As of 3.17 using find_package_handle_standard_args checks that the name
of the FindXXX file and the first argument are the same case.
Some modules used non-standard variables or the old singular variables
instead of plurals. This normalizes variable usage to the new-style.
Some CMakeLists.txt did custom error checking instead of propagating
find_package errors. These were changes to call find_package with
REQUIRED or without QUIET where needed and shortens the custom status
messages. This helps users who want to enable that functionality see
what precisely wasnt found.
This replaces the ffmpeg-encoded-output which had serious issues
(missing headers, muxing issues with non compliant mpegts streams) with
an output grafted from obs-ffmpeg-mux.
Simplify UI options by combining LookAhead Depth and Async Depth into
latency mode option. Ultra-low, low, and normal will set these two
encode parameters accordingly.
Don't allow unsupported Vulkan formats to fall back to B8G8R8A8.
Probably better to fail completely than do an illegal copy.
Also remove bad conversion for VK_FORMAT_A2R10G10B10_UNORM_PACK32.
Red and blue channels were reversed, and there's no DXGI equivalent.
Addresses #2796. We can do more later if justified.
Fix an issue in the way AVFrames were handled in the FFmpeg
encoder plugin, which could lead to tearing in the encoded
video due to data races in the AVFrame and AVBuffer.
This is fixed by calling av_frame_make_writable which ensures
the frame and its associated buffer are writable. If its not,
it will copy the AVFrame, create a new AVBuffer for it and
decrease the refcount of the old AVFrame and AVBuffer.
This way OBS always ends up with a usable buffer to write into
which is not still used by the encoder while avoiding a copy
when unnecessary.
This implements OSS audio input capturing support for OSS-capable OSes.
FreeBSD and DragonFly (not yet tested on) supports are added as a
starting point.
The error message in the process_packet function was prefixed with
"receive_audio" because it was previously from code in the encode_audio
function which was called from the receive_audio function. This is just
a string change to avoid being misled to believe that the error is
always audio related.
If we fallback to ffmpeg NVENC, the error from new NVENC might still be
present in the encoder structure. Given that this provides a lot more
actionable information to the user, let's use it if possible.
While discussing the Flatpak RFC [1], it was spotted that the
LUT filter couldn't open the file selection dialog. It was
explained, then, that the proper formats were either composed
of "User Label (file extensions)", or "file extensions", and
the LUT filter was setting "(file extensions)" without the
actual user label.
While this works on a standard Qt file selection dialog, it
cannot be properly formatted as a set of D-Bus filters, thus
breaking the sandbox integration.
Add a simple user label to the LUT file filter.
[1] https://github.com/obsproject/rfcs/pull/21#issuecomment-619106757
Enables a pipeline for texture-based encoding with QSV. Utilizes OBS
NV12 output for encode to avoid offloading them from GPU, which will
increase performance. The option to select old QSV pipeline still
remains and will fallback if new pipeline fails.
add_control_property() was previously static inline bool, but did not
return a value and failed to build on FreeBSD-CURRENT with Clang 10.0,
with:
error: non-void function 'add_control_property' should return a value
As os_gettime_ns() gets large the current scaling methods, mostly by casting
to uint64_t, may lead to numerical overflows. Sweep the code and use
util_mul_div64() where applicable.
Signed-off-by: Hans Petter Selasky <hps@selasky.org>
As of 3.17 using find_package_handle_standard_args checks that the name
of the FindXXX file and the first argument are the same case.
Some modules used non-standard variables or the old singular variables
instead of plurals. This normalizes variable usage to the new-style.
Some CMakeLists.txt did custom error checking instead of propagating
find_package errors. These were changes to call find_package with
REQUIRED or without QUIET where needed and shortens the custom status
messages. This helps users who want to enable that functionality see
what precisely wasnt found.
The Auto level lets the encoder decide which level to use.
Furthermore, some devices allow levels beyond 4.2. Therefore,
we add all levels allowed by H.264. This is of importance,
since using a resolution not supported by the specified level
can trigger undefined behavior on the hardware/kernel.
Previous fix in #2547 altered id before they were handed off to
callbacks and v4l2 wont strip its own flags from ids resulting in
invalid ids in the ioctl.
Instead add cleanup section and jump all branches there before looping.
(Jim edit: I'm changing this entire function so that it isn't horrible.)
v4l2_ioctl will clear the V4L2_CTRL_FLAG_NEXT_CTRL flag resulting in
infinite loops for any driver that reported readonly, disabled, or
volitile controls.
Not all devices return PKEY_AudioEngine_DeviceFormat properties.
Per MSDN, if the PROPERTYKEY referenced in key is not present in
the property store, this method returns S_OK and the vt member of
the structure pointed to by pv is set to VT_EMPTY.
PR #1937 (commit b9ad1ce) added QSV target usage options, but there was
a comma missing between two of the array entries. This resulted in
"faster" and "veryfast" becoming "fasterveryfast", which is not valid.
Code style changes were required by .clang-format.
This fixes a bug where games like "Don't Starve Together" wouldn't
capture because their actual render window is a completely different
window than their actual window on the screen.
So, because we already have the hook info by this point with the last
known window handle available, instead of using 0 here, we can just use
the window handle provided by the shared memory. And we didn't even
have to change the hook! That's nice.
RandR has two sets of screen geometry information:
1. CRTC. These are the physical scanout engines in the hardware
2. Monitors. These are the logical partitions of the screen.
By default, each CRTC gets mapped to a Monitor. However, some monitors
actually require two CRTCs to drive them due to limitations in the
scanout hardware. Users can also create 'virtual' monitors to support
VNC or other systems.
This patch makes the RandR code prefer the Monitor mechanism to the
older CRTC mechanism. If the server doesn't support a new enough RandR
version, the existing CRTC code is used instead.
The name of the monitor is also provided in place of the arbitrary
number to help users select the desired source.
Signed-off-by: Keith Packard <keithp@keithp.com>
Instead of erroring out completely when it can't determine if the
certificate is valid, proceed anyway. This matches how web browsers
treat failed cert revocation checks. schannel just has somewhat
paranoid defaults.
Make sure HWND tracking is cleaned up when Vulkan surfaces are
destroyed. Also use unbounded linked list to fix games that leak
surfaces on Alt+Tab like Doom.
Also replace CRITICAL_SECTION with SRWLOCK, both for claimed speed
benefit, and to remove initialization code.
Certain UWP programs can't obtain a normal window handle from their API
for whatever reason (this was observed with minecraft win10 edition), so
if the normal window handle on the map fails, try window handle 0
instead.
(This also modifies image-source, obs-text, text-freetype2, and UI)
This improves source definition versioning. To do this, it now stores
two identifier names. One "unversioned" which is the original name, and
one "versioned" with the version number appended.
This fixes both backward compatibility with older OBS versions, and
fixes the inability to use "add existing" in OBS itself on sources
created from older version definitions.
The problem with using obs_get_video_info for a default is that the user
can change that value. A default should not be a value that is subject
to change based upon the user's settings.
Instead, choose a sane "normal" value, like 1920x1080. Later on, this
should be replaced with a system that allows automatic resizing via the
user.
Sometimes the width/height would initially be at 0x0 when the swap chain
was alt-tabbing, causing the capture to fail full-stop when trying to
reacquire.