4e-3 was an extremely optimistic, and in practice too low, cutoff value,
resulting in very bad 3...

vulkan/stubs.c also depends on vulkan headers.

It should return empty string if it is not defined, but Looks like meson
returns ' ' in some cases.

When reducing the intensity we have to also reduce the saturation to
compensate, similar to how ...

Bleh, didn't trigger locally..

Typically the case for synthetic test profiles that rotate hue vectors
entirely onto different p...

Explicitly reset `csp->hdr.prim`, in case this contains other values
from e.g. the swapchain.

When forcing HDR levels, we currently override also the primaries
struct, which is quite an unde...

These XYZ values are normalized to the white point, so adjust the
display from the incorrect mcd...

To help debug issues related to ICC profiles without necessarily needing
to provide the ICC prof...

Need to take a proper fabsf of the crop dimensions.

Return an empty frame mix until the first frame's PTS value has been
reached.

Probably wouldn't happen anyway on any sane compiler, but nonetheless.

To avoid collisions with other vars named 'tmp'

Fixes: https://code.videolan.org/videolan/libplacebo/-/issues/284

In case output is flipped/mirrored.

Fixes: 3fd65b2655ede8498728d9940cf21ccc9ce0b4c3

Avoids excessively over-saturating when tone mapping upwards in
brightness, which generally lead...

These were forgotten in earlier changes.

This avoids using designated initializers (compat issues with C++20)

Since this comes with a significant performance and CPU cost enhancement
at only minor debuggabi...

Instead of implicitly clamping the output image to the originally
specified dst_rect, this allow...

Helper function to compute bounding boxes of transformations.

Mostly for cosmetic reasons, though I plan on extending at least one
instance of this.

After a few iterations I settled on this, which is essentially a
HDR-specific sharpening filter,...

I want to extend this with more "runtime"-provided (i.e. not
user-facing) parameters. Having lea...

Requires generation of a low-res feature map. To avoid excessive ringing
artefacts, hard-code th...

Make this way less hacky and also more easily extended.

If gray maps to solid black, as in some extreme benchmark/test profiles,
y_approx ends up as inf...

Lest the very fabric of society fall apart and the world descend into
chaos.

The upscaler tests were actually no-ops because the output crop was
effectively hard-coded as 1x...

1/6.0 etc. are not legal constants in GLES, need to write as 1.0/6.0.

Mostly for demonstration purposes.

Could be done better (e.g. mipmapping to reduce aliasing, arbitrary EWA
interpolation, ...) but ...

Just to make sure the shaders compile and the main logic branch is
covered.

For hooking the point immediately before alpha blending, for any shaders
that want to make a las...

Simple helper function for a common-ish operation.

Mostly just as a convenience to the user, to avoid having to hook
PL_HOOK_PRE_OUTPUT and ship a ...

Needed for new feature.

This only affects pass_output_target and does not make it into the
cache.

Removes the clunky stateful peak_detect_active bool in exchange for a
properly scoped temporary ...

Make it more explicit which case is which.

This variable was never currently set.

MinGW-w64 math functions are significantly slower than the UCRT ones.
In particular powf is over...

With linear interpolation, this gives enough precision even on a
256-element LUT, to pass the te...

To test impact of performance improvements to the 3DLUT generation code

Currently `include_directories` are ignored when used as `dependencies`
in `has_header_symbol`. ...

Prefer our Vulkan-Headers as they are always available and with correct
version. Also remove ver...

In the case of importing a planar texture, we only need to synchronize
the base texture, while t...

In case of emulated texture format we have to upload manually to reorder
the data.

Useful to know where the errors are coming from. Also log source with
PL_LOG_DEBUG same as other...

Ensures that red maps (approximately) to red, cyan to cyan, and so on.
This is not an exact test...

Apply a hue shift to forcibly ensure primaries map to primaries, e.g.
red to red, and so on. Thi...

Improves the visualization by allowing me to effectively rotate the
isohue vectors with respect ...

We need to avoid stripping alpha information when blending, which
requires both persisting alpha...

We currently error if the crop expr exceeds the video frame dimensions.
Handle this properly by ...

Wrong variable used in clamp.

Sometimes these are tagged as RGB. PNG in particular likes to give such
files for "gray" format,...

The target frame could accidentally be double-acquired under the old
logic, leading to deadlock....

For consistency with both the Vulkan spec and the revised version of the
lavu vulkan patchset. T...

To avoid discontinuity in black point handling.

Fixes: https://code.videolan.org/videolan/libpl...

Two for the price of one.

Very cheap for us to support, just treat it as an on-the-fly alpha mask.

Direct3D 11 runtime only guarantees that allocations within the maximum
resource size are suppor...

We need to clean up before returning.

Instead of half-assing it with a #define, which prevents the use of
mipmapped textures (e.g. in ...

Instead of accounting for the texture size manually in
sh_bind/setup_src.

We (unfortunately) st...

The old interlacing fields got deprecated.

We should *exclude* the target_pts = frame_pts + dur case, because that
would be the start of th...

Instead of returning EOF here early, only return EOF once the last frame
should no longer be vis...

And ignore safety clamp in this case.

Treat an EOF after partial frame as EOF, instead of forwarding this
frame to the output.

No reason for this not to exist, especially as it's useful to do linear
transitions/fades.

To avoid the trappings of users directly accessing `frame->user_data`,
which led to the current ...

This is no longer just a straight pointer to AVFrame, and absolutely
should be cleaned up. We ca...

And actually, exposing this implementation detail was a bad idea in the
first place.

Now that we also validate DXGI calls.

It is not recommended by FFmpeg as some decoders have issues.

DXGI debug interface encapsulate multiple message queues, which allows
to get validation not onl...

Older MinGW-w64 doesn't define IDXGIInfoQueue in dxgidebug.h.

Instead of universally enabling all supported QFs, this field lets us
whitelist the set of permi...

Avoids leaking this memory, and ties its lifetime explicitly to the
mapped AVFrame.

So I can add more fields to this than just the `AVFrame`. Motivated by a
desire to keep track of...

Wrong index was used resulting in intermittent register allocation
conflicts.

Fixes: https://gi...

This was bugged in the same way, due to mismatch between VK_VERSION_1_2
and VK_API_VERSION_1_2...

When using this as an extension (rather than promoted command), we
cannot omit the KHR suffix. F...

These macros expand to 1, effectively disabling the intended behavior.