This a new VA-API implementation of a H264 encoder.
It can control the GOP and parameter settings, while the MV searching,
VCL and the rate control algorithm are implemented by VA drivers and HW.
It supports most of the common usage options in H264, but still lacks
of look ahead, field, B frame weighted prediction, etc.
Co-authored-by: Victor Jaquez <vjaquez@igalia.com>
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>
The g_queue_clear_full() and g_array_copy() functions in the glib
may not be available for the current glib version check, so we add
helper functions to wrap it.
This should be deleted after the glib version bumps.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1051>
LOAD macro relies in m7 being zero for interleaving purposes. Using LOAD
on the m7 register makes it interleave with its new content instead of
with 0.
The effect of this bug was bobbing on some static lines that appeared
over fast-moving content.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1816>
The fd was in different meanings on windows:
POSIX read and write use the fd as a file descriptor.
The gst_poll use the fd as a WSASocket.
This patch use WSASocket as default on windows. This is a temporary measure, because IPC has many different implement. There may be a better way in the future.
See #1044
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1791>
The dynamic resolution changes when
the sequence starts when the decoder detects a coded frame with one or
more of the following parameters different from those previously
established (and reflected by corresponding queries):
1.coded resolution (OUTPUT width and height),
2.visible resolution (selection rectangles),
3.the minimum number of buffers needed for decoding,
4.bit-depth of the bitstream has been changed.
Although gstreamer parser has parsed the stream resolution.
but there are some case that we need to handle resolution change event.
1. bit-depth is different from the negotiated format.
2. the capture buffer count can meet the demand
3. there are some hardware limitations that the decoded resolution may
be larger than the display size. For example, the stream size is
1920x1080, but some vpu may decode it to 1920x1088.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1381>
v4l2videodec do some refactoring so that it can support
dynamic resolution change event.
1.wrap the setup process of capture as a function,
as decoder need setup the capture again when
dynamic resolution change event is received.
2.move the function "remove_padding"
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1381>
This allows downstream of a payloader to know the RTP header's marker
flag without first having to map the buffer and parse the RTP header.
Especially inside RTP header extension implementations this can be
useful to decide which packet corresponds to e.g. the last packet of a
video frame.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1776>
osxaudiodeviceprovider now probes devices more than once to determine
if the device can function as both an input AND and output device.
Previously, if the device provider detected that a device had any output
capabilities, it was treated solely as an Audio/Sink. This causes issues
that have both input and output capabilities (for example, USB interfaces
for professional audio have both input and output channels). Such devices
are now listed as both an Audio/Sink as well as an Audio/Source.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1385>
The RTP payload seems to be required as it carries the frame count
information. Also, gst_rtp_base_payload_allocate_output_buffer had
the second argument incorrect.
Strangely some devices like Shanling MP4 and Sony XM3 would still
work without this while some like the Sony XM4 do not.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1797>
While this is slightly more expensive (~48% slower per random number) it
does not cause any measurable difference when running through a complete
audio conversion pipeline.
On the other hand its random numbers are of much higher quality and on
spectrograms for 32 bit to 24 bit conversion the difference is clearly
visible.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1729>
The instant-rate value in the TrickMode enum is a
flag, but the other values are not. Move instant-rate
to the end of the enum and give it a value large enough
for it to be used without modifying the trick-mode
setting.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1788>
Update x264enc long-name to be more than just the name. Then the
description also was updated to be longer than the long-name, and
similar to the plugin description.
Finally, as I am here, H264 was replaced by H.264 and x264 is only a
plugin (not plugins).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1771>
They can't be used in any useful way. The type of every GstMemory is
always GST_TYPE_MEMORY and the subtyping relationship has to be
implemented on top of that via the associated allocator and mem_type
string.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1764>
Scenario:
- Source 1 requesting and waiting a clock id
- Source 2 requesting and waiting on a clock id
- Test attempting to crank both sources in the same GstHarness
gst_test_clock_crank() originally dropped locks between the retrieving
of the next clock id and advancing to the next clock id. This would
mean that both sources would race each other attempting to complete
their clock waits. Sometimes the operations would be performed in the
correct order, other times they would not and a FALSE return value would
be produced.
This would lead to an assertion in gst_harness_push_from_src() expecting
that all clock cranks to succeed.
Fix by ensuring that the clock wait produced is dealt with before
processing the next by not dropping the relevant locks after retrieving
the next clock id.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1299>
This is a minimal unit test the show that the stride extrapolation can work
with all pixel format we support. This minimal verify that the extrapolation
match the stride we set into GstVideoInfo with 320x240 for all the pixel
format we support. The tiles formats are skipped, since their stride is
set as two 16bit integers, and we also skip over palette planes.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1567>
As this element is single threaded, we only need to stop the objects to
allow changing the format again. Fixes assertion notably on shutdown and
on some other situation where the format may be set twice without
actually activating the element.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1567>
Many of the legacy APIs, specifically in the Linux Kernel, have a
single stride for the pictures. In this context, it is common
to extrapolate the other strides based on the selected pixel
format. Such function have been copy pasted from video4linux2
plugin into wayland, kms and v4l2codecs plugins.
This patch implements a generalized from of that function and
make it available to everyone through the video library.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1567>
Unlike other simple tiled formats, the Mediatek HW use different tile size
per-plane. The tile size is scaled according to the subsampling. Effectively,
using the name 16L32S to represent linearly layout tiles of size 16x32 bytes
in the Y plane, and 16x16 in the UV plane. In order to make this specificity
discoverable, a new SUBTILES flags have been added.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1567>
If the video4linux device supports norms but has no norm set, norm is
returned as an uninitialized variable after the ioctl call, leading to
gst_v4l2_tuner_get_norm_by_std_id() returning a random norm from the
supported norms. Catch this case and instead return NULL to indicate
that no norm is setup.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1625>
We have the d3d11screencapturesrc element in d3d11 plugin
which is obviously better than this element in terms of performance
and design, so we don't need to make people be confused by two separate elements.
Let's pick the better implementation and remove unnecessary one.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1750>
... instead of round(). Depending on framerate, calculated position
may not be clearly represented by using uint64, 30000/1001 for example.
Then the result of round() can be sliglhtly larger (1ns) than
buffer timestamp. And that will cause unnecessary frame delay.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1747>
It was assumed that the kernel parameters would match with the bitstream value
but instead the author when with another set of value. Surprisingly, this
makes no difference with the resulting fluster score.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1748>
If a serialized event arrives behind a buffer, it should not be send before
it. This fixes the pending event handling so that only early pending events,
the one that arrrived or was generated while the adapter was empty get send
before pushing buffer. All other events are not pushed after.
This issue lead the latency tracer to think our audio encoder did not have any
latency. This was testing with opusenc in a live pipeline.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1266>
Each stream may have its own segment timeline
(i.g., different segment.start or segment.base)
depending on edit-list and composition-to-decode atom.
Make sure whether time position of a stream has been actually
far behind than that of current target stream.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1352>
If present, add '-lsocket' and '-lnsl' to network_deps.
ext/curl/meson.build: add network_deps to dependencies
gst/festival/meson.build: same
sys/shm/meson.build: same
Fixes linking issues on Illumos distros.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1525>
Often, users will need to scale inputs (e.g.
with vaapipostproc) before they are submitted
to the vaapioverlay. However, this results in
multiple VPP passes/operations in the pipeline
which creates unnecessary process overhead.
This change allows for inputs to be submitted
at original scale to vaapioverlay with per-sinkpad
scale dimensions specified so they can be scaled
and blended/composited in a single VPP pass/operation
to avoid the unnecessary process overhead.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1380>
Don't set VAAPI vpp blend flags if alpha == 1.0,
i.e. fully opaque. This can avoid extra processing
overhead on some drivers that apply blending
unconditionally when flags are present, even if the
end result is the same without blend flags (i.e. all
opaque alpha channels).
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1380>
Note that AYUV and AYUV64 formats will be used to expand format
support, especially some packed YUV formats (e.g., Y410, YUY2)
are common DXGI formats used for hardware decoder/encoder on Windows
but those formats cannot be used as a render target. We need to handle
them differently without pixel shader help, using compute shader
for example.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1699>
The libjpeg-turbo internal state might not be correctly initialized for
the first frame in a stream, pull the frame stride from gstreamer frame
metadata instead, which is correct even for the first frame, and which
makes this code consistent with the surrounding lines.
Fixes: e6d83d8f96 ("jpegdec: Support libjpeg-turbo colorspace conversion")
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1687>
It is imperative that the libjpeg-turbo state is properly initialized
before jpeg_start_decompress() is called. Make sure cinfo.out_color_space
and cinfo.raw_data_out are set to their final values matching their peer
caps before calling jpeg_start_decompress().
Fixes: e6d83d8f96 ("jpegdec: Support libjpeg-turbo colorspace conversion")
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1687>
Pull out peer caps checking code into gst_jpeg_turbo_parse_ext_fmt_convert().
This code is used by libjpeg-turbo extras to determine whether peer is capable
of handling buffers into which libjpeg-turbo can directly decode data. This
kind of check must be performed before jpeg_start_decompress() is called in
gst_jpeg_dec_prepare_decode() as well as in gst_jpeg_dec_negotiate(), hence
the common code.
This commit does modify the code a little to make it easier to call from both
call sites without much duplication, hence the extra `if (*clrspc)` test.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1687>
This reverts commit 2aa2477208.
The commit is completely wrong, libjpeg-turbo is perfectly capable
of decoding I420 (YUV) to RGB. The test case provided alongside the
aforementioned commit passes without this revert because it decodes
image of JCS_YCrCb color space, so the new `if (clrspc == JCS_RGB)`
condition is false on that image, and the libjpeg-turbo decoding
does not get used. The real bug is hidden by that commit.
The real problem is in the call order of gst_jpeg_dec_prepare_decode()
and gst_jpeg_dec_negotiate(). The gst_jpeg_dec_prepare_decode() calls
jpeg_start_decompress() which sets up internal state of the libjpeg,
however, neither cinfo.out_color_space nor cinfo.raw_data_out are
set correctly yet. Those two are set up in gst_jpeg_dec_negotiate()
which is called a bit later. Therefore, the real fix is the set up
cinfo.out_color_space and cinfo.raw_data_out before calling
jpeg_start_decompress(). This is however a separate patch.
Fixes: 2aa2477208 ("jpegdec: only allow conversions from RGB")
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1687>
Remove all the d3d11 and dxgi header version dependent ifdef
and bump the minimum requirement to d3d11_4.h and dxgi1_6.h.
We are already failing support old Visual Studio (Windows SDK actually)
such as Visual Studio 2015. Note that our MinGW toolchain satisfies
the requirement.
From runtime point of view, this change should be fine since
we are checking OS version with IUnknown::QueryInterface()
everywhere in order to check API availability
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1684>
We make all MSDK encoders declare "memory:VAMemory" feature. Then
the pipeline such as:
gst-launch-1.0 -vf filesrc location=xxx.h264 ! h264parse ! \
vah264dec ! msdkh265enc ! fakesink
will choose VA memory caps between the VA decoder and MSDK encoder.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1087>
The MSDK encoder's query function is not set and it just forwards
all query to its base class. We now need to answer the context
query correctly. Other VA plugins need to query the VA display.
By the way, the current query of "gst.msdk.Context" is also missing.
The other MSDK elements must depend on the bin's context message(
sent in context_propagate()) to set their MsdkContext correctly.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1087>
The MSDK VPP's query function is not set and it just forwards
all query to its base class. We now need to answer the context
query correctly. Other VA plugins need to query the VA display.
By the way, the current query of "gst.msdk.Context" is also missing.
The other MSDK elements must depend on the bin's context message(
sent in context_propagate()) to set their MsdkContext correctly.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1087>
The MSDK decoder's query function is not set and it just forwards
all query to its base class. We now need to answer the context
query correctly. Other VA plugins need to query the VA display.
By the way, the current query of "gst.msdk.Context" is also missing.
The other MSDK elements must depend on the bin's context message(
sent in context_propagate()) to set their MsdkContext correctly.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1087>
We now can use the gst va lib's display to create our MSDK context,
and use its helper functions to simplify our code. The improved logic
is like this:
1. Every MSDK element should use gst_msdk_context_find() to find a MSDK
context from neighbour. If valid, reuse it.
2. Use gst_msdk_ensure_new_context(). It will first query neighbours
about the GstVaDisplay, if found(e.g. some VA element is connected),
use gst_msdk_context_from_external_display() to create a MSDK context.
3. Then, creating the MSDK context from scratch. It creates both the
display and MSDK context.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1087>
The VA display object from VA lib is a common defined object. which
contain the whole display things. It is easier to use, and more important,
we can share it with the other VA plugins and keep all the VA related
plugins working on the same GPU device.
We also delete the useless gst_msdk_context_get_fd() API.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1087>
The current manner for deciding the new temporal unit is based on
temporal delimiter(TD) OBU. We only start a new temporal unit when
the TD comes.
But some streams do not have TD at all, which makes the output "TU"
alignment fail to work. We now add check based on the relationship
between the different layers and it can successfully judge the TU edge.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1634>
Some streams may have problematic OBUs at the beginning, which causes
the parse fail to detect the alignment and return error. For example,
there may be verbose OBUs before a valid sequence, which should be
discarded until we meet a valid sequence. We should let the parse
continue when we meet such cases, rather than just return error.
Part-of: <https://gitlab.freedesktop.org/gstreamer/gstreamer/-/merge_requests/1634>
