FFmpeg YUV to RGB conversions require extra destination padding.
The Mantis issue might have been fixed when some YUV format conversions
were moved to the GPU, but this may fix other YUV formats.
Fixes https://obsproject.com/mantis/view.php?id=1177
This brings the script in the repo up to date with what is being
used to generate release builds for OBS Studio on Windows, as
well as fixing up a few issues.
If an image source isn't visible there is no point polling for changes
during the time it's hidden as no one will see them. Should the image
be on unreachable media such as an unavailable network share, this
check would cause the graphics thread to periodically stall trying to
detect changes of a non-visible image source.
If the image is modified while it's hidden, the updated file will now
be loaded as soon as the source is made visible - the update timer
intentionally ticks up while hidden.
Partially fixes some of the behavior described in
https://github.com/obsproject/obs-studio/issues/3007, a full fix
likely requires significant changes to support OS-specific file change
notification APIs.
Modules are now unloaded while OBS core is still active, allowing
modules that call into libobs in their unload function to continue to
work. This changes the behavior of such calls - previously they were a
no-op since the OBS core would be gone, but they are now executed with
the OBS core still being active.
libobs functions check for a null obs_core inconsistently. If the core
is null, the functions silently return with no indication of an error to
the API user. This commit removes all null checks in libobs functions
that require libobs to be initialized. For OBS Studio, we are
(hopefully!) using the API properly so this should have no impact.
Use general layout and 0 access masks for external synchronization as
specified in the spec.
Also set pipeline stages for maximum synchronization just in case
because it doesn't seem like the pipeline stages are specified.
"Vulkan-incompatible APIs will require the image to be in the GENERAL
layout whenever they are accessing them."
"Whilst it is not invalid to provide destination or source access masks
for memory barriers used for release or acquire operations,
respectively, they have no practical effect. Access after a release
operation has undefined results, and so visibility for those accesses
has no practical effect. Similarly, write access before an acquire
operation will produce undefined results for future access, so
availability of those writes has no practical use. In an earlier version
of the specification, these were required to match on both sides - but
this was subsequently relaxed. These masks should be set to 0."
Batch files into clang-format to prevent needless processes. Now its
twice as fast in real time and uses 4x less resources. The old
implementation is also slower (but pretty close to) running a single
invocation of clang-format on all files.
before
time ./formatcode.sh
real 0m3.860s
user 0m20.975s
sys 0m6.694s
after
time ./formatcode.sh
real 0m1.486s
user 0m5.426s
sys 0m0.203s
This commit adds a basic check to see if the v4l2 stream requires a
restart. This is primarly implemented on the basis of the fact that
the options requiring a restart have values stored in the v4l2_data
structure. A restart is only needed if any of the values have changed
since the last update.
Since libobs internally rounds to various powers of 2 to support SSE
functions, resolutions that are too low can get rounded to zero. libobs
will fail to startup with a zero resolution, so if a user accidentally
entered anything that rounded to zero OBS would become unusable without
manual fixing by editing the profile .ini.
All strings were treated as partial matches before, which caused a
false positive with any executable beginning with "time", notably
affecting the game "Timelie" which used Timelie.exe.
