[Ffmpeg-devel-irc] ffmpeg-devel.log.20130211

[burek]

[01:11] <cone-654> ffmpeg.git 03Michael Niedermayer 07master:8491ac3ad2e9: mlpdec: check ch_assign more completely
[01:11] <cone-654> ffmpeg.git 03Ronald S. Bultje 07master:55e35c98d5f3: h264 intra pred: add missing includes.
[01:11] <cone-654> ffmpeg.git 03Ronald S. Bultje 07master:c7e3e55429cb: Move ff_emulated_edge_mc prototypes to videodsp.
[01:11] <cone-654> ffmpeg.git 03Ronald S. Bultje 07master:5e74a2b68563: bit_depth_template: don't depend on dsputil.h.
[01:12] <Darkarnium> Hey Compn, sorry, I didn't see your message last night until late
[01:18] <Darkarnium> Pretty damned nasty, and only being used in a development environment as a PoC at the moment: http://pastebin.com/3LNcery6
[01:19] <Darkarnium> Worth noting, I'm using the snprintf and av_dict_set and 'headers' as 'cookie' does not seem to apply for my test. Whether I specify the options as per RFC6265 or as just an attribute value pair.
[01:30] <wm4> someone is reporting broken decoding with h264_vda in recent ffmpeg git (might be my fault, but I don't know and it used to work fine)
[02:36] <cone-654> ffmpeg.git 03Ronald S. Bultje 07master:fd6a021d8e3d: get_bits: return pointer to buffer that is the result of the alignment.
[11:31] <Tjoppen> someone's improving my cinepak encoder toy. this makes me happy :)
[11:50] <durandal_1707> i want that filter! - http://slowmovideo.granjow.net/videos.html
[11:53] <Tjoppen> ooh
[11:59] <durandal_1707> michaelni: when gbrp16 output?
[12:09] <cone-615> ffmpeg.git 03Paul B Mahol 07master:5d7009dac28b: evrcdec: fix wrong condition
[12:10] <michaelni> ill look into adding gbrp16 support if noone else does ...
[12:17] <someone-noone> Hello! Question related to vda_h264 decoding. Does VDADecoderDecode in sync mode returns frames in display order? Docs are saying that frames are always in decode order, however ffmpeg reorders it only in async mode and do not reorder in sync mode. I'm confused. Is there smth I've missed in sources?
[12:23] <michaelni> maybe ask dilaroga or xidorn quan
[12:27] <someone-noone> thanks. How often are they online on this channel?
[12:27] <someone-noone> or better try mail?
[12:28] <cone-615> ffmpeg.git 03Diego Biurrun 07master:d46d87d1d3de: sparc: dsputil_vis: Fix silly variable name search and replace typo
[12:28] <cone-615> ffmpeg.git 03Michael Niedermayer 07master:c3a62d3507b2: Merge remote-tracking branch 'qatar/master'
[12:31] <michaelni> probably better mail, ive seen dilaroga here but not common enough for waiting to be a good idea IIRC
[12:41] <cone-615> ffmpeg.git 03Michael Niedermayer 07release/0.5:272e7f6443b7: huffyuvdec: Check init_vlc() return codes.
[12:41] <cone-615> ffmpeg.git 03Michael Niedermayer 07release/0.5:ac476bfa9f90: huffyuvdec: Skip len==0 cases
[12:41] <cone-615> ffmpeg.git 03Reinhard Tartler 07release/0.5:5c9d2d837783: Bump version number for the 0.5.10 release
[12:41] <cone-615> ffmpeg.git 03Anton Khirnov 07release/0.5:0f6d4da8def2: bmpdec: only initialize palette for pal8.
[12:41] <cone-615> ffmpeg.git 03Mina Nagy Zaki 07release/0.5:4475a7d88b89: lavfi: avfilter_merge_formats: handle case where inputs are same
[12:41] <cone-615> ffmpeg.git 03Michael Niedermayer 07release/0.5:6b97e76dfca8: avsdec: Set dimensions instead of relying on the demuxer.
[12:41] <cone-615> ffmpeg.git 03Aneesh Dogra 07release/0.5:c28c631d29a8: bytestream: add a new set of bytestream functions with overread checking
[12:41] <cone-615> ffmpeg.git 03Anton Khirnov 07release/0.5:d1729c3715af: avidec: use actually read size instead of requested size
[12:41] <cone-615> ffmpeg.git 03Michael Niedermayer 07release/0.5:4fac60d56863: cavsdec: check for changing w/h.
[12:41] <cone-615> ffmpeg.git 03Anton Khirnov 07release/0.5:2ae6bdbb9b17: avidec: return 0, not packet size from read_packet().
[12:41] <cone-615> ffmpeg.git 03Anton Khirnov 07release/0.5:d4e4234147c9: yuv4mpeg: return proper error codes.
[12:41] <cone-615> ffmpeg.git 03JindYich Makovi
ka 07release/0.5:6731776795a8: h264: avoid stuck buffer pointer in decode_nal_units
[12:41] <cone-615> ffmpeg.git 03Luca Barbato 07release/0.5:5235db68c062: mpegaudiodec: fix short_start calculation
[12:41] <cone-615> ffmpeg.git 03Alex Converse 07release/0.5:80f89a9b40cc: tiffenc: Check av_malloc() results.
[12:41] <cone-615> ffmpeg.git 03Luca Barbato 07release/0.5:1f1b2f18067b: yuv4mpeg: reject unsupported codecs
[12:41] <cone-615> ffmpeg.git 03Janne Grunau 07release/0.5:2e1474fd9988: lavf: avoid integer overflow in ff_compute_frame_duration()
[12:41] <cone-615> ffmpeg.git 03Dale Curtis 07release/0.5:c3761b661874: Fix uninitialized reads on malformed ogg files.
[12:41] <cone-615> ffmpeg.git 03Luca Barbato 07release/0.5:fe4409a396d7: oggdec: check memory allocation
[12:41] <cone-615> ffmpeg.git 03Janne Grunau 07release/0.5:a49599b1255e: h264: check context state before decoding slice data partitions
[12:41] <cone-615> ffmpeg.git 03Justin Ruggles 07release/0.5:7aeb281aa507: shorten: check for realloc failure
[12:41] <cone-615> ffmpeg.git 03Michael Niedermayer 07release/0.5:9def5c466648: shorten: Use separate pointers for the allocated memory for decoded samples.
[12:41] <cone-615> ffmpeg.git 03Luca Barbato 07release/0.5:4f8f4458a5a8: vp56: release frames on error
[12:41] <cone-615> ffmpeg.git 03Luca Barbato 07release/0.5:b9500bf864e9: vp6: properly fail on unsupported feature
[12:41] <cone-615> ffmpeg.git 03Michael Niedermayer 07release/0.5:10ec2308b0d4: Merge remote-tracking branch 'qatar/release/0.5' into release/0.5
[12:57] <durandal_1707> michaelni: what was idea of noise filter? and how addgrain (film grain) relates to it?
[13:12] <cone-615> ffmpeg.git 03Carl Eugen Hoyos 07master:182e342fdbc7: Print mxf Universal Label if the user requested verbose logging.
[13:19] <durandal_1707> shouldn't colormatrix filter actually use libswscale?
[13:21] <michaelni> the idea of noise was to add different kinds of noise and grain ...
[13:24] <durandal_1707> i don't think there is way in lavfi to overlay frames with differen opacity
[13:52] <durandal_1707> how should i name blend filter? mix/merge/blend
[13:56] <iive> mix sounds too audio related.
[13:58] <durandal_1707> iive: its called amix
[14:00] <iive> the filter may be called amix, but I do associate the mix word with audio.
[14:12] <durandal_1707> ubitux: want to add different edge detectors (not just canny) to edgedetect?
[15:23] <ubitux> durandal_1707: i called the filter "edgedetect" and not "canny" for that reason, so feel free to add some enhancement
[15:26] Action: ubitux is going back tomorrow :(
[15:44] <durandal_1707> ubitux: i already have bunch of other work, and you wrote edgedetect and are most qualified person to extend it
[15:46] <cone-615> ffmpeg.git 03Paul B Mahol 07master:8160864b0027: lavfi/amerge: make description consistent with documentation
[16:02] <ubitux> durandal_1707: i have no plans for it atm
[16:21] <cone-615> ffmpeg.git 03Derek Buitenhuis 07master:c177f2ec4a21: libfdk-aacenc: Actually check for upper bounds of cutoff
[17:14] <durandal_1707> ubitux: your opinion on blend filter?
[17:14] <durandal_1707> overlay already have some blend code - very limited one though
[17:16] <durandal_1707> also i dont understand this complicated logic with overlay/alphamerge request_frame thing
[17:57] <durandal_1707> michaelni: i added yuv444 support to overlay and it is broken, works if i add format=yuv444p to begin and end any pix fmt of filter chain
[17:58] <durandal_1707> this is strange
[18:01] <durandal_1707> filter expect that overlay and main are of same format but it does not happen to actually be in config_input
[18:10] <durandal_1707> negate/lut debug filter output is boring spam
[18:11] <durandal_1707> it inserts scaler 444 -> 420 in middle
[18:11] <durandal_1707> perhaps ffplay nonsense
[18:15] <durandal_1707> indeed, it happens with ffplay only
[18:25] <durandal_1707> so i want basically VV->V filter without pts nonsese stuff invading my code
[18:26] <durandal_1707> is it doable?
[18:30] <durandal_1707> another thing is that filters write to first frame, I dont find this much useful for blend and want to have two input read only frames and output writeonly
[18:34] <michaelni> you can set permission on the inputs
[18:34] <michaelni> or a filter can output read only buffers so they wont be written into
[18:34] <durandal_1707> yes, I know that, but what about this PRESERVE and pts nonsense
[18:35] <durandal_1707> i just read pixels from 2 inputs and write pixel to output
[18:36] <durandal_1707> and both 2 inputs should have same format
[18:37] <michaelni> for same format just make them use the same list and you should get the same format
[18:41] <durandal_1707> michaelni: how? see bug with ffplay and overlay when i added yuv444p
[18:44] <michaelni> i think vf_overlay uses seperate lists for its 2 inputs not sure thats the issue though
[18:46] <durandal_1707> formats are set with query_formats and its global for all inputs/outputs
[18:59] <durandal_1707> so filter_frame() can take two inputs at same time?
[19:03] <durandal_1707> looks like i need to do some kind of buffering
[19:04] <durandal_1707> this is nonsense, why would any filter would need to do this?
[19:07] <durandal_1707> i will just copy-paste code
[19:13] <durandal_1707> saste: about il: feel free to simplify options if you think that per plane options do not make sense
[19:14] <durandal_1707> saste: what would be template of filter that takes 2 readonly inputs and 1 write only output?
[19:14] Action: durandal_1707 thinks that having filter templates could help novice devs
[19:20] <saste_> durandal_1707, overlay?
[19:21] <durandal_1707> overlay writes to first input
[19:21] <durandal_1707> that is useless for blend
[19:23] <durandal_1707> well not always useless, but i imagine scenario where it would be useful to have original input
[19:24] <durandal_1707> blend is just geq but instead on single frame it operates on VV->V
[19:29] <cone-615> ffmpeg.git 03Michael Niedermayer 07master:0aa1d848ece6: ffv1enc: better heuristic to calculate initial states
[20:14] Action: durandal_1707 now that is priority ^
[22:04] <llogan> why is rbg24 not automatically chosen for gif output instead of the current AV_LOG_ERROR message?
[22:05] <llogan> it is confusing to users
[22:05] <llogan> *rgb24
[22:18] <llogan> Compn: i'm not against your request, FYI.
[22:31] <Compn> llogan : i know, i'm just pointing out -ss 30 is seconds, so -ss 30:30 is minutes+seconds
[22:35] <llogan> gif appears to support pal8, (if you use -f image2) but that message does make that apparent.
[23:00] <llogan> s/does/does not/
[23:07] <cone-615> ffmpeg.git 03Ronald S. Bultje 07master:972771dcf27d: h264chroma: remove duplicate 9/10 bit functions.
[23:39] <BBB-work> michaelni, one more for you, a file encoded with "./ffmpeg -i bear_mpeg4_mp3.avi -acodec copy -vcodec mpeg4 -flags mv4+qpel+mv0+cbp+aic -trellis 1 -dia_size 1040 -g 250 -bf 8 -b_strategy 2 -cmp sad -subcmp rd -mbd 2 -precmp sad -last_pred 4 -subq 8 -bidir_refine 4 -qns 3 bear_mpeg4asp_mp3.avi" apparently fails to playback properly in ffplay (stutters a lot) because of a bug in compute_pkt_fields() for files with b-frames that are not
[23:39] <BBB-work>  h264
[23:39] <BBB-work> [let's continue that here]
[23:40] <michaelni> ok
[23:42] <cone-615> ffmpeg.git 03Ronald S. Bultje 07master:b582af1ed778: Use simple nop codes for <= sse (rather than <= mmx).
[23:45] <cone-615> ffmpeg.git 03Stefano Sabatini 07master:4c69cafceea5: doc/filters: fix position of il filter docs
[23:45] <cone-615> ffmpeg.git 03Stefano Sabatini 07master:af60e451509a: doc/filters: apply minor fixes to il filter docs
[23:45] <cone-615> ffmpeg.git 03Stefano Sabatini 07master:64353be81135: doc/filters: rework idet filter description
[23:56] <michaelni> BBB-work, i see the stuttering too without encoding with -bf 8
[00:00] --- Tue Feb 12 2013