see lines 129 to 136)
[14:37] <arwa_> Okay, thanks :)
[14:37] <ubitux> also make sure you haven't made the same mistake in the other pp based filters
[14:37] <arwa_> So, I need to fix the code?
[14:37] <ubitux> yes
[14:38] <ubitux> pp7 has the same problem
[14:38] <arwa_> Yes, it is giving the same kind of results for pp7 also.
[14:38] <ubitux> please send a patch for both of them
[14:38] <arwa_> I will write a patch correcting it.
[14:38] <ubitux> (1 patch for each filter please)
[14:38] <ubitux> thank you
[14:39] <arwa_> Okay :)
[14:40] <ubitux> btw, you might want to factorize that code between all the pp filters
[14:40] <ubitux> same for the dither tables
[14:41] <arwa_> by factorize, do you mean including it in internal.h?
[14:41] <ubitux> the prototype maybe, but not the code
[14:42] <ubitux> not sure where it could belongs, maybe a libavfilter/utils.c, dunno
[14:42] <ubitux> or simply avfilter.c
[14:42] <JoshX> ok building now with the fixes you suggesed in place, wm4 :)
[14:42] <JoshX> *suggested
[14:43] <ubitux> arwa_: you understand the issue about the aligned thing? (not how to fix it, but more the meaning of this) or you want more explanations?
[14:43] <arwa_> I would like to know more
[14:44] <arwa_> I didn't exactly understand.
[14:44] <ubitux> ok so, most pp filters (maybe all?) code needs to work by block of 8x8 minimum
[14:44] <ubitux> to avoid some boundary checks (they just process the whole 8x8 block
[14:45] <ubitux> so when the frame doesn't match that "grid" (because of non multiple of 8 dimensions) there is no special code
[14:45] <ubitux> but we still want to output a frame that is not a multiple of 8
[14:45] <arwa_> Oh, okay.
[14:45] <ubitux> we just want to buffer to have a sufficient size to avoid writing or reading out of memory
[14:46] <ubitux> so basically we request a frame in larger dimensions
[14:46] <ubitux> and then actually fix the dimension to match the one frome the input, hiding these stashing areas
[14:46] <arwa_> Oh, nice!!
[14:46] <ubitux> and lavfi asserted here saying that the out link was configured to match the exact same dimension as the input
[14:46] <ubitux> but you were sending frame that were larger
[14:47] <ubitux> (because aligned to the grid)
[14:47] <ubitux> anyway, hope that clears things up
[14:47] <wm4> ubitux: does lavfi have any magic mechanism to request padded frames from preceding filters?
[14:48] <ubitux> not sure what you mean :p
[14:48] <wm4> well if a filter "requests" a larger frame than the input size, what exactly happens?
[14:49] <arwa_> Thank you. It is giving me an overall idea of what is happening.
[14:50] <ubitux> wm4: i wonder, maybe there is a pool somewhere so that frame is reused or something :)
[14:50] <ubitux> but i don't know
[14:51] <ubitux> /* TODO: set the buffer's priv member to a context structure for the whole
[14:51] <ubitux>  * filter chain.  This will allow for a buffer pool instead of the constant
[14:51] <ubitux>  * alloc & free cycle currently implemented. */
[14:51] <ubitux> seems not :(
[14:51] <wm4> heh
[14:51] <ubitux> i thought we had some at one point
[14:52] <wm4> there's one in lavc
[14:52] <ubitux> it probably died as a casuality of the evil plan or something
[14:53] <ubitux> yeah michael added one in lavfi 4 years ago
[14:54] <ubitux> but it disappeared somehow, too bad
[14:54] <cone-110> ffmpeg.git 03Paul B Mahol 07master:87577f5508ca: avfilter/vf_telecine: use the name 's' for the pointer to the private context
[14:57] <ubitux> lol we still have that unused pool structure
[14:58] <ubitux> would be nice to have a pool of 4 or 8 frames
[14:59] <wm4> ideally there'd be a pool that can be used by both lavfi and lavc and API users
[15:00] <ubitux> there is a lavu/AVBufferPool could be used
[15:00] <ubitux> but the FramePool itself is local to avcodec afact
[15:00] <wm4> yeah
[15:00] <wm4> it's messed directly into utils.c
[15:00] <wm4> (who needs code reuse)
[15:01] <ubitux> could be an interesting work to factorize them
[15:01] <ubitux> but not my priority right now, we might want to open a ticket or something
[15:14] <JoshX> hmm could it be that what i am trying to do (and did against a source from the beginning of december) is already in the codebase now?? 
[15:15] <nevcairiel> anything is possible. what are you trying to do?
[15:16] <JoshX> i'm trying to make filenames with the timestamp to the millisecond using the segement feature
[15:16] <JoshX> and it only did %d before
[15:16] <nevcairiel> wasnt strftime support added recently
[15:16] <JoshX> but now there seems to be a strftime feature
[15:16] <JoshX> ghe possibly :)
[15:16] <JoshX> it would fix my problem ;)
[15:17] <nevcairiel> i guess the answer is yes then
[15:17] <JoshX> haha
[15:17] <wm4> yeah, always check git master before trying to fix or add something
[15:17] <JoshX>     { "strftime",          "set filename expansion with strftime at segment creation", OFFSET(use_strftime), AV_OPT_TYPE_INT, {.i64 = 0 }, 0, 1, E },
[15:17] <JoshX> just found this debugging my code ;)
[15:17] <JoshX> well my idea wasn't half bad then ;)
[15:17] <JoshX> now to see how to use it then ;)
[15:18] <ubitux> it was added in 2013
[15:19] <wm4> that feature sucks a bit
[15:20] <wm4> you can only have a strftime format, or ffmpeg's own one
[15:20] <JoshX> its not in the examples, manpage, help, whatever??
[15:20] <Daemon404> i want a feature to disable filename expansion
[15:20] <wm4> can't mix them
[15:20] <Daemon404> i should send a patch
[15:20] <Daemon404> because our filename expansion is applied to urls
[15:20] <Daemon404> and complete breaks escaped urls
[15:20] <wm4> lol
[15:20] <wm4> maybe it shouldn't be applied to URLs
[15:20] <Daemon404> im sure some person will claim to use it
[15:21] <ubitux> ah i was actually mistaking it for the strftime img2 thing
[15:21] <wm4> no, someone will claim that someone else might use it
[15:21] <wm4> (ffmpeg business as usual)
[15:21] <Daemon404> yep
[15:21] <Daemon404> theoretical "users"
[15:21] <wm4> every feature is sacred, every feature is good
[15:22] <nevcairiel> speaking of segment, i've had a rather random failure with it, where the webserver serving the segments and playlist would return an empty playlist .. i think its a race caused by the code just re-writing the playlist in that instant ...
[15:22] <nevcairiel> didnt I see a function in some other muxer to write to a tmp file and then replace the original in one operation after
[15:22] <nevcairiel> i should copy that
[15:24] <JoshX> ah.. my main problem now is that strftime does not provide milliseconds...
[15:25] <Daemon404> microseconds is not enough?
[15:26] <Daemon404> is you file like 9 billion fps?
[15:26] <JoshX> haha
[15:26] <JoshX> nope
[15:26] <nevcairiel> micro is actually smaller than milli Daemon404 :p
[15:26] <Daemon404> the sad part is that i grew up in a fully metric and SI country
[15:26] <Daemon404> and forgot that
[15:26] <JoshX> what would be the strftime pattern for YYYYmmdd_HHmmssfff?
[15:26] <nevcairiel> shouldnt %f be milli
[15:26] <JoshX> it it in a lot of languages
[15:26] <JoshX> lets see
[15:27] <Daemon404> yeah it could be MS
[15:27] <Daemon404> %f is micro, so you can round
[15:28] <JoshX> 20150128_152740%f.mp4
[15:28] <Daemon404> hmm
[15:28] <JoshX> well.. 
[15:28] <Daemon404> pubs.opengroup.org/onlinepubs/009695399/functions/strftime.html
[15:28] <Daemon404> POSIX strftime has no micro
[15:28] <JoshX> indeed
[15:28] <wm4> linux4lyfe http://linux.die.net/man/3/strftime
[15:28] <JoshX> you need to use gettimeofday
[15:28] <wm4> MORE FEATURES
[15:28] <nevcairiel> are you really going to produce more than one segment per second so that milliseconds are required
[15:29] <JoshX> i produce mp4 video which need to be synced to output of another system so the starttime must be ms-based
[15:29] <JoshX> and then i count frames to get to the exact position i need to be
[15:29] <nevcairiel> and you are syncing over the filename? that seems weird
[15:30] <JoshX> can i get the timecode to the ms per frame in mp4 then?
[15:31] <JoshX> it would be much better for me if i could put the exact timecode per frame..
[15:32] <JoshX> either in a seperate index file or embedded in the mp4
[15:32] <JoshX> i thought about abusing a subtitle field for that..
[15:33] <JoshX> nevcairiel: any idea?
[15:43] <cone-110> ffmpeg.git 03Paul B Mahol 07master:231bf4d1c21c: avfilter/vf_colorbalance: use the name 's' for the pointer to the private context
[15:43] <cone-110> ffmpeg.git 03Paul B Mahol 07master:abcf8610b8b3: avfitler/vf_il: use the name 's' for the pointer to the private context
[15:43] <cone-110> ffmpeg.git 03Paul B Mahol 07master:f27cda48ec64: avfilter/vf_extractplanes: use the name 's' for the pointer to the private context
[15:43] <cone-110> ffmpeg.git 03Paul B Mahol 07master:6121968a9e87: avfilter/af_biquads: use the name 's' for the pointer to the private context
[15:43] <cone-110> ffmpeg.git 03Paul B Mahol 07master:67aaa7056d2b: avfilter/avf_avectorscope: use the name 's' for the pointer to the private context
[15:43] <cone-110> ffmpeg.git 03Paul B Mahol 07master:fca50464dc7f: avfilter/vf_colorchannelmixer: use the name 's' for the pointer to the private context
[15:44] <ubitux> haha
[15:44] <ubitux> i can drop my most-boring-thing-ever-and-i-dont-want-to-finish-this-shit i guess now
[15:44] <ubitux> +branch
[15:45] <durandal_1707> there are still some filters left
[15:45] <durandal_1707> i keep mistyping filter
[15:46] <ubitux> if you're into a rename rampage in lavfi, you could rename the "picref" into frames
[15:46] <saste> what about doing useful work instead?
[15:46] <ubitux> :D
[15:46] <ubitux> yeah, if you want to do useful stuff, you can add the frame pool we talk about today
[15:47] <durandal_1707> that is too hard
[15:47] <ubitux> really?
[15:47] <ubitux> it needs a bit of work but it doesn't look that hard
[15:48] <ubitux> i wonder how much the pool mecanism helps in comparison to modern malloc implementation
[15:57] <wm4> ubitux: there's also some hidden cost, because these large allocations will go through mmap, which means the kernel has to zero pages in the background
[15:57] <wm4> but all in all it probably doesn't matter that much
[15:59] <durandal_1707> michaelni: i added av_frame_make_writable() right after cloning and it doesn't help
[16:00] <wm4> no, you need to call it right before you write to a frame
[16:01] <durandal_1707> huh, why?
[16:03] <wm4> because that's how reference counting works
[16:03] <durandal_1707> ahh it WORKS now
[16:58] <JoshX> anyone who can help me with the timecode thing? i need a way to reproduce the time of the first frame to the ms when using the segment filter to be able to jump to a frame in a video file based on a timestamp to the ms
[16:59] <ubitux> durandal_1707: .needs_writable? or it's not necessary all the time?
[16:59] <JoshX> i'm saving a h264 stream from an ip camera in 15 minut chunks and i just need the timecode of the first frame (or the timecode of any frame) or an index file with frametimes/numbers
[16:59] <JoshX> i can live with any of the 3 :)
[17:13] <ubitux> hey, the void & cluster ordered ("blue noise") dithering is kind of interesting
[17:14] <ubitux> i'm probably going to add this one instead of the ordered one
[17:15] <anshul_mahe> ubitux: are u intrested to check that c608 patch, I was planning to move on at the patch by sam
[17:17] <ubitux> sam?
[17:17] <ubitux> you mean shan?
[17:17] <anshul_> yes, I forgot his name
[17:17] <wm4> ubitux: mpv does use something related to void & cluster (although it's static)
[17:18] <ubitux> yes that's the point :)
[17:18] <ubitux> error diffusion is a nightmare for gif encoding, or still frames in general
[17:18] <ubitux> so i need to have a good static one
[17:18] <ubitux> (assuming that's what you meant by static)
[17:19] <ubitux> anshul_: i don't plan to do anything, but i'm happy to review some patches i guess
[17:19] <ubitux> anshul_: also, do we have a FATE test for CC or not yet?
[17:19] <anshul_> no should I give it priority over shan patch
[17:20] <ubitux> because in addition to improve the coverage, it will also show how the markup evolve :)
[17:20] <anshul_> ok, So I would make one fate test first
[17:21] <ubitux> yeah, ideally
[17:21] <anshul_> do we have fate test for other subtitles, I would take one for refrence fate test
[17:21] <ubitux> yes, almost all of them
[17:21] <ubitux> anshul_: look at 6dc99fdf0e4328ca2992b7bfcd139f7b1b636bf8
[17:22] <ubitux> it includes a test
[17:24] <durandal_1707> ubitux: if i got i right it should not be necessary all the time, like when repeat_pict is used
[17:24] <cone-51> ffmpeg.git 03Arwa Arif 07master:622936424c69: avfilter/vf_fspp: Fix frame-alignment in FSPP
[17:24] <cone-51> ffmpeg.git 03Michael Niedermayer 07master:375a0273cec4: avfilter/vf_fspp: check count before calling row_idct()
[17:24] <cone-51> ffmpeg.git 03Michael Niedermayer 07master:a6f9a5d0f635: avfilter/x86/vf_fspp: Fix loop condition for column_fidct()
[17:24] <ubitux> durandal_1707: ok
[17:25] <anshul_> ubitux: thanks
[17:25] <ubitux> anshul_: make fate-foobar GEN=1 is your friend
[17:25] <ubitux> make sure you select a sample with advanced "markup"
[17:25] <ubitux> and make sure the dependencies for that sample are correct
[19:06] <cone-51> ffmpeg.git 03Paul B Mahol 07master:c099783316cf: avformat/nsvdec: remove case which is no longer possible
[20:35] <cone-51> ffmpeg.git 03Clay McClure 07master:6a808f5ae17f: libdc1394: Add support for MONO8 (gray) video mode
[20:35] <cone-51> ffmpeg.git 03Michael Niedermayer 07master:37984ca13334: Merge commit '6a808f5ae17f1fcdbcfb18055872c12aef70ffff'
[21:03] <llogan> the xcb x11grab -y avoption seems to conflict with the -y global option. can someone confirm?
[21:04] <llogan> ffmpeg -y -f x11grab -x 100 -y 50 -i :0.0 outptu.mp4
[21:04] <llogan> console "outptu" then squawks ":0.0: Protocol not found"
[21:06] <llogan> does the same w/o -y flobal
[21:06] <llogan> *global
[21:16] <cone-51> ffmpeg.git 03Luca Barbato 07master:3c18a7b18807: avio: Do not consider the end-of-buffer position valid
[21:16] <cone-51> ffmpeg.git 03Michael Niedermayer 07master:0b4fc4bacd86: Merge commit '3c18a7b18807de81566381a1bcbe9f6103c0296b'
[21:16] <cone-51> ffmpeg.git 03Michael Niedermayer 07master:58ed1528574f: avformat/aviobuf: Allow seeking to the end of the buffer for read only buffers
[21:33] <michaelni> llogan, you can pass x/y in the filename +%d,%d style
[21:34] <llogan> i know, but the -x and -y are listed as AVoptions and that may confuse users who will attempt it that way
[21:45] <michaelni> llogan, indeed, feel free to remove or document, whatever you think is better
[21:52] <iive> :O
[21:55] <llogan> michaelni: i'd prefer that the AVoption does not get confused with the global one (and keep it documented) but i don't know how to do that.
[21:56] <llogan> for now i just didn't include -x and -y in indevs.texi
[22:24] <ubitux> arwa: it might be interesting to add the reference image as well
[22:25] <ubitux> (because sometimes we wonder what we would ideally get from that square garbage :D)
[22:25] <arwa> Yes, I have added the reference image.
[22:25] <ubitux> ah, cool :)
[22:25] <arwa> I have one ques about pp filter.
[22:26] <arwa> I am confused on what options are offered by the filter.
[22:26] <arwa> I read the documentation, but I am not able to change the parameters.
[22:26] <arwa> For example:
[22:28] <arwa> pp=hb|40 is not working
[22:28] <arwa> Is there some syntax error?
[22:33] <michaelni> -vf pp=hb|40 would need some quoting like with '' or "" 
[22:35] <michaelni> you can use : instead of | IIRC
[22:37] <arwa> Okay thanks :)
[22:39] <michaelni> both  -vf 'pp=hb|40' and  -vf pp=hb:40  should work
[22:44] <arwa> It i working. Thank you.
[22:54] <cone-51> ffmpeg.git 03Michael Niedermayer 07master:61928b68dc28: h264: Do not share rbsp_buffer across threads
[22:54] <cone-51> ffmpeg.git 03Michael Niedermayer 07master:72db3c79dfdd: Merge commit '61928b68dc28e080b8c8191afe5541123c682bbd'
[23:02] <arwa> I am getting error with this - 'pp=hb/vb/dr/fq=8'
[23:02] <arwa> [Parsed_pp_0 @ 0xb8e1980] Option 'hb/vb/dr/fq' not found
[23:16] <ubitux> heh, hannibal looks so sweet after deblocking :3
[23:17] <arwa> I know!! :P
[00:00] --- Thu Jan 29 2015