We don't know if there's a way to get window color space, so we've been
using the monitor color space of the window. This toggle forces OBS to
ignore the monitor color space, and assume the window is SDR.
This reverts commit a36b5bee99.
After this commit was merged, apparently devices were not functioning
correctly for some people. Especially regarding their internal settings.
Just revert it for now.
Apparently this annoying, stupid variable was leaking before the ComPtr
was added, which ironically was necessary to ensure devices worked.
After that ComPtr commit some devices stopped capturing properly.
Basically, it implies that this pointer needs to stick around while the
device is in use.
(Jim note: This was one of the most painful things I've ever had to
debug)
It's about time to get rid of this being labeled as "(new)".
Also rename the FFmpeg variant. And make it more explicit when the
FFmpeg encoder is being used in the log file.
With HEVC and H264 settings being near-identical, it was impossible to
figure out which codec was being used by context alone. This applies to
both ffmpeg output and jim-nvenc.
Fixes#6976.
This was necessary since the default background color was meant to be
clear, but wasn't set properly. The bug is fixed in macOS 13 Beta 6, so
we can remove this statement.
Move multiplication to when its passed to the encoder, so that bitrate
is kept in kbps. Changed for both for H264 and HEVC. Other encoders
(x264 and NVENC) already display bitrate in kbps in the log,
so it makes sense to mimic this with AMF. It's difficult to tell the
exact bitrate with bps.
This goes back to a slightly older variant of the
SubmitInput/QueryOutput handling that doesn't use AMF's timeout
property. Older devices do not like it when you change the query timeout
on the fly and will lock up. So instead, wait one millisecond when the
AMF input is full, which appears to fix the issue according to testers.
Also adds a loop timeout in case it goes in an infinite loop (which it
shouldn't anymore, but still)
Big thanks to Flaeri for testing the old code, and Yukari for patiently
testing a whole bunch of builds.
All this does is it uses the same exact code AMD uses with their own
example FFmpeg muxer code. Although instead of adding to the PTS, it
subtracts from the DTS.
The memory leak was introduced by a commit ba68eda59 to use
av_packet_alloc because av_init_packet got deprecated.
Also removes a boolean flag `new_packet` and use the pointer `packet`,
which is introduced by ba68eda59, to indicate there is a new packet.
Co-authored-by: Norihiro Kamae <norihiro@nagater.net>
Parentheses are needed due to operator precedence (although the previous
code happened to work because POLLIN has the value 1).
Fixes: 91f986ec99 ("linux-v4l2: Check udev fd events")
Users with AMD CPUs and Intel dGPUs (Arc) would find that QSV is
restricted to 1200p and missing features due to to this check failing.
Once the encoder gets rewritten for AV1 support we will fix this
properly.
udev_event_thread calls poll with two fds: the udev fd, and an eventfd
used for shutdown. On FreeBSD we were hanging on shutdown in
udev_monitor_receive_device after receiving the eventfd event.
Now, if the udev fd reports no events skip the
udev_monitor_receive_device call.
At shutdown the loop will exit via os_event_try().
Currently we use signals based on os_event's but the udev code blocks in
a 1s select (now poll) that cannot be signalled without an FD. This adds
an extra eventfd which will be signalled along with the os_event to
trigger the poll.
Also switches from select to poll to avoid issues with large fd values.
Converting the color space was likely accidental as it isn't explicitly
set. Unfortunately, not setting it means that it gets set to the
default, which is Rec. 709 and thus a conversion takes place when having
any other space. This conversion leads to a massive performance penalty
that isn't necessary.
Adding all windows manually means that any windows opened after the
session was started won't appear. This instead excludes the Control
Center from Display Capture. While unfortunate, it seems to be the only
thing on the display that can't be disabled.
Since this was included in things like the inject helper and game
capture DLLs, it added an additional export to those programs. Instead,
simply share the source files directly in the projects that need it.
Co-authored-by: jpark37 <jpark37@users.noreply.github.com>
This resets the RTX greenscreen filter if the FX returns a CUDA error.
In case of a CUDA error 1099, the error can keep repeating locking up
the filter which is inoperant. So if such an error is detected we reset
the FX.
Signed-off-by: pkv <pkv@obsproject.com>
Replace TRANSCONDING with TRANSCODING, because it is likely that the
typo compatibility will be removed in a future AMF header update. This
should also minimize confusion and improve search.
This reverts commit 0532a5c1f0.
Ubuntu 21.10 is from OBS 28 onwards not longer supported and as such we
can raise the requirement of the PipeWire dependency to 0.3.33.
If the test process freezes for whatever reason, OBS would block
indefinitely on reading the pipe. This commit adds a 2.5 second timeout
on the AMF test process after which it will self-terminate.
In the event that simply loading the AMF DLL causes initialization code
to run that might crash, don't load the DLL before the AMF test process
has succeeded. Instead, we load the DLL as data to see if it exists,
then run the AMF test, then load the DLL for real.
Regular SDR/HDR stingers, and SDR track matte should work. HDR track
matte might work, but would take a carefully crafted video that takes
the SDR white level into account, and this hasn't been tested.
When compiling with the new flag these are required, if we dont fail
here we instead get a more cryptic failure that the targets are
unassigned later in the file. This should make it easier to find the
flag to turn off or packages being searched for.
This fixes a bug where an update of the parent source breaks the filter.
A signal handler is added to listen to updates of the parent.
This triggers a reset of the RTX FX.
Signed-off-by: pkv <pkv@obsproject.com>
By default, new mpegts output is used; but to allow CI on linux not
to be broken, we allow use of old mpegts output.
Up to ubuntu 22.04 there is no librist package available.
A manual compile is then required but the CI scripts would need to be
updated.
This also allows easy fallback in case of fatal bugs in new output.
Signed-off-by: pkv <pkv@obsproject.com>
Currently the ffmpeg_mpegts_muxer output is integrated with ffmpeg-mux.
Both use obs native encoders in contrast with obs-ffmpeg-output which
relies on avcodec library.
This allowed easy implementation of SRT, RIST & HLS protocols through
avformat library.
The main drawback is that obs-ffmpeg-mux exe doesn't allow for easy
debugging nor logging of the protocols.
It was written initially as a separate binary designed for recording so
that if obs fails for some reason, the recording can still terminate
gracefully.
In this commit the ffmpeg_mpegts_muxer is rewritten so that a pipe to
the ffmpeg-mux binary is not used any more.
The muxing to mpegts is still delegated to avformat.
But it can be traced more easily in all its steps.
Also the protocol part for SRT & RIST is implemented natively.
Custom avio_contexts for SRT & RIST are used to that end.
This allows to pass our own implementation of librist and libsrt
libraries instead of relying on avformat. This is very advantageous :
- this allows better logging.
- this allows better bug fixing and maintainance without having to rely
on hypothetical upstream fixes.
One immediate bonus of native implementation is that fixes bugs which
were not previously fixable.
Fixes: SRT & RIST auto-reconnect partly broken #6749
Fixes: SRT: OBS unusable and uncloseable after starting stream to
invalid srt server #5791
Signed-off-by: pkv <pkv@obsproject.com>
After changing the media source from video file to audio file,
the last video's frame (preloaded) maybe still rendered.
This commit ensures that when media source changed to
another one without video, clean the source's video output,
by simply add judgement to check whether the media has video.
Switch to Visual Studio 2022 and clang-format 13.
The Ubuntu 22.04 image currently has clang-format 12.0.1, 13.0.1, and
14.0.0. VS2022 17.2 ships clang-format 13.0.1, so let's use that for now
until VS2022 17.3 is available with clang-format 14.
This commit also makes the necessary changes to allow the clang-format
check to pass.
This commit also updates the obs-browser submodules with the necessary
changes.
Please visit the submodule repo for blame history:
https://github.com/obsproject/obs-vst
This also replaces the obs-vst .clang-format with obs-studio's.
This commit depends on the previous commit, which removes the submodule
separately as Git complains otherwise.
Co-authored-by: Alex Anderson <anderson.john.alexander@gmail.com>
Co-authored-by: Anton <camotank12345@gmail.com>
Co-authored-by: Blue Cat Audio <support@bluecataudio.com>
Co-authored-by: Cephas Reis <c3r1c3@nevermindonline.com>
Co-authored-by: Colin Edwards <colin@recursivepenguin.com>
Co-authored-by: Florian Zwoch <fzwoch@gmail.com>
Co-authored-by: Fogmoon <i@fogmoon.com>
Co-authored-by: Gol-D-Ace <Gol-D-Ace@users.noreply.github.com>
Co-authored-by: Igor Bochkariov <ujifgc@gmail.com>
Co-authored-by: Jesse Chappell <jesse@sonosaurus.com>
Co-authored-by: Keen <523059842@qq.com>
Co-authored-by: Kurt Kartaltepe <kkartaltepe@gmail.com>
Co-authored-by: Matt Gajownik <matt@obsproject.com>
Co-authored-by: Matt Gajownik <matt@wizardcm.com>
Co-authored-by: Richard Stanway <notr1ch@users.noreply.github.com>
Co-authored-by: Ryan Foster <RytoEX@gmail.com>
Co-authored-by: follower <github@rancidbacon.com>
Co-authored-by: gxalpha <beckmann.sebastian@outlook.de>
Co-authored-by: jp9000 <obs.jim@gmail.com>
Co-authored-by: jpark37 <jpark37@users.noreply.github.com>
Co-authored-by: mntone <sd8@live.jp>
Co-authored-by: tytan652 <tytan652@tytanium.xyz>
Co-authored-by: wangshaohui <97082645@qq.com>
Co-authored-by: wangshaohui <wang.shaohui@navercorp.com>
Please visit the submodule repo for blame history:
https://github.com/obsproject/obs-vst
NOTE: You may need to manually delete plugins/obs-vst to pull this diff.
When the ScreenCaptureKit-based source is generated, the window ID
read from the settings will be empty and interpreted as window ID "0".
This window ID doesn't represent a valid window, but the capture stream
is nevertheless successfully initialized with that value, but will not
generate any output.
Worse, the activated stream should be destroyed when a new stream is
created, but creation will throw an error.
This commit ensures that a valid window ID is used if no good default
value is available, ensuring a working (and destroyable) stream.
Also removes any remaining usage of the old window-utils.
Initializing a display stream with a content filter which excludes an
empty list of windows leads to a broken stream. Initializing with a list
of all available windows as a workaround re-enables functionality.
Reduces the amount of calls to the update callback (also reduces amount
of calls to check for available capture content).
Also moves some code to only update data for currently selected capture
type.
Whereas the `availability` checks will correctly detect macOS 12.5,
the `__MAC_OS_X_VERSION_MAX_ALLOWED` macro is dependent on the platform
SDK. The most current platform SDK is 12.3, hence why this version
needs to be checked for.
If the API to query the maximum throughput isn't available, it was
intended to fall back to balanced. This code caused it to vall back to
speed instead because if the API isn't available, max_throughput will be
0. This fixes it to make it only fall back to balanced instead.
This adds support for RTX video Effects SDK.
The Background Removal filter relies on NVIDIA RTX AI Greenscreen FX.
A front speaker is separated from a background by applying
an alpha mask computed by AI.
Signed-off-by: pkv <pkv@obsproject.com>
The NVIDIA Audio Effects SDK currently interferes with the Video
Effects SDK by not releasing a CUDA context.
Till this is fixed upstream, we do it manually. This also requires
an SDK update to 1.1.0.5 for NVIDIA Audio Effects SDK.
Signed-off-by: pkv <pkv@obsproject.com>
This updates the support of NVIDIA Audio Effects SDK to 1.1.0.5
Additionally, we add support for some CUDA functions required for
fixing a bug.
Signed-off-by: pkv <pkv@obsproject.com>
SCK's application list has one entry with empty name and bundle ID which
belongs to WindowServer. This is not worth exposing since an empty entry
would confuse users and nothing would get captured if selected.
Marking a source as deprecated will make it harder for users to find
when creating a new source. We don't currently plan on removing
traditional display and window capture, however ScreenCaptureKit will be
the preferred capture method going forward.
on the capabilities of audio capture refer to the documentation https://developer.apple.com/documentation/screencapturekit/ or watch the session "Meet ScreenCaptureKit".
Currently audio capture typically happens via virtual devices set up with 3rd party products. This requires additional installation and configuration by users.
Including the audio with the existing CMBuffers will make it easier to manage audio sync with video frames. Additionally, you can capture distinct audio tracks from multiple apps at once,
like a call with friends + screen content.
macOS 13 on a 2019 16" MacBook Pro
macOS 13 on a 2022 M1 Pro Macbook Pro
Tested with multiple games + Discord, FaceTime, Apple Music and Safari tabs playing audio content.
* Bug fix (non-breaking change which fixes an issue)
* Performance enhancement (non-breaking change which improves efficiency)
* Additional control over what audio content to capture
* Logic to manage audio & frame sync
Replaces the usage of SO_SNDTIMEO with TCP_USER_TIMEOUT on Linux.
- Noted as more effective than SO_SNDTIMEO by multiple sources.
- Drops TCP connection on timeout instead of returning send() errors,
meaning more predictable reconnect timing and reconnect.
- Timeout period actually reflected in practice. SO_SNDTIMEO takes
double the timeout time than actually requested to time out on
unix, whereas WinSock actually waits the specified time.
When trying to update to mbedTLS 3.2.1, I ran into failures with RTMPS
output. After consulting an mbedTLS implementation example, I determined
that we were not setting up the SSL/TLS context config in the correct
order, causing the connection to fail. Performing the setup in the
recommended order fixes that.
Adds support for texture-based AMD encoding, with both H264, HEVC, and
HDR support. Falls back to FFmpeg when texture-based encoding cannot be
used for whatever reason.
(Jim note: This is based upon obsproject/obs-studio#4538 by AMD/Luxoft
with fewer files, FFmpeg fallback for software encoding, and HDR
support. I also went to lengths to ensure that FFmpeg command line
parameters also works with it)
Co-authored-by: Jim <obs.jim@gmail.com>
cfad34d - Use auto release refs
64c1796 - Handle SSL cert date invalid error in browser panels
394d6f7 - Back up then restore POSIX signal handlers on CefInitialize
161ffd2 - Replace type mismatch with range-based for loop
Add aja-source version 2 to differentiate buffering settings, which is
implemented in a255b0f74.
To register two versions, remove to return obs_source_info and register
the source inside aja-source.cpp.
Also revise aja-output.cpp to register aja-output to keep consistency of
function type.
This commit rounds the timestamp to the integer multiple of fps_den.
Prior to this change, since the timestamp in MFX is defined in units of
90 kHz, timestamps for 60000/1001 fps is not accurately expressed. Also
the encoder sometimes returns the decode time stamp slightly off.
Hence, the timestamp in OBS becomes not a multiple of fps_den after
converting back from the timestamp in MFX.
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.
`absolutePathForAppBundleWithIdentifier` and `openFile` have both been
deprecated since macOS 11.0. This pr uses the correct modern APIs when
OBS is compiled for macOS 11.0 or more recent and falls back to the
legacy APIs when either compiled against or run on older macOS versions.
Both rate control methods only work on hardware encoders and will error
out when selected for software encoders, so we shouldn't show them there
Co-Authored-By: Sebastian Beckmann <beckmann.sebastian@outlook.de>
The session_set_bitrate method already returns an OSStatus to confirm
that everything has been successful. As such, a second check to make
sure that the bitrate has indeed changed is redundant and only adds
unnecessary code.
It is expected that capture might fail, like when a window is minimized.
In these cases we dont want the log to be spammed by X11 errors so
install a noop error handler while we try to import.
Direct access to `[AVCaptureDevice devices]` is deprecated since macOS
10.15. An `AVCaptureDeviceDiscoverySession` is to be used instead, which
this PR implements.
When running on older macOS versions or compiling against older platform
SDKs, the deprecated method is used.
PS: The formatting is not proper ObjC style, but that's clang-format's
doing with our current settings.
This commit fixes what is arguably a long-winded series of previous
commits that have possibly caused just as many problems as they have
fixed. I'll spare the details, but basically, there's no reason that
any of the RTMP object should ever be used across socket sessions.
This provides a slight enhancement by removing the `RTMP_Init` call
in `rtmp_stream_create()`, since it effectively just initializes TLS
just for `try_connect` to deinitialize it before it is even used.
This also fixes the current `SO_RCVTIMEO` timeout functionality by
making sure that `RTMP_Reset` is called last.
To improve readability of the capture thread function
`AJASource::CaptureThread`, audio part is split into two functions.
Behavior is not changed.
Also fixed one of warnings.
When too many audio frames are captured and at the same time the ring
buffer on the card is wrapped, memory overrun was not captured before
DMA transfer, which resulted in corruption of memory contents. This
commit moves the overrun check before the DMA transfer.
This fixes a bug where the new circular buffer code is hit by protocols
such as rist or srt although the buffer was meant for files.
Signed-off-by: pkv <pkv@obsproject.com>
It is reported that the encoder does not follow the bitrate settings.
This commit tries to correct the timestamps so that the bitrate control
in the encoder works.