Those functions are actually causing more problems than
helping out with the preview because the width/height is
updated immediately while the actual size of the frames
displayed changes later.
Due to the plugin creating a thread to retrieve and output the
captured image data, care must taken to not modify data the
thread reads from the outside while the thread is running.
In previous revisions some settings accessed by the capture
thread were written to in the update function which could cause
image corruption and in the worst case crashes.
The members of the data struct are now split into two groups,
those that are used by the thread while it is running and must
not be changed from the outside, and those can be changed at any
time.
This adds a function to prepare the source_frame struct for use
with obs_source_output_video. Since all of the values except for
the timestamp and data pointers are known in before it makes
little sense to compute them over and over again.
Due to the fact that v4l2 uses a single continuous memory segment
for multi planar formats we can also precompute memory offsets for
the planes.
By using the wrapper functions supplied by libv4l2 we gain
support for formats not natively supported by obs.
The library intercepts certain system calls to transparently
enable recoding.
Until a proper fix is found to support devices that use stepwise
or continuous values for framesize and framerate this adds fixed
values that can be selected.
For devices that support discrete values those are still queried
and used.
Autoselection is only implemented for frame rate as it doesn't make
sense to autoselect the video format (Any fills the role of library
supplied autoselection) or the resolution (it may make sense to
implement resolution matching based on the current transformation
settings in the future though)
Use warn/info/debug helper macro functions to output
warnings/information/debug log data that's preformatted to include the
module and the current output name.
NOTE: In texture_setimage, I had to move variables to the top of the
scope because microsoft's C compiler will give the legacy C90 error of:
'illegal use of this type as an expression'.
To sum it up, microsoft's C compiler is still utter garbage.
Instead of using strings for setting names, use macros. Macros prevent
mis-spellings, prevent usage of the wrong name, and if you get it wrong,
it will simply fail to compile, helping to ensure that the right setting
value is being used.
Similar to the shader functions, the effect parameter functions take
the effect as a parameter. However, the effect parameter is pretty
pointless, because the effect parameter.. parameter stores the effect
pointer interally.
Since adding a source now opens the config dialog for that source
the convenience gain from having a device automatically start capturing
is outweighed by the inconvenience from having captured images
unintentionally recorded/streamed
The locale parameter was a mistake, because it puts extra needless
burden upon the module developer to have to handle this variable for
each and every single callback function. The parameter is being removed
in favor of a single centralized module callback function that
specifically updates locale information for a module only when needed.
Having the value stored here is somewhat pointless, so this is one step
in fixing the locale handling. Locale should be handled by the modules
themselves with their own loaded locale lookup information.
Due to the recent change with matrices, the way space is represented by
matrices is a bit different. The matrix stack represents the view
matrix, rather than the model matrix, so the position is more
representative of the camera (view) itself, rather than that of the
objects (model).
Common services that require specific encoder settings (specified in the
'recommended' section of the json file) will now have those encoder
settings applied before the encoders start up.
This fixes the issue with services like twitch that weren't getting
their required 2-second keyframe maximum interval set.
Apparently, despite the fact that Apple added
kAudioHardwarePropertyTranslateUIDToDevice in 10.8, it's not actually
usable in 10.8, only 10.9. So, instead of being able to use it like a
normal, sane person, we have to enumerate all devices manually and find
the AudioDeviceID ourselves. A slight annoyance and a mark against
apple's competence, but audio devices should now be working again on
10.8 at least, so whatever.
fryshorts
benklett
jp9000
Palana
BtbN
Anry
Christoph Hohmann
Azat Khasanshin