[Ffmpeg-devel-irc] ffmpeg-devel.log.20150614

[burek]

[01:10:42 CEST] <cone-077> ffmpeg 03Timo Rothenpieler 07master:f17d973413ec: avfilter/vf_colorkey: Add colorkey video filter
[03:01:37 CEST] <cone-077> ffmpeg 03Michael Niedermayer 07master:612f8cae0640: avcodec/jpeg2000dec: Support RPCL
[03:29:06 CEST] <cone-077> ffmpeg 03James Almer 07master:20abb1afe5d7: softfloat: make av_div_sf() inline
[03:46:52 CEST] <cone-077> ffmpeg 03Michael Niedermayer 07master:4a3bc0d9a5a0: avcodec/jpeg2000dec: Implement "Reset context probabilities on coding pass boundaries"
[05:28:53 CEST] <philipl> Hilarious observation number 2 - marking a long term reference as a long term reference results in glitches.
[05:29:04 CEST] <philipl> The nvidia example app doesn't set IsLongTerm at all.
[05:59:35 CEST] <philipl> nevcairiel: http://fate-suite.ffmpeg.org/hevc-conformance/LTRPSPS_A_Qualcomm_1.bit
[05:59:44 CEST] <philipl> Does this one work correctly for you with dxva?
[05:59:57 CEST] <philipl> There are glitches in here that show up in both my code and the nvidia example.
[06:00:28 CEST] <philipl> POC 223 in particular. You can actually see mis-handling of a long term ref (to POC 192).
[06:09:47 CEST] <cone-077> ffmpeg 03Philip Langdale 07master:01fac84fe435: avcodec/nvenc: Enable YV12 input format
[06:09:48 CEST] <cone-077> ffmpeg 03Philip Langdale 07master:1b19d0c63281: avcodec/hevc: Track long and short term RPS size for VDPAU
[10:22:25 CEST] <nevcairiel> philipl: that passes fine
[10:30:15 CEST] <nevcairiel> i never really tested LTs at all, but they are just implemented the same as STs in DXVA2 with another flag attached to them, so apparently it works.
[11:21:17 CEST] <cone-586> ffmpeg 03Clément BSsch 07master:1dee7a31e050: avfilter: bump minor after colorkey addition
[13:01:58 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:132bf0bbb7dc: avcodec/jpeg2000dec: Print all cdx/cdy values not just the 2nd and 3rd
[13:01:59 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:ad072a134fc0: avcodec/jpeg2000dec: Handle format that has planes matching YUVA420P structure but seems not to be that
[15:10:10 CEST] <cone-586> ffmpeg 03Andreas Cadhalpun 07master:ed0b1db64011: doc: fix spelling errors
[15:16:53 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:742a26c4898f: avcodec/jpeg2000dec: Print error messages for more error conditions
[16:50:22 CEST] <nevcairiel> Sometimes I wonder why random people can even set the priority and the like :p
[16:57:00 CEST] <ubitux> pigoz: yes someone is on VT and is going to submit a patch soon
[19:03:48 CEST] <cehoyos> michaelni: Hi, Isn't p0_05.j2k CMYK?
[19:04:13 CEST] <michaelni> could be
[19:06:22 CEST] <cehoyos> I used extractplanes and all four planes did not look like like transparency information to me.
[19:06:40 CEST] <cehoyos> I am just guessing here.
[19:13:38 CEST] <Daemon404> open in ot photoshop and see?
[19:14:44 CEST] <cehoyos> I don't have photoshop here but I can show you command lines to produce four gray png that are apparently YCbCrW 4:2:0
[19:16:46 CEST] <cehoyos> (That's at least how Fabrice calls it, I cannot find another reference for "YCbCrW")
[19:22:59 CEST] <cone-586> ffmpeg 03Paul B Mahol 07master:20ee65ef73b7: avfilter/af_afade: use av_clipd() instead of nested FFMAX & FFMIN
[19:25:49 CEST] <gouessej> Hi
[19:26:07 CEST] <gouessej> I can't use the FFMPEG FTP account
[19:26:15 CEST] <gouessej> I get the following error: 421 Too many connections (1) from this IP
[19:27:06 CEST] <Daemon404> you tried to connect too many times
[19:27:10 CEST] <Daemon404> wait a little bit and try again
[19:29:16 CEST] <cehoyos> Or use http://www.datafilehost.com/
[19:30:20 CEST] <gouessej> @cehoyos I have succeeded to send the text file :s
[19:30:35 CEST] <gouessej> @cehoyos Which version of Totem do you use?
[19:32:32 CEST] <cehoyos> An ancient one but I don't think that matters: You are (that is what I guess, I simply cannot reproduce) seeing bugs in some applications and are reporting them on the wrong bug tracker.
[19:34:49 CEST] <gouessej> I use ffmpeg to convert the mp4 files into ogv
[19:35:05 CEST] <Daemon404> this probbably belongs in #ffmpeg fwiw
[19:35:39 CEST] <cehoyos> gouessej: just post your command line and console output on the bug tracker and I will try to reproduce.
[19:37:02 CEST] <gouessej> @cehoyos: When I use the same command line than you with the latest static build, I have the same problem, Totem doesn't play the resulting ogv file
[19:37:11 CEST] <cehoyos> gouessej: just post your command line and console output on the bug tracker and I will try to reproduce.
[19:42:28 CEST] <gouessej> @cehoyos Done. Thanks. What more can I do?
[19:56:56 CEST] <pigoz> ubitux: nice, dilaroga already posted patches to mpv.
[19:57:40 CEST] <ubitux> ah great; he sent me a mail today about it
[21:17:34 CEST] <wm4> that ticket is just going to be like the aac one, isn't it
[21:28:50 CEST] <jamrial> no. the aac one at least should end up being productive
[21:36:34 CEST] <nevcairiel> or so we hope
[21:36:42 CEST] <nevcairiel> that one here is just silly
[21:46:51 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:91c8025c44ec: avcodec/jpeg2000dec: Show the progression order at debug level for all variants
[21:46:52 CEST] <cone-586> ffmpeg 03wm4 07master:2222f419da99: tls_gnutls: fix hang on disconnection
[21:46:53 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:6b08dc393b52: avcodec/jpeg2000dec: Print warning if the mqc pointer mismatches at the end
[22:45:30 CEST] <ubitux> boring easy task if looking for an easy first contrib: check the ff_formats_ref() calls in lavfi
[22:45:57 CEST] <ubitux> if +someone
[22:46:02 CEST] <ubitux> +is
[23:00:42 CEST] <durandal_1707> ubitux: still writing that filter?
[23:00:49 CEST] <ubitux> yeah
[23:00:55 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:59c107292c75: avcodec/jpeg2000: Fix band coords
[23:00:56 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:d9841e53e2eb: avcodec/jpeg2000: Change precinct coords to same origin as bands
[23:00:57 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:a206c13289e6: avcodec/jpeg2000: Change codeblock coords to same origin as bands
[23:00:58 CEST] <ubitux> first draft almost done
[23:00:58 CEST] <cone-586> ffmpeg 03Michael Niedermayer 07master:ec5164aa310e: avcodec/jpeg2000: Fix regression with multiple tiles
[00:00:00 CEST] --- Mon Jun 15 2015