See #2611
When transitioning to PLAYING we call IDeckLinkInput::StartStreams() (via GstDecklinkInput::start_streams). This can
fail for various reasons, including if the hardware device is already in use.
Previously, we logged StartStreams() failure as an element error but otherwise continued to successfully transition to
PLAYING. Now, if StartStreams() returns an error we fail the state change.
If StartStreams() fails then a subsequent call to StopStreams(), when transitioning PAUSED to READY, will also fail.
Previously, if StopStreams() failed then the state change also failed. Unfortunately this prevented the element from
later being disposed, which in turn meant the associated hardware resources was never freed. Consequently, if a
decklinkvideosrc failed to transition PAUSED to READY, then the associated hardware device could not subsequently be
used by any other decklinkvideosrc.
Now, we log an element error if StopStreams() fails but otherwise consider the state change to have succeeded. This
means that the element can be disposed and the associated hardware resource released.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9578>
plugin_init() will not get called if element/feature registration
happens manually, such as when using linking only specific plugin
features with gstreamer-full. That is possible when plugins contain
static features.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9496>
Since the sequence of IMMDeviceEnumerator::EnumAudioEndpoints()
followed by IMMDevice::Activate() is not atomic, Activate() may fail
if the enumerated device becomes invalidated before probing.
In such cases, retry device probing
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9468>
VA encoders, at reconfiguration, have to check if the rate-control was changed
by the user, but since user parameters setting are in another thread, the
comparison was racy.
This patch locks the object to compare the current rate-control with the one set
by the user.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9480>
When the base videodecoder class re-attempts a negotiation after flush, the
vabasedec `need_negotiation` flag isn't necessarily set to TRUE, because in that
situation the input state hasn't changed.
By always chaining up we are sure that buffer pool negotiation will always be
attempted.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9457>
When the base videodecoder class re-attempts a negotiation after flush, the
vabasedec `need_negotiation` flag isn't necessarily set to TRUE, because in that
situation the input state hasn't changed.
By always chaining up we are sure that buffer pool negotiation will always be
attempted.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9457>
When the base videodecoder class re-attempts a negotiation after flush, the
vabasedec `need_negotiation` flag isn't necessarily set to TRUE, because in that
situation the input state hasn't changed.
By always chaining up we are sure that buffer pool negotiation will always be
attempted.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9457>
When the base videodecoder class re-attempts a negotiation after flush, the
vabasedec `need_negotiation` flag isn't necessarily set to TRUE, because in that
situation the input state hasn't changed.
By always chaining up we are sure that buffer pool negotiation will always be
attempted.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9457>
Add d3d12fisheyedewarp element that performs fisheye image dewarping
using D3D12. A UV remap LUT texture is generated via a compute shader,
and the actual remapping is performed in a pixel shader using this LUT
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9441>
If enabled, wasapi2src/sink will post a warning message instead of an error,
when device failures occur, such as open failure, I/O error,
or device removal.
The element will continue to produce/consume audio buffers and behave as if
a capture/render device were active, allowing pipeline to keep running even when
no audio endpoint is available
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9326>
... even if hardware does not support interlaced encoding at bitstream level.
Although interlacing information is not written in the bitstream,
that information can be signalled via container, thus allow interlaced
stream.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9328>
Add a new action signal to allow applications to capture
the most recently rendered frame directly from the swapchain
back buffer.
Unlike the existing "last-sample" property, which exposes
the raw input sample before any sink-side processing, this
signal captures the final displayed image after any internal
image processing (e.g., UV remap, color balance, overlay) has been
applied.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9423>
Previous version passed just enough data to the parser to detect
main/main-10, but main-422-10 is one of the range extensions profiles.
Those need a few more bits to be accurately detected, and since those
were just uinitialized memory previously, we'd incorrectly end up with
main or main-10 when the encoder was in fact giving us 4:2:2 10bit
output.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9348>
