Commit Graph

41 Commits

Author SHA1 Message Date
Jonathan Thomas
0c2008461a Fixed seek method to fallback on both streams, if one stream fails to seek. This fixes MP3s... which have a video stream, but only a single video packet. Also fixed the CheckWorkingFrames method to better determine when a frame is completed. 2012-07-09 00:41:17 -05:00
Jonathan Thomas
dfae7825ca Refactored FFmpegReader code a bit, fixed many, many memory leaks, and many valgrind errors. Now the memory seems very stable, and the seeking and stream walking seems very solid. 2012-07-08 23:26:44 -05:00
Jonathan Thomas
8067cb657a Fixed a bug when seeking, and not reseting the cache's current frame. Removed some debug code. 2012-07-07 00:58:05 -05:00
Jonathan Thomas
e9c243ddeb Audio optimizations and some debug code for seeking 2012-07-06 16:52:13 -05:00
Jonathan Thomas
21aa4d6b01 Replaced last_video_frame and last_audio_frame with a single last_frame variable, and refactored some related code. This fixes a big audio seeking bug. 2012-07-06 15:17:57 -05:00
Jonathan Thomas
457df5b2ff Changed the method of determining if a frame is completed or not. 2012-07-06 02:34:18 -05:00
Jonathan Thomas
e610d343f4 Added some debug messages about when a frame is "complete" 2012-07-05 02:29:43 -05:00
Jonathan Thomas
25b3d70a9a Refactored the cache class to only apply the max frames against "previous" frames, and allow unlimited number of future frames, to account for different types of files (some which have hundreds of tiny frames, which need large future cache). 2012-07-05 00:01:42 -05:00
Jonathan Thomas
53e16efd63 Fixed a small audio buffer issue, and removed more debug code. 2012-07-04 04:01:59 -05:00
Jonathan Thomas
34a07b8d82 All known memory leaks are now fixed, and the memory seems to hold steady when looping through frames for many many minutes at a time. Fixed a bug with av_free_packet, which was breaking after an audio decode. Removed lots of debug statements. 2012-07-04 03:07:26 -05:00
Jonathan Thomas
e8d9dba8ce Started adding clean up code, to reduce memory footprint 2012-07-03 16:58:07 -05:00
Jonathan Thomas
6553f49874 Made CheckSeek() threadsafe 2012-07-03 02:59:38 -05:00
Jonathan Thomas
a662abcfd7 Fixed many bugs, including an audio crackle between packets, and fixed many multi-threading issues. 2012-07-03 02:42:47 -05:00
Jonathan Thomas
eea7b0c076 Protected audio decoding in a omp_critical, and removed some debug code 2012-07-02 17:25:07 -05:00
Jonathan Thomas
ed6b50709d More OpenMP debug code 2012-07-02 01:25:32 -05:00
Jonathan Thomas
fde25cec5a Lots of experimental changes with the FFmpegReader, and caching of packets. 2012-07-02 00:51:10 -05:00
Jonathan Thomas
815204d235 1) Successfully got OpenMP working with the video processing method.
2) Added new "processing" flag, to help determine if a frame is being worked on in a thread
3) Added a new cache Display() method, for debugging reasons
2012-07-01 01:43:06 -05:00
Jonathan Thomas
311c04de23 Another openmp tweak 2012-06-29 02:32:19 -05:00
Jonathan Thomas
40680f7201 Fixed a few openmp bugs in the FFmpegReader, but there are still many issues 2012-06-29 02:02:12 -05:00
Jonathan Thomas
62df1bc7ac openmp changes 2012-06-18 09:26:14 -05:00
Jonathan Thomas
b775fff325 Got ffmpeg and juce compiling and mostly everything working again. FFmpeg is the newest version, and so is JUCE. 2012-06-16 02:12:48 -05:00
Jonathan Thomas
063c1481bd Fixed a bug calculating # of samples per frame, to use FPS instead of video time base. 2012-02-26 17:22:24 -06:00
Jonathan Thomas
eae0c0bbfb FIxed a bug related to the CheckFPS method, if a video is too short to use this method, it needed to exit the function, and seek back to frame 1. 2012-02-26 16:45:50 -06:00
Jonathan Thomas
403c7e512e Fixed 2 bugs related to audio samples. Each frame can now determine it's own # of samples, due to some sample rates not being divisible evenly into some frame rates. Also, a bug was fixed that injected a bad sample into a frame's samples, when a frame received samples from multiple audio packets. 2012-02-26 16:40:53 -06:00
Jonathan Thomas
ba0245ac82 fixed a bug where resampled audio was not completly copied into the audio buffer. 2011-12-21 16:49:33 -06:00