Always add GstTensorMeta to buffer instead of re-using existing GstTensorMeta
when it's already present. The reason for the change is when we have cascaded
inference elements overwriting (tflite) or leaking (ONNX) GstTensorMeta when we
have cascaded inferences. We didn't create and API to append to GstTensorMeta
and I think it's more convenient to have a new GstTensorMeta for each
inferences. It's also more clear that a group of tensor was produced by one
inference but doesn't limit tensordecoders from using tensors produced by
multiple inference. I don't see much value in having all tensors data always
inside one GstTensorMeta since appending would mean re-allocation of the tensors
array anyway.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9564>
LCEVCdec supports different pixel aspect ratios other than 1/1. This change
forwards the pixel aspect ratio of the base picture to the LCEVC decoder,
and also updates the output pixel aspect ratio caps base on the one from the
enhanced frame.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9463>
The output resolution is not always twice as big as the input resultion divided
by the pixel aspect ratio. This is the case for LCEVC '0D' mode, where the
output resolution is the same as the input resolution, and the only enhancement
is the picture being clearer.
This patch uses LCEVC_PeekDecoder() after sending the LCEVC enhancement data to
know what the output resolution will be before allocating the output picture.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9463>
Even though the LCEVC decoder works fine without this, it is recommended to
set read access to base pictures that are sent to the decoder, and write access
to enhanced pictures that are received from the decoder.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9463>
.. in decoders.
Inline session parameters allows to not create session parameters handlers for
every new stream parameters (such as SPS and PPS for H.264, for example), but
instead to pass them as a chained structure in the decoding main structure. This
is completely align with GStreamer decoder base classes.
Even that the previous approach is kept, if the devices doesn't support video
maintenance2, it shows a lot of validation errors.
Also it was required to add another parameter when enabling extension to verify
if the extension is linked with a device feature and if it is enabled.
Bump Vulkan API (and driver version for both decoders and encoders) to 1.4.306
Also bumped the ABI_CHECK_TAG because the CI finally catches up with the vulkan
video symbols that are not exposed by a public header (tough they are binary
public).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9585>
Some extensions need to have enabled certain feature in the device. This patch
does that check by adding a new field in the extension list which is a function
that can be mapped to gst_vulkan_physical_device_has_feature_*() functions.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9585>
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>
The responsibility to change the framerate to the one
requested on the VideoTrack should always be the last time effect in it.
This introduces a new `capsfilter` at the end of video effect that
will be in passthrough most of the time.
t show
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9013>
There was a case where we could end up having more buffers on the sinkpads
than what was expected, depending on timings. This ensure that we have a fixed
number of buffers produced by the testsrc, and we play until EOS.
Replace theora encoder/decoder chain with fakevideodec for testing
QoS handling in videorate rate property tests. This simplifies the
test pipeline and removes dependency on theora codecs.
Also adjust test expectations to match RGBA format output from
fakevideodec instead of I420 from theoradec, and limit buffer
counts to make tests more deterministic.
This also fixes the rate we set for 2.0 test as we were actually testing
0.5 there.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9013>
When executing a stop action, ensure that any pending wait_message_action
is cleared and unreffed to prevent it from being executed after the
scenario has been stopped. This prevents potential crashes or unexpected
behavior when stopping a scenario that has a pending wait action.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9013>
Use the local context variable that was retrieved earlier instead of
action->priv->context which is always NULL at this point since we
reset it right before. This ensures we invoke the callback on the
correct context.
And make sure that the wait message still has a reference when setting done
otherwise it is unref before calling the context invoke function
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9013>
Add support for an 'overrides' field in validatetest meta structures that
allows changing issue severity levels during test execution. This enables
tests to pass when encountering known issues by downgrading their severity.
The overrides field accepts an array of change-severity structures with:
- issue-id: The issue ID to override
- new-severity: New severity level (critical, warning, issue, ignore)
Currently only change-severity overrides are supported. The feature follows
the same pattern as expected-issues and is only available in .validatetest
files, not .scenario files.
Includes comprehensive documentation and a test case demonstrating the
functionality.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9013>
Refactor the videorate element to properly separate upstream and downstream
time domains when the `rate` property is set.
The previous implementation had issues with time domain conversions when
changing rates, especially during seeks after which we were basically confusing
input and output timestamps.
This also fixes rate changes as we are now tracking the wanted input timestamps
and not confusing them with the output, so this way we do not drop buffers when
the rate is changed while playing, meaning that the related tests have been
fixed
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9013>
When the dots tracer is detected among active tracers, switch from
SHOW_ALL to SHOW_FULL_PARAMS to prevent parameter ellipsization.
The dots tracer is designed to work with gst-dots-viewer which can
handle long parameter values, so we preserve the full text in dot files.
This improves the viewing experience when using the dots tracer with
the gst-dots-viewer tool.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9547>
This checks if a tracer has registered to any hook after
initialization and automatically registers it to the "none" hook if not
so the lifetime of those tracer is the same as tracer who registers to hooks.
For example the `dots` tracer was being freed right after initialization
which was unexpected.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9547>
When GStreamer pipeline graphs contain references to other pipeline dot files,
users should be able to navigate between related pipeline views by clicking
on these references. This enables exploring complex pipeline hierarchies
where one pipeline references sub-pipelines.
The implementation detects text elements containing pipeline-dot references,
styles them as web links, and handles click events to navigate to the
referenced pipeline while preserving existing text selection and drag
functionality for other elements.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9547>
Enable text selection on SVG text elements by changing CSS user-select
properties from 'none' to 'text' and cursor from 'default' to 'text'.
Add JavaScript event handling to intercept mousedown events on text
elements, preventing dragscroll interference while allowing normal text
selection. This preserves the existing drag-to-pan functionality for
non-text areas while making text elements selectable and copyable.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/9547>