dcpomatic/src/lib, branch v2.11.56 DCP-o-matic DCP tools https://git.carlh.net/cgit/dcpomatic/atom?h=v2.11.56 2018-02-16T21:57:13Z AudioDecoder::_positions is at the resampled rate, if applicable; 2018-02-16T21:57:13Z Carl Hetherington cth@carlh.net 2018-02-16T21:54:01Z urn:sha1:7147957ea5485a70580135cf32c43a05ed97a7f2 fix its initial setup. Fixes problems when seeking deep into resampled content; before this commit the audio time stamps got out of whack sufficiently that the butler would fill with video before any audio made it past the range checks. Make Config::read() fallback match Config::set_defaults() for 2018-02-13T01:08:14Z Carl Hetherington cth@carlh.net 2018-02-13T01:08:14Z urn:sha1:08feee25150ff15bad67087c80e44ba21e6ebd0d _default_dcp_content_type. Add SWS_ACCURATE_RND to sws flags. Without this, conversions to RGB in 2018-02-12T23:47:02Z Carl Hetherington cth@carlh.net 2018-02-12T23:47:02Z urn:sha1:b17a5963f9c79e3e99890f32129105dd49fb0c29 crop_scale_window_test leave the 6 pixels at the right side of the image uninitialised (sws_scale doesn't fill them in with anything) so you get garbage in the output. SWS_ACCURATE_RND fixes it. There are whisperings that this is slow, so the impact of this patch may need to be measured in some "real" situations. Tone some debug messages down slightly. 2018-02-12T23:21:34Z Carl Hetherington cth@carlh.net 2018-02-12T23:21:34Z urn:sha1:d7f731ef2a3e099cf7b9af1de514e8b56b71dfa1 Save decoding resolution reduction between launches (#1195). 2018-02-10T23:00:15Z Carl Hetherington cth@carlh.net 2018-02-10T22:43:09Z urn:sha1:0af38a850f967fd1848850a18432abc6314e9cc4 Use av_guess_frame_rate rather than av_stream_get_r_frame_rate; 2018-02-10T21:39:18Z Carl Hetherington cth@carlh.net 2018-02-10T21:39:18Z urn:sha1:17f9254f68fabfe5d8136d256b69a65fc58724ee helps with #1030. I hope this isn't a bad idea; this commit should be reverted if there any hints of misread video frame rates. Fix odd behaviour when analysing single bits of audio which start at non-zero positions (#1203). 2018-02-10T02:05:14Z Carl Hetherington cth@carlh.net 2018-02-10T02:05:14Z urn:sha1:9d4906db6e143248cf1e814d539aa2c7a50721dc Always pre-roll when seeking in DCPs; I don't think the performance 2018-02-09T17:07:16Z Carl Hetherington cth@carlh.net 2018-02-09T17:07:16Z urn:sha1:11a6decd652158b2288de17dc1aeafcd860fdfef hit is anything to worry about (especially compared to the time taken to decode the next frame). Updated ru_RU translation from Igor Voytovich. 2018-02-09T01:00:37Z Carl Hetherington cth@carlh.net 2018-02-09T01:00:37Z urn:sha1:10b5f18a721d9c6cde3da3562158a4726fd24d65 typeid() doesn't seem to downcast, so just check types in the take_settings_from methods (#1192). 2018-02-08T20:57:03Z Carl Hetherington cth@carlh.net 2018-02-08T20:57:03Z urn:sha1:d9825793692a41374b184f423994667b663d8526