[Ffmpeg-devel-irc] ffmpeg-devel.log.20150720

burek burek021 at gmail.com
Tue Jul 21 02:05:03 CEST 2015

[00:11:23 CEST] <cone-506> ffmpeg 03Michael Niedermayer 07master:5c583b141980: avcodec/dds: Flip more R<->B cases
[00:15:11 CEST] <genio> I'm not certain if this is a bug, but it seems like one.  https://www.cwhitener.com/test.avi is a fairly crappy avi file that I can't control how is being created.  When I try to ffmpeg -i test.avi I get tons of errors and cannot transcode the file to mp4 or any other.  However, VLC plays the original without issue. 
[00:47:16 CEST] <cone-506> ffmpeg 03Carl Eugen Hoyos 07master:d8d2f934bd00: lavc: Improve documentation for rc_max_rate and bit_rate.
[00:47:17 CEST] <cone-506> ffmpeg 03Carl Eugen Hoyos 07master:32d8726a31b7: lavf/rtpenc_jpeg: Warn if number of present quantization tables is not two.
[00:47:18 CEST] <cone-506> ffmpeg 03Carl Eugen Hoyos 07master:c239fcb94893: lavc/jpeg2000: Signal lossless / bit-exact video stream.
[02:22:49 CEST] <cone-506> ffmpeg 03Michael Niedermayer 07master:749f85496ccf: avcodec/ffv1enc: fix assertion failure with unset bits per raw sample
[02:22:50 CEST] <cone-506> ffmpeg 03Michael Niedermayer 07master:bf1fd59c6a22: avformat/avidec: Fix demuxing avi with broken index
[02:23:27 CEST] <michaelni> genio, should be fixed
[02:24:31 CEST] <genio> so it works for you on latest?
[02:45:06 CEST] <rcombs> wm4: I threw together a dumb partial implementation of MOV sidx parsing: https://gist.github.com/46cd5f5cc53d31a7ff55
[02:45:33 CEST] <rcombs> right now backwards seeking is broken for, uh, reasons of some sort
[02:45:51 CEST] <rcombs> but it sort of works!
[02:46:11 CEST] <rcombs> someone yell at me if I'm using structures for something entirely different from their intended purposes
[02:54:41 CEST] <cone-506> ffmpeg 03Michael Niedermayer 07master:cb53db144c10: avformat/mp3dec: Rename sync() function to avoid build failure on android
[02:58:07 CEST] <jamrial> genio: he just commited a fix to git, so do a git pull and recompile to get it
[03:49:37 CEST] <cone-506> ffmpeg 03Michael Niedermayer 07n2.7.2:HEAD: avformat/mp3dec: Rename sync() function to avoid build failure on android
[04:43:21 CEST] <cone-506> ffmpeg 03Luca Barbato 07master:e93ca480c913: parseutil: Add more resolution aliases
[04:43:22 CEST] <cone-506> ffmpeg 03Michael Niedermayer 07master:9ebe041e1ca8: Merge commit 'e93ca480c91397f82f2ea5b6a8b82ac6d0e2bfd2'
[05:08:18 CEST] <cone-506> ffmpeg 03James Almer 07master:6de5b6cd8616: avfilter/af_astats: use UINT64_C instead of the LLU suffix
[05:11:11 CEST] <jamrial> keeping vs2012 alive :p
[05:33:19 CEST] <genio> jamrial: ahh, I overlooked it as I tend to skip over colors in IRC
[05:33:22 CEST] <genio> thanks
[08:39:58 CEST] <jamrial> lol, i fix af_astats for vs2012 only to find af_dynaudnorm was broken as well
[08:42:08 CEST] <jamrial> erf() and copysign(), which are not available on vs2012
[08:43:01 CEST] <nevcairiel> do we even have replacements of those
[08:43:06 CEST] <jamrial> nope
[08:43:19 CEST] <jamrial> anyone willing to write them? :p
[08:43:36 CEST] <nevcairiel> i dont evne know what those do
[08:44:03 CEST] <nevcairiel> copysign is probably easy
[08:44:09 CEST] <nevcairiel> but erf
[08:49:54 CEST] <jamrial> maybe http://www.johndcook.com/blog/cpp_erf/
[08:51:03 CEST] <nevcairiel> should just add a configure check, make it depend on it, done
[08:51:08 CEST] <nevcairiel> if anyone cares, they cna do it =p
[09:02:32 CEST] <jamrial> sounds good. will send a patch in a moment
[09:06:08 CEST] <jamrial> there
[09:07:22 CEST] <jamrial> wonder which of the other new filters will blow up next
[09:14:19 CEST] <wm4> rcombs: seems promising
[09:15:09 CEST] <wm4> unfortunate that it appears to require a completely new seek implementation
[09:16:00 CEST] <rcombs> wm4: yeah, I tried just adding index entries
[09:16:33 CEST] <rcombs> that doesn't work because the mov demuxer expects those to point to samples within the data atoms, rather than at atoms to be parsed
[09:16:44 CEST] <rcombs> on the bright side, I know the mov demuxer a bit better now
[09:16:54 CEST] <nevcairiel> not sure t hats a bright side
[09:17:00 CEST] <rcombs> though not quite well enough to work out why backseeking isn't working (yet)
[09:17:05 CEST] <rcombs> nevcairiel: heh
[09:18:03 CEST] <wm4> rcombs: what do you mean with backseeking? simply seeking to an earlier timestamp?
[09:18:08 CEST] <rcombs> yeah
[09:18:16 CEST] <rcombs> which means it's some sort of state problem
[09:18:23 CEST] <wm4> ugh
[09:19:02 CEST] <rcombs> maybe something in mov_read_tfhd
[09:19:18 CEST] <rcombs> which is one of the things that makes me wonder if I'm not using its structures the way they were intended
[09:19:55 CEST] <rcombs> (they're laid out pretty much perfectly for what I'm doing and the values I've looked at match up with what that code expects, but somehow shit's ending up fucked anyway)
[09:20:46 CEST] <JEEB> qtff pdf doesn't help?
[09:21:05 CEST] <JEEB> unless you're not dealing with mov, in which case it's isobmff
[09:21:33 CEST] <rcombs> JEEB: I'm more concerned about mov.c's structs at this point than the actual MP4 things involved, which I think I understand well enough at this point
[09:22:10 CEST] <rcombs> I'm definitely reading the sidx atom well enough for the files I'm working with
[09:22:37 CEST] <rcombs> (I'm completely ignoring the timescale setting right now but that's just because I'm lazy and it should be just a couple lines to fix)
[09:23:41 CEST] <nevcairiel> i'm sl ightly confused by the patch, you call mov_seek_fragment, but once its done, it still runs through the normal seek functions?
[09:23:45 CEST] <nevcairiel> maybe those screw you up?
[09:23:50 CEST] <rcombs> nevcairiel: yes
[09:24:02 CEST] <rcombs> nevcairiel: mov_seek_fragment gets you to the right fragment and parses its headers
[09:24:28 CEST] <rcombs> and then the regular seek function handles the AVIndex-based seeking, with the index populated from the trun atom
[09:25:06 CEST] <rcombs> oh, and I wasn't quite sure how to correctly match between track_ids and AVStream->ids
[09:26:47 CEST] <rcombs> I might have an issue in mov_read_default's logic for deciding when it's finished
[10:48:40 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:3bb17d2692cd: avcodec/golomb: get_ur_golomb_jpegls: Fix reading huge k values
[10:48:40 CEST] <cone-809> ffmpeg 03Andrey Utkin 07release/2.6:4c2a1e673bfd: rtpenc_jpeg: handle case of picture dimensions not dividing by 8
[10:48:40 CEST] <cone-809> ffmpeg 03Martin Storsjö 07release/2.6:94a3d2e49caa: rtsp: Make sure we don't write too many transport entries into a fixed-size array
[10:48:40 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:862ba8b4b4d8: avformat/nutdec: Return error on EOF from get_str()
[10:48:40 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:627040156193: avformat/nutdec: Fix recovery when immedeately after seeking a failure happens
[10:48:40 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:bd9313d9f840: avformat/nutdec: Check X in 2nd branch of index reading
[10:48:40 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:d4d8c3553d08: nutdec: fix infinite resync loops
[10:48:41 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:9c3d306ba437: nutdec: stop skipping bytes at EOF
[10:48:42 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:cb921dd47f74: nutdec: abort if EOF is reached in decode_info_header/read_sm_data
[10:48:43 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:4db46cf5208f: avformat/wavdec: Increase probe_packets limit
[10:48:44 CEST] <cone-809> ffmpeg 03Rodger Combs 07release/2.6:308bcf53f9d6: avformat/wavdec: Increase dts packet threshold to fix more misdetections
[10:48:45 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:4947158b1f34: avcodec/put_bits: Update size_in_bits in set_put_bits_buffer_size()
[10:48:46 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:665b34df730b: avcodec/mpegvideo_enc: Update the buffer size as more slices are merged
[10:48:47 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:63716cebb797: avcodec/put_bits: Assert that there is enough space left in skip_put_bytes()
[10:48:48 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:9e3723b848c1: avcodec/bitstream: Assert that there is enough space left in avpriv_copy_bits()
[10:48:49 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:3ea2ffc284d4: ffmpeg_opt: Set the video VBV parameters only for the video stream from -target
[10:48:50 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:c40aa4704f51: avfilter/x86/vf_fspp: Fix invalid combination of opcode and operands
[10:48:51 CEST] <cone-809> ffmpeg 03Luca Barbato 07release/2.6:b46ae71f29a1: h264: Make sure reinit failures mark the context as not initialized
[10:48:52 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:9591f2fe94fc: swresample: Check the return value of resampler->init()
[10:48:53 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:d6d328c2947c: libopenjpegenc: add NULL check for img before accessing it
[10:48:54 CEST] <cone-809> ffmpeg 03Ganesh Ajjanagadde 07release/2.6:9d749d69b224: swresample/dither: check memory allocation
[10:48:55 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:19cef664fed6: avformat/mxfenc: Accept MXF D-10 with 49.999840 Mbit/sec
[10:48:56 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:7050d07562c0: Revert "avformat/rtpenc: check av_packet_get_side_data() return, fix null ptr dereference"
[10:48:57 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:ed5bfeca04fb: swresample/swresample: Cleanup on init failure.
[10:48:58 CEST] <cone-809> ffmpeg 03Nedeljko Babic 07release/2.6:c8377cc49b2f: libavutil/softfloat: Fix av_normalize1_sf bias.
[10:48:59 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:c2595b448571: avcodec/atrac3plusdec: consume only as many bytes as available
[10:49:00 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:2b31264e1a6b: avcodec/alsdec: Check for overread
[10:49:01 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:e354ec95d6bb: avcodec/adpcm: Check for overreads
[10:49:02 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:e21db6e3399f: avformat/ffmdec: Check ffio_set_buf_size() return value
[10:49:03 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:6b4c6b7ff5c1: avcodec/jpeg2000dec: Check that coords match before applying ICT
[10:49:04 CEST] <cone-809> ffmpeg 03Simon Thelen 07release/2.6:3c7e41df27e2: libavutil/channel_layout: Correctly return layout when channel specification ends with a trailing 'c'.
[10:49:05 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:33ec70785f4a: avdevice/lavfi: do not rescale AV_NOPTS_VALUE in lavfi_read_packet()
[10:49:06 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:0afa8a6c859b: avcodec/h264_slice: Use AVFrame diemensions for grayscale handling
[10:49:07 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:d15f4f5f22ee: takdec: ensure chan2 is a valid channel index
[10:49:08 CEST] <cone-809> ffmpeg 03Deliang Fu 07release/2.6:35e46e74bc24: avformat: Fix bug in parse_rps for HEVC.
[10:49:09 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:1c5d25f2c008: ffmpeg_opt: Check for localtime() failure
[10:49:10 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:274121179f40: mov: abort on EOF in ff_mov_read_chan
[10:49:11 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:c7ef69c23c01: libopenjpegdec: check existence of image component data
[10:49:12 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:b477f97ab984: arm: only enable setend on ARMv6
[10:49:13 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:c10871aca817: vp9: change type of tile_size from unsigned to int64_t
[10:49:14 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:31dd78363138: vp8: change mv_{min,max}.{x,y} type to int
[10:49:15 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:3ea15a4547f8: sonic: set avctx->channels in sonic_decode_init
[10:49:16 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:008229b7f978: h264: er: Copy from the previous reference only if compatible
[10:49:17 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:7b1c5fa5e579: examples/demuxing_decoding: use properties from frame instead of video_dec_ctx
[10:49:18 CEST] <cone-809> ffmpeg 03wm4 07release/2.6:c049e979b605: avio: fix potential crashes when combining ffio_ensure_seekback + crc
[10:49:19 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:a5e8ca7e73f8: ffmpeg: Free last_frame instead of just unref
[10:49:20 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:cd85dd1373d6: postproc: fix unaligned access
[10:49:21 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:47e950848599: avcodec/dpxenc: implement write16/32 as functions
[10:49:22 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:476cf3eb7e6f: matroskadec: validate audio channels and bitdepth
[10:49:23 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:5019f8d10c08: matroskadec: check audio sample rate
[10:49:24 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:f2e3d5adb3ca: swr: Remember previously set int_sample_format from user
[10:49:25 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:80cec4e4960b: swscale/rgb2rgb_template: Disable shuffle_bytes_2103_c on big endian
[10:49:26 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:12e166dacb54: swscale/rgb2rgb_template: Implement shuffle_bytes_0321_c and fix shuffle_bytes_2103_c on BE
[10:49:27 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:e920af2cdf3f: swscale/rgb2rgb_template: Fix signedness of v in shuffle_bytes_2103_c()
[10:49:28 CEST] <cone-809> ffmpeg 03Sebastien Zwickert 07release/2.6:be6a569e3ab8: vda: unlock the pixel buffer base address.
[10:49:29 CEST] <cone-809> ffmpeg 03James Almer 07release/2.6:860dc1e19b99: swscale/x86/rgb2rgb_template: add missing xmm clobbers
[10:49:30 CEST] <cone-809> ffmpeg 03James Almer 07release/2.6:48127e2ad6e2: swscale/x86/rgb2rgb_template: fix signedness of v in shuffle_bytes_2103_{mmx,mmxext}
[10:49:31 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:3b136179a308: ffmpeg: Do not use the data/size of a bitstream filter after failure
[10:49:32 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:49ad1658b2db: swscale/swscale_unscaled: Fix rounding difference with RGBA output between little and big endian
[10:49:33 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:0324008de61b: wmavoice: limit wmavoice_decode_packet return value to packet size
[10:49:34 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:63ef4014c5ff: avcodec/pngdec: Only allow one IHDR chunk
[10:49:35 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:873b08f4111a: avcodec/pngdec: Require a IHDR chunk before fctl
[10:49:36 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:917544b2eace: avcodec/pngdec: Check values before updating context in decode_fctl_chunk()
[10:49:37 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:1ea58c6c6a88: avfilter/vf_transpose: Fix rounding error
[10:49:38 CEST] <cone-809> ffmpeg 03James Zern 07release/2.6:002bbc36875d: vp9/update_prob: prevent out of bounds table read
[10:49:39 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:11142fb83bb8: avcodec/h264_slice: Use w/h from the AVFrame instead of mb_w/h
[10:49:40 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:1674c5beafc4: avcodec/aacsbr: check that the element type matches before applying SBR
[10:49:41 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:4fad45ad146c: avcodec/aacsbr: Assert that bs_num_env is positive
[10:49:42 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:a4cb97067b4c: avcodec/rawenc: Use ff_alloc_packet() instead of ff_alloc_packet2()
[10:49:43 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:1ca226989c31: wmalosslessdec: avoid reading 0 bits with get_bits
[10:49:44 CEST] <cone-809> ffmpeg 03Andreas Cadhalpun 07release/2.6:6d6ca8b82e34: wmalosslessdec: reset frame->nb_samples on packet loss
[10:49:45 CEST] <cone-809> ffmpeg 03Chris Watkins 07release/2.6:77f5302465ca: oggparsedirac: check return value of init_get_bits
[10:49:46 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:27a88f9d45d6: avcodec/mpegvideo: Clear pointers in ff_mpv_common_init()
[10:49:47 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:6c98d5e7c83f: avcodec/utils: use a minimum 32pixel width in  avcodec_align_dimensions2() for H.264
[10:49:48 CEST] <cone-809> ffmpeg 03Anton Khirnov 07release/2.6:c5c141bc699f: bytestream2: set the reader to the end when reading more than available
[10:49:49 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:0cb600b7bfb7: avcodec/alac: Clear pointers in allocate_buffers()
[10:49:50 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:df6f2daccf66: avcodec/sanm: Reset sizes in destroy_buffers()
[10:49:51 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:9d0f7a8d78d6: avcodec/pthread_frame: check avctx on deallocation
[10:49:52 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:fca28d35e493: ffmpeg: Fix cleanup with ost = NULL
[10:49:53 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:77eddee37535: ffmpeg: Fix crash with ost->last_frame allocation failure
[10:49:54 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:e604662d0476: avformat/mov: Fix deallocation when MOVStreamContext failed to allocate
[10:49:55 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:7dbfeeeb9448: ffmpeg: Fix cleanup after failed allocation of output_files
[10:49:56 CEST] <cone-809> ffmpeg 03Zhang Rui 07release/2.6:5db095793a28: avutil/fifo: Fix the case where func() returns less bytes than requested in av_fifo_generic_write()
[10:49:57 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:fb47b6cb4b89: swscale/utils: Clear pix buffers
[10:49:58 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:ed6d8e185a4e: avcodec/pthread_frame: clear priv_data, avoid stale pointer in error case
[10:49:59 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:b7f1d51b9c11: avfilter/af_aresample: Check ff_all_* for allocation failures
[10:50:00 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:620b3e680c38: avcodec/rv34: Clear pointers in ff_rv34_decode_init_thread_copy()
[10:50:01 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:0de6cf5fdd9e: avcodec/diracdec: Check for hpel_base allocation failure
[10:50:02 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:e3a521d817cb: avcodec/vp8: Fix null pointer dereference in ff_vp8_decode_free()
[10:50:03 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:e8ce2770b12a: avcodec/vp8: Check buffer size in vp8_decode_frame_header()
[10:52:22 CEST] <nevcairiel> that fruity guy just doesnt get it
[10:52:24 CEST] <nevcairiel> oh well
[11:14:13 CEST] <durandal_1707> not again
[11:23:55 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07release/2.6:827a8bfc64a4: Update for 2.6.4
[12:00:49 CEST] <cone-809> ffmpeg 03Paul B Mahol 07master:7967474bd972: avfilter/af_amerge: increase max number of channels from 32 to 64
[13:42:31 CEST] <BBB> does anyone know what is the serial number in AVPacket and in ffplay.c lines like d->finished = d->pkt_serial; or in the parameter packet_queue_get()?
[13:42:51 CEST] <BBB> (like, what is it used for)
[13:45:39 CEST] <nevcairiel> i dont see a serial number in avpacket
[13:48:31 CEST] <nevcairiel> seems like a thing ffplay invented
[13:49:31 CEST] <BBB> hm, youre right
[13:49:33 CEST] <BBB> so what is it?
[13:49:43 CEST] <BBB> it seems to get it from the clock, not from the packet
[13:50:28 CEST] <nevcairiel> no clue
[13:51:08 CEST] <BBB> not even that, its just its private packet list, an incremental counter
[13:51:09 CEST] <BBB> MyAVPacketList
[13:51:40 CEST] <BBB> so& I dont know who added that, annotate has too many cosmetics to be useful, but what is it useful for? michaelni, do you know?
[13:55:00 CEST] <Compn> http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=a687acbbf0869def24f516b0147e9ff93695c347
[13:55:02 CEST] <Compn> maaaaybe ?
[13:55:39 CEST] <nevcairiel> that whole patch doesnt even have a single serial in it =p
[13:56:13 CEST] <Compn> oop
[13:56:50 CEST] <nevcairiel> anyway BBB, http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=2a4c7e6540c4549d6a47cd7cb288d4d910a99736
[13:56:56 CEST] <nevcairiel> i'll leave you to interpret whatever that means
[13:57:21 CEST] <Compn> http://git.videolan.org/?p=ffmpeg.git;a=commitdiff;h=2a4c7e6540c4549d6a47cd7cb288d4d910a99736
[13:57:24 CEST] <Compn> argh
[13:57:26 CEST] <Compn> nev beat me
[13:58:03 CEST] <Compn> The purpose of the serial field is to accompany the decoded data during the
[13:58:03 CEST] <Compn>  decoding process to know if the decoded data belongs to the data stream after
[13:58:03 CEST] <Compn>  the latest packet queue flush.
[13:58:05 CEST] <Compn> hehe
[14:02:29 CEST] <michaelni> Compn, yes, that sounds about correct, serial was for figuring out to which side of a flush a packet belongs IIRC
[14:32:36 CEST] <BBB> d->finished = d->pkt_serial
[14:32:39 CEST] <BBB> so what does that mean then?
[14:32:44 CEST] <BBB> I still dont understand that line
[14:33:59 CEST] <BBB> it seems like an eof identifier?
[14:37:03 CEST] <cone-809> ffmpeg 03hSÇ 07master:4bd619339762: avcodec: loongson constants redefined with macros
[14:37:04 CEST] <cone-809> ffmpeg 03hSÇ 07master:fdac5ff682f0: avcodec: loongson relocate constants of idctdsp and h264pred
[14:54:27 CEST] <ubitux> where is the codec guessed from the format?
[14:54:35 CEST] <ubitux> (for encoding)
[15:02:13 CEST] <BBB> AVOutputFormat.video_codec?
[15:02:16 CEST] <BBB> and audio_codec
[15:02:22 CEST] <BBB> or you mean wheres the code consuming that field?
[15:04:56 CEST] <ubitux> so it's set automatically when allocating an output format?
[15:05:00 CEST] <ubitux> thx, will look at this
[15:05:22 CEST] <BBB> ffmpeg_opt.c has a call to av_guess_codec
[15:05:35 CEST] <BBB> which takes an AVOutputFormat as input
[15:05:40 CEST] <BBB> I think that does what youre looking for
[15:05:54 CEST] <ubitux> sounds perfect indeed
[15:06:12 CEST] <ubitux> thx
[15:17:04 CEST] <goncalomsdias> any expert in decklink cards? I'm having segmentation fault while sending one file to output. It works if I pass the option -an
[15:31:22 CEST] <michaelni> durandal_1707, tblend test fails on arm it seems: http://fate.ffmpeg.org/report.cgi?time=20150720042949&slot=armv7l-panda-gcc4.4-armv6j
[15:39:14 CEST] <durandal_1707> michaelni: perhaps multiplying with 1.0 causes problems?
[15:47:02 CEST] <cone-809> ffmpeg 03Donny Yang 07master:0030613d6925: avcodec/apngdec: Fix typos in decoder causing incorrect results
[15:52:57 CEST] <durandal_1707> michaelni: does "apng: Fix decoding images with the PREVIOUS dispose op" hang on smp?
[15:56:49 CEST] <michaelni> durandal_1707, tblend passes if i outcomment av_clip_uint8_arm()
[15:57:18 CEST] <durandal_1707> Huh
[16:01:01 CEST] <durandal_1707> its not commented here...
[16:01:34 CEST] <durandal_1707> you mean you commented it?
[16:02:13 CEST] <genio> michaelni: Your fix for my crappy video seems to be working perfectly.  Thank you for the quick turnaround.
[16:02:18 CEST] <genio> michaelni++
[16:02:35 CEST] <goncalomsdias> any expert in decklink cards? I'm having segmentation fault while sending one file to output. It works if I pass the option -an
[16:04:37 CEST] <michaelni> durandal_1707, yes i commented it out and it worked, i dont know if thats random luck or something wrong with it
[16:06:15 CEST] <durandal_1707> compare output of tblend=all_mode=difference128 in hex editor?
[16:07:13 CEST] <michaelni> it doesnt need hex editor it had a strong pink color tone
[16:19:18 CEST] <durandal_1707> michaelni: check if that arm function is actually correct...
[16:28:20 CEST] <mathieu_> Hi, I reported a bug in gstreamer-ffmpeg, and I was asked "Can you start a discussion with upstream FFMPeg please."
[16:28:27 CEST] <mathieu_> now what do I do? :)
[16:28:47 CEST] <J_Darnley> Tell us whet the bug is?
[16:29:32 CEST] <J_Darnley> A link to your bug report?
[16:29:52 CEST] <mathieu_> https://bugzilla.gnome.org/show_bug.cgi?id=752523
[16:30:11 CEST] <mathieu_> it's regarding the drawing of Motion Vectors on h264 video
[16:31:44 CEST] <goncalomsdias067> while using the decklink output format as show in the documentation, I get segmentation fault unless I pass the option -an. How can I have some help with this? should I open one bug report or is this already know issue?
[16:32:31 CEST] <J_Darnley> goncalomsdias067: No idea.  Maybe check the issues on trac.
[16:35:21 CEST] <goncalomsdias067> already did. I was able to found something once but didn't kept the url. Now I can't find it, but I know it was something old
[16:35:55 CEST] <J_Darnley> mathieu_: I am not familiar with that code but you might want to show how we can trigger that crash with ffmpeg
[16:36:47 CEST] <J_Darnley> For instance, what does the debug-mv option correspont to in ffmpeg or libavcodec?
[16:36:54 CEST] <goncalomsdias067> well, I guess you would need to have the decklink card
[16:38:29 CEST] <mathieu_> it corresponds to the avctx->debug_mv setting
[16:38:45 CEST] <durandal_1707> ubitux have you made that image with all colors?
[16:38:53 CEST] <J_Darnley> goncalomsdias067 https://duckduckgo.com/html/?q=site:trac.ffmpeg.org+decklink has lots of hits but they all seem to match --enable-decklink
[16:38:58 CEST] <michaelni> durandal_1707, seems to still fail even if i replace av_clip_uint8_arm by pure C code
[16:39:10 CEST] <ubitux> durandal_1707: not yet but it's trivial
[16:39:18 CEST] <ubitux> durandal_1707: you even have some code in haldclut iirc
[16:39:26 CEST] <mathieu_> there's that 1 line that is in both Gstreamer & FFMpeg, and commenting out fixes Gstreamer
[16:39:42 CEST] <mathieu_> however I'm not aware of the consequences, since I don't know the code well
[16:40:12 CEST] <mathieu_> a quick check with valgrind does not reveal any memory leaks by removing that line
[16:41:26 CEST] <goncalomsdias067> the first link seems to be the same problem I'm having. it only works if the option -an is used
[16:44:39 CEST] <durandal_1707> michaelni: you mean by av_clip?
[16:50:49 CEST] <goncalomsdias067> http://trac.ffmpeg.org/ticket/3480 seems that the ticket already exists
[17:16:46 CEST] <cone-809> ffmpeg 03Niklesh 07master:a927aacbe605: movtextdec.c: Correct the highlight tags
[17:20:11 CEST] <michaelni> durandal_1707, fixed tblend, that was a silly bug
[17:27:34 CEST] <durandal_1707> going to push it or send a patch?
[17:30:01 CEST] <cone-809> ffmpeg 03Djordje Pesut 07master:5499467d5d3d: avcodec: Template creation for AAC decoder (SBR-module)
[17:30:02 CEST] <cone-809> ffmpeg 03Jovan Zelincevic 07master:b0414da90d6d: avcodec: Table creation for AAC_fixed_decoder (SBR-module)
[17:30:03 CEST] <cone-809> ffmpeg 03Djordje Pesut 07master:f85bc147fb87: avcodec: Implementation of AAC_fixed_decoder (SBR-module)
[17:30:04 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:d82d11397f61: avutil/arm/intmath: return int for uint8 / uint16 clip
[17:30:58 CEST] <michaelni> durandal_1707, pushed it, feel free to change/solve it differently if you like
[17:36:14 CEST] <durandal_1707> michaelni: so other tests on arm also passes now?
[17:37:12 CEST] <durandal_1707> bunch of code uses those functions
[17:40:01 CEST] <ubitux> so, what's the quality flag for jpeg in ffmpeg? qscale? qmin/qmax? global_quality? something else?
[17:41:05 CEST] <michaelni> durandal_1707, just tested the one failing
[17:41:41 CEST] <cone-809> ffmpeg 03Ivan Uskov 07master:fec0485c4380: Adding myself as maintainer for qsv*
[17:42:05 CEST] <durandal_1707> ubitux: IIRC qmin qmax
[17:42:11 CEST] <ubitux> :/
[17:42:43 CEST] <ubitux> doesn't make much sense for a single picture
[17:49:33 CEST] <durandal_1707> ubitux: image with all colors 8bit would be 4096x4096 size
[17:50:00 CEST] <durandal_1707> But how to best order colors?
[17:50:30 CEST] <durandal_1707> could be video source filter
[17:52:47 CEST] <ubitux> yeah 4096x4096 is pretty fine
[17:52:52 CEST] <ubitux> wait i think i have one here
[17:53:30 CEST] <ubitux> -rw-r--r-- 1 ux ux  35M Jun 29 12:11 rgb24.png
[17:53:32 CEST] <ubitux> mmmh
[17:54:28 CEST] <ubitux> http://b.pkh.me/rgb24.png
[17:58:00 CEST] <ubitux> durandal_1707: we could generate it in fate
[17:58:14 CEST] <ubitux> i wonder if it couldn't be done in ppm with awk
[17:58:24 CEST] <ubitux> gonna write a hugez file though
[17:58:41 CEST] <ubitux> durandal_1707: hey, what about a source filter?
[17:59:05 CEST] <durandal_1707> I told you already...
[17:59:11 CEST] <ubitux> -f lavfi allcolors=rgb or allcolors=bgr
[17:59:14 CEST] <ubitux> oh, ok my bad
[17:59:20 CEST] <ubitux> well then do that
[18:00:06 CEST] <durandal_1707> another one of size 256x256 and 256 frames
[18:19:14 CEST] <cone-809> ffmpeg 03James Almer 07master:e1778fb657ca: x86/vf_ssim: fix some instruction comments
[18:19:15 CEST] <cone-809> ffmpeg 03James Almer 07master:e3851169eedc: x86/vf_ssim: add ff_ssim_4x4_line_xop
[18:19:16 CEST] <cone-809> ffmpeg 03James Almer 07master:736a3860555a: configure: check for erf() and copysign()
[20:14:46 CEST] <cone-809> ffmpeg 03Ivan Uskov 07master:264ba3d847f9: libavcodec/qsvdec.c: missed MFXVideoDECODE_Close() call
[20:26:44 CEST] <rcombs> wm4: sure enough, it was an issue in mov_read_default's logic for deciding when it's finished
[20:27:06 CEST] <rcombs> it was finishing too early because I forgot to reset `found_mdat`
[20:28:21 CEST] <wm4> so it works now?
[20:30:08 CEST] <rcombs> yup
[20:30:12 CEST] <rcombs> for the files I'm testing, anyway
[20:30:33 CEST] <rcombs> (I'm using mpv and youtube-dl with `--ytdl-format bestvideo+bestaudio`)
[20:32:28 CEST] <wm4> awesome
[20:32:30 CEST] <rcombs> https://gist.github.com/e649a7bc0c0f9c7459a2 still a couple things I'm pretty sure I'm doing wrong
[20:32:59 CEST] <rcombs> like matching AVStream indexes to MOV track IDs
[20:33:37 CEST] <rcombs> and I think things might go a little bit wonky if you have multiple streams in a single file (shouldn't be anything actually harmful, but it'll probably print some warnings)
[20:34:18 CEST] <rcombs> but I can probably send it as-is and see if someone who knows mov.c a bit better can work those out
[20:35:28 CEST] <rcombs> ended up using dash.js as a reference, which feels a little bit silly
[20:35:53 CEST] <rcombs> (for parsing sidx, that is)
[20:37:18 CEST] <rcombs> fun fact, they only parse the low 32 bits of the offset fields because 64-bit ints are pretty much impossible to work with in JS
[20:37:34 CEST] <rcombs> (I mean, you can store 2 32-bit ints and do math on them and all, but you can't e.g. use that as an array index)
[20:38:07 CEST] <wm4> lol
[20:43:50 CEST] <wm4> rcombs: yeah, I'd say just post a patch
[21:28:05 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:1f171de8cae9: mpegvideo_enc: Drop unnneded initialization
[21:28:06 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:16b68bd2ee5f: Merge commit '1f171de8cae9d0c527c5f0bae2a597dbd65fd3a4'
[21:34:27 CEST] <ubitux> anyone to propose libav to cherry-pick [t]blend filters instead of writing an half assed diff filter?
[21:37:12 CEST] <D404|Ghetto> nobody cares because nobody uses libav's labfi
[21:37:15 CEST] <D404|Ghetto> lavfi*
[21:38:29 CEST] <BtbN> I wonder what will happen with libav in the long turn, now that basicaly nobody will use new versions anymore
[21:39:04 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:95e2317ed855: roqvideoenc: Drop unneeded initialization
[21:39:05 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:6497aab022bf: Merge commit '95e2317ed85502dd8d96bcd9b12084dbfb8f9e8e'
[21:52:07 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:91767360d86a: a64multienc: Do not entangle coded_frame
[21:52:08 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:05fb4c9aaf84: proresenc: Do not entangle coded_frame
[21:52:09 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:a99f0ad194fd: Merge commit '91767360d86ac786571593ab11c7291010ab3829'
[21:52:10 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:f2a581e2eeec: Merge commit '05fb4c9aaf84b59f8ab1ce8d4c0f49dd12113024'
[21:52:12 CEST] <ubitux> already acked, great
[21:52:24 CEST] <ubitux> guess we'll have to deprecate it immediately and alias it to blend
[21:54:31 CEST] <jamrial> or just not merge it at all and make "diff" an alias for tblend
[21:54:45 CEST] <ubitux> that's what i meant by the alias
[21:55:00 CEST] <ubitux> drop the code, alias it to blend, and add a deprecated warning
[21:55:12 CEST] <jamrial> yeah, i was mainly saying to not merge it at all rather than merge, deprecate and drop
[21:55:25 CEST] <jamrial> but guess you implied the same anyway, heh
[21:55:44 CEST] <ubitux> i'd better have michaelni spend some time on more useful stuff than dealing with this in the merge
[21:57:22 CEST] <jamrial> he will merge the commit for metadata purposes anyway, but can make it so ultimately no code is changed
[21:58:34 CEST] <ubitux> i'd better have the code in, so we are not merging it in a hurry
[22:01:44 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:56672aeaee19: svq1enc: Do not entangle coded_frame
[22:01:45 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:033144c0d13c: Merge commit '56672aeaee19216d2fd8eeb9964b1f71f0af2919'
[22:13:28 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:7fc8d8a1b3da: libxvid: Do not entangle coded_frame
[22:13:29 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:d6006dd9f0d4: libvpxenc: Do not entangle coded_frame
[22:13:30 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:70c648db8948: Merge commit '7fc8d8a1b3da1666176b3d92e989ee74f3ef014e'
[22:13:31 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:96504b3f72a7: Merge commit 'd6006dd9f0d4d01023359230212f1f9fa4800e5b'
[22:21:52 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:f3b4a92241a3: libtheoraenc: Keep coded_frame.key_frame a write-only variable
[22:21:53 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:efc618aff9d6: qtrleenc: Keep coded_frame.key_frame a write-only variable
[22:21:54 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:bde7f5b135c7: Merge commit 'f3b4a92241a33d444f7f6018ebc12e2e3a2d335f'
[22:21:55 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:256430cf384d: Merge commit 'efc618aff9d68d2ddc323a5c5f892ac71951f162'
[22:27:55 CEST] <philipl> __gb__: Will anyone from Intel be adding vaapi-hwaccel support for hevc/vp8/vp9?
[22:33:07 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:10a9149de242: ffv1enc: Keep coded_frame.key_frame a write-only variable
[22:33:08 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:b97e3e11a9c2: Merge commit '10a9149de242c7bbc4e130d3d7c593b89e20f80e'
[22:48:23 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:75c1ed2e4c72: ffv1enc: Use input frame to set SAR and interlacing
[22:48:24 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:91f9b6579ac6: flashsvenc: Keep coded_frame.key_frame a write-only variable
[22:48:25 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:5cc3faf6198b: Merge commit '75c1ed2e4c7250f5e75a11d0140727054ff0136b'
[22:48:26 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:439c9760b046: Merge commit '91f9b6579ac684c4b51c4cd0dbaed0a4f8295edf'
[23:19:51 CEST] <rcombs> ubitux: https://ffmpeg.org/pipermail/ffmpeg-devel/2015-July/175805.html can ffprobe even use avpriv functions? (even if it could, I'd be against it)
[23:20:17 CEST] <rcombs> ubitux: though that function could be moved into ffprobe.c, I suppose
[23:20:25 CEST] <nevcairiel> technically it can,  but it really shouldnt
[23:21:32 CEST] <rcombs> though I might be misunderstanding what that's in re:
[23:22:26 CEST] <nevcairiel> but there isnt even a new function in that patch, how can it become avpriv
[23:26:19 CEST] <ubitux> ah you need it in ffprobe?
[23:26:45 CEST] <rcombs> nevcairiel: the next one adds a new function
[23:26:57 CEST] <rcombs> so I assumed that was what he meant
[23:27:12 CEST] <rcombs> but it's only used in ffprobe so I'm just moving it into ffprobe.c
[23:32:24 CEST] <jamrial> well, ffserver is using ff_ functions, even worse than avpriv :p
[23:33:41 CEST] <ubitux> yeah no avpriv in ffprobe
[23:34:08 CEST] <ubitux> av_ is fine as long as it's easy to keep compatibility and api is well defined and useful
[23:34:50 CEST] <rcombs> I just yanked it because it's not particularly useful
[23:35:01 CEST] <rcombs> now just building and testing
[23:44:31 CEST] <baptiste> rcombs, sidx patch looks great :)
[23:44:55 CEST] <baptiste> quick question though, AFAIR, the index timestamps are dts, not pts, is that correct ?
[23:44:59 CEST] <rcombs> baptiste: know what the correct solution for the ID matching?
[23:45:04 CEST] <rcombs> *is?
[23:45:11 CEST] <baptiste> I think your solution is good
[23:45:18 CEST] <rcombs> just +1?
[23:45:21 CEST] <baptiste> people use track id generally
[23:45:23 CEST] <rcombs> that felt wrong
[23:45:34 CEST] <baptiste> it's correct, mov starts at 1
[23:45:49 CEST] <rcombs> and every mov stream is guaranteed to have a corresponding AVStream?
[23:46:26 CEST] <baptiste> well no
[23:46:46 CEST] <rcombs> baptiste: re: DTS vs PTS, I don't believe it should matter here, since the relevant timestamps should always correspond to keyframes?
[23:47:05 CEST] <baptiste> it might matter for search I believe
[23:47:32 CEST] <rcombs> and AFAIK you can't ref between fragments
[23:47:45 CEST] <baptiste> you can search any timestamp, and you can remove the keyframe flag
[23:47:58 CEST] <durandal_1707> michaelni: how to display continuous yuv colors on 4096x4096 without any duplicate
[23:48:07 CEST] <rcombs> yes, and the AVIndex code handles all that
[23:48:25 CEST] <ubitux> durandal_1707: yuv444?
[23:48:27 CEST] <rcombs> but the fragment seek code just has to get you into the fragment containing your target timestamp
[23:48:52 CEST] <durandal_1707> ubitux: yes
[23:49:08 CEST] <baptiste> yes, if you search for dts 4 and return the fragment with pts 4, that might cause an issue
[23:49:50 CEST] <rcombs> can you have PTS 4 in one fragment and DTS 4 in another?
[23:50:02 CEST] <baptiste> I believe so
[23:50:46 CEST] <cone-809> ffmpeg 03Vittorio Giovara 07master:d6604b29ef54: Gather all coded_frame allocations and free functions to a single place
[23:50:47 CEST] <cone-809> ffmpeg 03Michael Niedermayer 07master:b1fad7ac200d: Merge commit 'd6604b29ef544793479d7fb4e05ef6622bb3e534'
[23:52:11 CEST] <baptiste> I just think it might not be a good idea to mix pts and dts in that list :)
[23:52:41 CEST] <rcombs> you mean in MOVFragmentIndex?
[23:56:37 CEST] <baptiste>                      e->timestamp = current_dts;
[23:56:42 CEST] <baptiste> AVIndexEntry
[23:57:18 CEST] <rcombs> I don't touch those, though
[23:58:27 CEST] <baptiste> yeah I notice now
[00:00:00 CEST] --- Tue Jul 21 2015

More information about the Ffmpeg-devel-irc mailing list