Completely revamped the entire media i/o data and handlers. The
original idea was to have a system that would have connecting media
inputs and outputs, but at a certain point I realized that this was an
unnecessary complexity for what we wanted to do. (Also, it reminded me
of directshow filters, and I HATE directshow with a passion, and
wouldn't wish it upon my greatest enemy)
Now, audio/video outputs are connected to directly, with better callback
handlers, and will eventually have the ability to automatically handle
conversions such as 4:4:4 to 4:2:0 when connecting to an input that uses
them. Doing this will allow the video/audio i/o handlers to also
prevent duplicate conversion, as well as make it easier/simple to use.
My true goal for this is to make output and encoder plugins as simple to
create as possible. I want to be able to be able to create an output
plugin with almost no real hassle of having to worry about image
conversions, media inputs/outputs, etc. A plugin developer shouldn't
have to handle that sort of stuff when he/she doesn't really need to.
Plugins will be able to simply create a callback via obs_video() and/or
obs_audio(), and they will automatically receive the audio/video data in
the formats requested via a simple callback, without needing to do
almost anything else at all.
When the first async video frame is used it would not set audio timing,
moved that code into obs_source_getframe. Also, might consider renaming
obs_source_getframe. "Query frame" instead perhaps? Will sleep on it,
might not even bother.
- Add preliminary (yet to be tested) handling of timestamp invalidation
issues that can happen with specific devices, where timestamps can
reset or go backward/forward in time with no rhyme or reason. Spent
the entire day just trying to figure out the best way to handle this.
If both audio and video are present, it will increment a reference
counter if video timestamps invalidate, and decrement the reference
counter when the audio timestamps invalidate. When the reference
counter is not 0, it will not send audio as the audio will have
invalid timing. What this does is it ensures audio data will never go
out of bounds in relation to the video, and waits for both audio and
video timestamps to "jump" together before resuming audio.
- Moved async video frame timing adjustment code into
obs_source_getframe instead so it's automatically handled whenever
called.
- Removed the 'audio wait buffer' as it was an unnecessary complexity
that could have had problems in the future. Instead, audio will not
be added until video starts for sources that have both async
audio/video. Audio could have buffered for too long of a time anyway,
who knows what devices are going to do.
- Fixed a minor conversion warning in audio-io.c
- In the audio I/O code, if there's a pause in the program or its
threads (especially the audio thread), it'll cause it to sample too
much data, and increase line->base_timestamp to a potentially higher
value than the next audio timestamp that may be added to the line.
This would cause it to crash originally, because it expects audio
data that is within the designated buffering limit.
Because that audio data cannot be filled by that data anyway, just
ignore the audio data until it goes back to the right timing (which
it will as long as the code that is using the line accounts for its
current system time)
- Often, timestamps will go "back" in time with certain.. terrible
devices that no one should use. When this occurs, timing is now
reset so that the new audio comes in directly after the old audio
seamlessly.
- Audio data was just being popped to the "front" of the mix buffer, so
instead it now properly pops into the correct position in the mix
buffer (proper mixing still needs to be implemented)
- Added a test audio sinewave test source that should just play a sine
wave of the middle C note. Using unsigned 8 bit mono to test
ffmpeg's audio resampler, seems to work pretty good.
- Fixed a boolean trap in threading.h for the event_init function, it
now uses enum event_type, which can be EVENT_TYPE_MANUAL or
EVENT_TYPE_AUTO, to specify whether the event is automatically reset
or not.
- Changed display names of test sources to something a little less
vague.
- Removed te whole "if timestamp is 0 just use current system time"
when outputting source audio, if you want to use system time you
should just use system time yourself. Using 0 as some sort of
"indicator" like that just makes things confusing, and prevents you
from legitimately using 0 as a timestamp for your audio data.
- Circular buffer code wasn't correctly handling the splitting of
newly placed data segments, the code was untested and turned out to
just be backwards. It now copied the data to the back and front of
the buffer properly.
For one, I added a new member gs_window for future use.
The member is "display" which represents our connection to X11.
Ideally, we should use this specific connection to deal with our Window.
For now, it's disabled. Read comment for more information.
Secondly, wxGTK apparently doesn't map our window in some cases.
This causes the window ID passed to be bad and will stop (or segfault)
our program. This might be related to the first commit above.
For now, all this commit does is realize the window manually.
- Mixing still isn't implemented, but the audio system should be able
to start up, and mix at least once audio line for the time being.
Will have to write some test audio sources to verify things are
working properly, and build the rest of the output functionality.
- Using a recursive mutex fixes issues where objects need to enter the
main libobs sources mutex while already within the mutex in the same
thread. Otherwise it would keep getting locked on itself on
destruction.
- Apply the volume specified with the audio data packet before
inserting the audio data into the circular buffer. Added functions
for multiplying the volume with all the different audio bit depths.
(Could probably be greatly optmimized later)
- Added a volume variable to the obs_source structure and implemented
functions for manipulating source volume.
- Added a volume variable to the audio_data structure so that the
volume will be applied when mixing.
- Made it so that when a source is added or removed from a scene it
will add a reference to sourceSceneRefs (std::unordered_map). Each
source adds a reference to that every time they are added to a scene,
and releases a reference from it when they are removed from a scene.
When the value reaches 0, the source is no longer in any scenes, and
is then marked for removal and destroyed.
Before, I was using the source internal reference counter, which is a
really bad thing to do because I don't know what might actually be
referencing it. So using a separate discrete reference counter for
the number of scenes it's in is better in this case.
- Check the actual debug extension instead of checking GL 4.0 support;
it appears that despite having GL 4.0 support, AMD does not implement
the standard debug extensions even when the *_CONTEXT_DEBUG_BIT_ARB
flag is set on the context.
- Also, fixed a place where spaces where used for indents instead of a
tab.
- Removed the dependency on windows.h for windows. I feel it's an
unnecessarily large dependency to have to add to all source files
when the only thing that's needed to make the windows version compile
the debug functions is just the __stdcall call convention keyword.
On top of increasing compile time due to the large number of headers
it includes from all the windows API headers, it also adds a lot of
potential name conflicts, as I was getting a number of name conflicts
for lots of names like near/far, which were used in old legacy 16bit
windows code.
fixes a deadlock in obs_free_video/obs_video_thread where
video_output_stop would signal the update event before obs_video_thread
enters video_output_wait (the thread calling obs_free_video would
block on pthread_join and obs_video_thread would block on
pthread_cond_wait)
- Added an extra comment regarding source removal
- Made is so that when a scene item is removed from the sources list
box, it removes the item based upon the GetClientData rather than by
its string name. This is done because there may be duplicate sources
in a particular scene, and it's important to remove the right one.
- When the remove source tool icon is clicked, it will now remove the
source from the scene.
- Fixed a bug where the scene item removal callback would add the scene
item to the list instead of removing it.
- Changed AddSourcePopup to AddSourcePopupMenu. Name actually confused
me once despite being the writer, so it was clearly a bad name.
Sources can now be added to scenes via user interface. It's a little
convoluted because everything has to work through OBS signals to ensure
that plugins/etc can modify the scenes/sources exernally.
Also, when switching scenes, it will properly list sources for the
scene you changed to.
The WX_UTF8 macro is primarily for convenience so you can input a
UTF-8 string into a wxWidgets function without having to do
wxString(string, wxConvUTF8) each time.
Palana
jp9000
computerquip
Ján Mlynek
Benjamin Klettbach