Commit Graph
68 Commits
Author SHA1 Message Date
itsmattkc 412ea9cede ffmpegdecoder: fixed bug with frame identification 2020-02-24 02:01:42 +11:00
itsmattkc 8f5a92af9c ffmpegframecache: removed unnecessary debug line 2020-02-24 02:01:21 +11:00
itsmattkc 6956e159e6 ffmpegdecoder: remove debug messages 2020-02-23 22:18:59 +11:00
itsmattkc d3bc12b2f5 ffmpegdecoder: use stricted memory interval 2020-02-23 22:18:10 +11:00
itsmattkc c7cb227abb ffmpegdecoder: fixed memory leak introduced in previous commit 2020-02-23 21:24:07 +11:00
itsmattkc b1cb8489df memory: try caching frameptrs instead of avframes so we have more control 2020-02-23 21:20:33 +11:00
itsmattkc 48cdb3a6dd memorymanager: rewrote entire memory management system 2020-02-23 20:29:51 +11:00
itsmattkc 547990b920 ffmpegdecoder/memorymanager: fixed race condition in garbage collection 2020-02-21 23:15:09 +11:00
itsmattkc b0c9711974 ffmpegdecoder: fixed double mutex lock
Error() would call Close() getting stuck in a double mutex lock that could
stall workers. This commit fixes that.
2020-02-21 20:04:36 +11:00
itsmattkc 6c33e85162 decoder: don't start indexing any stream that doesn't need to be 2020-02-21 20:04:08 +11:00
itsmattkc dcedf12dbe ffmpegdecoder: stop clear_timer on deconstruction 2020-02-21 14:25:27 +11:00
itsmattkc befaf34a4c oiiodecoder: use QStringLiteral()
Minor code improvement/optimization.
2020-02-21 14:21:53 +11:00
itsmattkc e74c4c2d1c global: implemented rudimentary memory manager to use available system memory
without forcing the OS to page swap
2020-02-21 14:20:57 +11:00
itsmattkc 2e38e7e3fb indexmanager: moved files 2020-02-21 10:47:41 +11:00
itsmattkc 7797bf4971 ffmpegdecoder: added helper functions for memory handling 2020-02-21 01:18:21 +11:00
itsmattkc 4532c3884e ffmpegdecoder: fixed cast issue on gcc 2020-02-21 00:21:31 +11:00
itsmattkc 2299fda73a ffmpegdecoder: handle media that starts at a non-zero pts 2020-02-21 00:09:45 +11:00
itsmattkc 162c91ec52 ffmpegdecoder: allow probe duration pickup to be cancelled 2020-02-20 23:54:40 +11:00
itsmattkc bc2be7a2a4 ffmpegdecoder: use simpler loop to find duration if the container doesn't
provide one
2020-02-20 23:52:23 +11:00
itsmattkc 73e639757e ffmpegdecoder: fix bug that slowed down low fps footage unnecessarily 2020-02-20 23:34:49 +11:00
itsmattkc 3bd1c885cf decoder: round double to int 2020-02-20 19:24:59 +11:00
itsmattkc e373c40853 ffmpegdecoder: handle eof in retrieve 2020-02-20 16:50:03 +11:00
itsmattkc 3e776b62e1 oiiodecoder: added new mutex structures 2020-02-20 16:38:23 +11:00
itsmattkc 1cce8abd85 frame: added bool function returning whether the frame is allocated or not 2020-02-20 16:37:43 +11:00
itsmattkc 2145efa2a0 decoder/renderer: use memory cache and single decoder to optimize frames for
cycles
2020-02-20 15:32:12 +11:00
itsmattkc 71048f40a8 ffmpegdecoder: removed all old code to clean up 2020-02-20 02:22:59 +11:00
itsmattkc 95ca19d547 ffmpegdecoder: began work on the index process working in tandem with the
retrieval process
2020-02-20 00:06:20 +11:00
itsmattkc e68e89c40b renderer/decoder: use same background mechanism to conform audio as well 2020-02-19 16:45:34 +11:00
itsmattkc a719618b31 waveinput: better handling of WAV data sections 2020-02-19 12:45:21 +11:00
itsmattkc 56adf5005f decoder: provide conform progress signal 2020-02-19 12:45:01 +11:00
itsmattkc 08e6c6ffc0 decoder: move conform function to base class
There's no reason the Conform() function can't live in the base class since its
functionality isn't necessarily specific to FFmpeg.
2020-02-19 12:28:37 +11:00
itsmattkc 53551246da decoder: signal indexing progress
Better UI interactions, actually shows the progress of the index function in
the task manager now.
2020-02-19 12:19:42 +11:00
itsmattkc 03bdf01357 decoder/renderer: no longer index automatically as part of the retrieve
functions

Indexing is a lengthy process and had a high chance of getting RenderWorkers
stuck doing it rather than being responsive to cache requests. This commit
introduces a system where workers never index media, but instead signal that
media is not ready to their RenderBackends which ensure that the media gets
indexed and re-queues the affected frames when those indexes are ready.
2020-02-19 11:44:29 +11:00
itsmattkc 5e75e9b3b0 waveinput/output: allowed external access to data length value 2020-02-19 11:40:54 +11:00
itsmattkc 22046535ca various: added hooks so that various lengthy functions could be cancelled from another thread 2020-02-17 02:10:12 +11:00
itsmattkc 35ab379e9f ffmpegdecoder: updated comment to better describe error handling 2020-02-16 10:28:18 +11:00
itsmattkc cb3ef5d994 decoder: ensure indextask is started in the taskmanager's thread 2020-02-16 10:27:23 +11:00
itsmattkc cdfba418fc ffmpegdecoder: added fixme comments for errors that aren't handled 2020-02-16 00:45:36 +11:00
itsmattkc 3efbcc28ef ffmpegdecoder: improved index reliability 2020-02-15 23:39:33 +11:00
itsmattkc f5de19dd6b ffmpeg: allow decoders immediate access to frames even while indexing
Previously, FFmpeg decoders needed to wait for the initial index to finish
before they could retrieve frames. Now they can retrieve frames while the
index is occurring, provided the appropriate frame has already been indexed (if
not, the retrieve will need to wait still).
2020-02-15 13:57:29 +11:00
itsmattkc 9de164cfba ffmpeg: disable multithreading in favor of our own
Our renderer system is already heavily multithreaded and prioritized to keep the
main thread active. As a result, FFmpeg receives little benefits from being
multithreaded and is actually detrimental to our main thread as it hasn't been
prioritized to keep the main thread as active as possible.
2020-02-14 10:08:19 +11:00
itsmattkc 5d58479f0a oiiodecoder: fixed erroneous processing of extension_list 2020-01-26 14:43:45 +11:00
itsmattkc 490a634473 oiiodecoder: fixed segfault caused by some OIIO decoders if they received
unexpected data

When "probing" footage, Olive runs it past all of its decoders until one
responds that it can decode this file. However this caused issues when some
OIIO decoders would erroneously pick up incompatible files, try to read them
and segfault the entire app (notable OpenJPEG with MPEG-4 and RLA with WAVE
audio). We now use the file extension to check with OIIO if it "should" be
compatible before actually testing if it is.

This is not a "perfect" solution (i.e. someone could recreate the segfault by
renaming an MPEG-4 file to an image extension like .JPG), but is probably as
much as can be done Olive-side and should filter out all segfaults under normal
circumstances.
2020-01-26 11:54:49 +11:00
itsmattkc 8b0c7edfec OIIO: added support for OIIO 1.x
OIIO 2.x uses std::unique_ptr while 1.x uses raw pointers. Olive can now handle
both, manually destroying the raw pointers when necessary if running on OIIO
1.x.
2020-01-26 11:33:45 +11:00
itsmattkc 149d125e39 msvc: updated code for higher warning level 2020-01-24 12:00:55 +11:00
itsmattkc c0300371c7 ffmpegdecoder: fixed flawed debug line 2020-01-20 04:58:39 +11:00
itsmattkc 4697b98980 fixed gcc compile issues 2020-01-19 20:39:48 +11:00
itsmattkc 1101b58ee8 allow setting of pixel format/ocio method/sample format in preferences
Implementation isn't perfect yet, viewer/renderer doesn't update yet when
the preference is changed so a sequence needs to be re-opened for the change to
take effect.
2020-01-18 03:37:36 +11:00
itsmattkc f8f932112c fixed bug where the node remove with exclusive deps command would never run 2020-01-17 13:56:45 +11:00
itsmattkc 653709dc84 try to index media immediately on import
Since indexing is supplemental, it can be made a background task that can occur
at more or less any time. So we run it as soon as possible.
2020-01-17 03:58:29 +11:00