When changing a subproject (e.g. graphics-hook), the generated binaries
are not installed without also building win-capture. This commit enables
the subprojects to be treated as "plugins" which are installed to rundir
and also the installation directory without any additional workarounds.
The `add_target_resource` function uses unnecessary path components
when copying files to the rundir, resulting in files added that way
not to end up where OBS expects them to.
The generated binaries created by sub-targets also need to be copied
as part of `win-capture`'s data files, which was easy to fix as the
target exists before CMake switches into the subdirectories.
Doesn't make sense to grab the oldest frame when not capturing overlays,
because the overlay was already rendered in it. This will grab the
latest frame (before the overlay is rendered).
Use D3D12_RESOURCE_STATE_PRESENT for CreateWrappedResource arguments. We
should assume the application will set the backbuffer to PRESENT, and
expect that to be preserved across the Present() call.
Remove unnecessary calls to ReleaseWrappedResources.
Clean up (void**) casts with modern patterns.
Previous fix attempted to place swap_recurse count in gl_data struct,
where it could be unexpectedly memset to 0. Separate member from struct,
and make thread-local for additional safety.
Attempt to schedule shared texture copies against the command queue that
the game's swap chain uses to try to reduce artifacts. The heuristics
for obtaining the queue are not perfect, so provide a toggle to use the
previous behavior.
Modify game capture shared textures to be typeless if they could
potentially need SRGB and non-SRGB views in the future.
These capture APIs have been updated: D3D 10/11/12, Vulkan.
D3D8 capture does not use shared textures.
D3D9 and GL interop do not support typeless textures.
The new game capture DLL should be compatible with old versions of OBS.
Also removed a lot of dead code around pointless SRV/RTV support.
For game capture, hook DXGI release function to release D3D objects if
the related swap chain is also being destroyed.
An added bonus is that the game capture hook will handle swap chain
recreation for applications that don't use ResizeBuffers.
Rivatuner Statistics Server calls vkGetInstanceProcAddr for
vkCreateDevice with a NULL instance. According to spec, this should
return NULL, but RTSS will dereference this and crash.
This change returns function pointers for every intercepted function
regardless of instance, which is enough to keep RTSS alive.
Prevents multiple queues in the same family from racing on frame data.
Removal of vk_family_data also fixes broken capacity calculation. Two
queue families can share the same index if one is "protected-capable"
and the other isn't.
Make fixes mostly to stabilize CTS. Tested on Windows with RTX 2080 Ti.
Use separate linked-list maps for devices and queues.
Process queues upfront in OBS_CreateDevice, and not lazily in
OBS_GetDeviceQueue.
Handle failed node allocation for OBS_CreateInstance, OBS_CreateDevice,
and OBS_CreateSwapchainKHR.
Handle VK_NULL_HANDLE for OBS_DestroySurfaceKHR and
OBS_DestroySwapchainKHR.
Add temporary code path CTS_WORKAROUND to deal with CTS sharing surfaces
between instances.
Fix return value for OBS_DestroyInstance.
Add missing _freea calls, required for _malloca.
Make sure the queue passed to OBS_QueuePresentKHR supports transfer to
safely call vkCmdCopyImage.
Update GetInstanceProcAddr/GetDeviceProcAddr to conform closer to spec.
Remove extra macros because I dislike indirection more than verbosity.
jpark37