Cropping was not handled properly when frames have to be copied to
xvimage's buffer pool, first because the crop meta were dropped, and
second because the allocated frame size in xvimage's buffer pool were
smaller than the incoming frame.
This patch updates xvimagesink's video info when propose_allocation()
is called, and copies the GstVideoCropMeta from source frame to
destination one.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/1088>
Buffer pool is created every time setcaps() is called, but it's
required only when upstream doesn't use it, so it's only needed to
copy frames onto XV buffers.
This patch delay the creation of the buffer pool until it's frame copy
is required.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-base/-/merge_requests/1088>
The problem is for uploading YUV frames using derived images, is that
derived images imply tiling, so frames are wrongly uploaded.
Though derived for reading might work we cannot know the Intel graphics
generation to validate the caching. Overall, it's safer to disable derived
images for i965.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/2127>
iHD driver sets a tiled DRM modifier if surface's usage hint is set to
VPP_WRITE. This result in a garbled rendering when using glimagesink.
This patch changes the usage hint to generic if the caps feature is
DMABuf. Either way only iHD driver, so far, uses the usage hint flag.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/2127>
skip_try_fmt_probes is set to TRUE for v4l2src to skip interlace-mode and
colorimetry when probe caps. gst_v4l2_object_set_format_full() will add
colorimetry back to caps when iterating over the negotiated caps. There is
one case that v4l2src is first in preview state then starts recording.
v4l2src caps will change with an additional interlace-mode structure after
renegotiation, then v4l2src needs to reset. But this camera driver can't
orphan buffer pool, it causes require buffer failed as streaming is still
in active state.
To fix this, also need to add interlace-mode back to caps for camera to
avoid reset.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/-/merge_requests/961>
This MR provides a transform element that leverage ONNX runtime
to run AI inference on a broad range of neural network toolkits, running
on either CPU or GPU. ONNX supports 16 different providers at the
moment, so with ONNX we immediately get support for Nvidia, AMD, Xilinx
and many others.
For the first release, this plugin adds a gstonnxobjectdetector element to
detect objects in video frames. Meta data generated by the model is
attached to the video buffer as a custom GstObjectDetectorMeta meta.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-bad/-/merge_requests/1997>
* Try to make variable and function names more clear.
* Add plenty of comments describing the logic step-by-step.
* Improve the logging around this, making the logs easier to read and
understand when debugging these issues.
* Revise the logic of packets that are actually beyond saving in doing
the following:
1. Do an optimistic estimation of which packets can still arrive.
2. Based on this, find which packets (and duration) are now hopelessly
lost.
3. Issue an immediate lost-event for the hopelessly lost and then add
lost/rtx timers for the ones we still hope to save, meaning that if
they are to arrive, they will not be discarded.
* Revise the use of rtx-delay:
Earlier the rtx-delay would vary, depending on the pts of the latest
packet and the estimated pts of the packet it being issued a RTX for,
but now that we aim to estimate the PTS of the missing packet accurately,
the RTX delay should remain the same for all packets.
Meaning: If the packet have a PTS of X, the delay in asked for a RTX
for this packet is always a constant X + delay, not a variable one.
* Finally ensure that the chaotic "check-for-stall" tests uses timestamps
that starts from 0 to make them easier to debug.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gst-plugins-good/-/merge_requests/952>
The VP9 spec defines the MAX_TILE_WIDTH_B64(64), which is the maximum
width of a tile in units of superblocks. So the max width of one tile
should not be larger than 64x64=4096. When the width exceeds 4k, we
need to split it into multiple tiles in columns. The current vp9 encoder
does not handle this correctly.
The command such as:
gst-launch-1.0 videotestsrc ! video/x-raw,width=7680,height=4320 ! \
vaapivp9enc ! fakesink
will crash.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer-vaapi/-/merge_requests/424>