This was accidentally inverted...

Fixes unnatural black flickering in some scenes where encoding noise
pushes the background between 0 and 1 PQ. Adding a 1% PQ cutoff allows
censoring such values to 0.

I could also have hard-coded this constant but I decided it's simple
enough to expose, and it being tunable may help with tracking down
related issues in the future, and also allows users to opt out of this
functionality altogether.

Fixes: https://yabb.jriver.com/interact/index.php/topic,136378.msg950343.html#msg950343

It's called pl_render_*DEFAULT*_params for a reason.

Code 95% generously donated by ChatGPT. Bugs in remaining 5%.

What a headache this file was growing into...

The macOS SDK has conflicting defines with the `PAGE_SIZE` name.
Seems to affect both x86 and ARM.

Testing revealed the LUT size of 256 to have no performance downside
compared to 64, and it still fits into the hard limit for literal LUTs.
The maximum error (at ~1000x upsampling) is also on the order of 1e-4,
far surpassing any threshold of perceptibility.

Similarly, 1e-3 polar cutoff is not only completely below the dithering
noise floor at 8-bits SDR, but also below JND (~10 bits). Going any
higher than this starts to lead to visible/measurable artifacts, and
going any lower is just slower for no benefit.

Reducing option bloat, one commit at a time.

I created this for personal testing, but I ended up cleaning it up
enough to the point where it can reasonably be published.

Tests the upload/download performance in a multi-GPU scenario, including
between different VkDevices created on the same GPU. (Just pass the same
GPU name twice to test this)

Useful for benchmarking e.g. scalers on still images.

Should really have been part of the previous API bump, but oh well. We
can't support this easily in the pl_frames_infer() case, and it's also
not used by any downstream client currently. It's much simpler to just
set ICC profiles to NULL.

Re-gains us the functionality of pl_icc_params.

Addition of the pl_cache means we now persist ICC 3DLUTs across
restarts, in which case we seriously need to use the actual hash of the
profile to identify ICC profiles. Otherwise there will be cache
collisions all over the place.

Fixes: b38e1c84

Allows forcing the use of a scale pass to force all residual subpixel
offsets, instead of implicitly ignoring them.

Closes: videolan/libplacebo#291
Closes: https://github.com/mpv-player/mpv/issues/11948

For debugging purposes.

Fixes: https://github.com/haasn/libplacebo/issues/202

With xxhash, 50 MB is plenty fast enough compared to the general vulkan
init etc. overhead, and gives us plenty of room to store a number of ICC
3DLUTs and gamut mapping 3DLUTs.

xxhash is fast enough to not really worry about a few-MB large objects
slowing down init process.

For debug purposes.

Probably won't work for ICC/3DLUT cache but these probably want their
own custom mechanism anyway, since they're quite large.

No longer used by any current demo.

To be logarithmic.

This fixes leaking those headers in pkg-config file while we don't want
to do that. Especially because we prefer internal headers over system
ones.

rc.exe does not support --codepage= syntax and there is no way in meson
to ask what resource compiler it uses. Use pragma instead.

Unfortunatelly due to https://github.com/llvm/llvm-project/issues/63426
llvm-rc will use wrong code page.

To include less and not define min/max macros.

This is the wrong layer to solve this problem, and in particular,
hard-coding SDR 1.0 here prevents tone-mapping e.g. 200 nits scene to
50 nits projector. No longer needed hack now that we clamp in the tone
mapper directly.

See previous commit.

Allows scaling down the intensity of the saturation mapping step.

Sigh, no real recourse other than to brute force the mapping.

Still need to add a custom UI for the advanced tunables, but I'll defer
that to a future commit.

Fixes: https://github.com/haasn/libplacebo/issues/194

Kacper Michajłow authored 1 year ago and Niklas Haas committed 1 year ago

Fixes IsWindows10OrGreater() usage.

To completely disable smoothing, since even a smoothing period of 1.0
implies a small amount of smoothing.

This does technically regress the status quo behavior when passing 0
(which would previously see it being defaulted to 100), but with the way
options are handled these days (via the PEAK_DETECT_DEFAULTS macros) I'm
hoping to phase out the use of such implicit defaults to begin with.

It also doesn't break any current client (I checked all).

Starting from meson v1.2.1 it is unallowed to link to uninstalled static
libraries. This is how declare_dependency(link_with: ...) is treated. In
fact we want to use UCRT import lib only internaly and only when linking
shared library and/or executable. We don't want for it to be treated as
dependency of anykind, just as a directive for linker. Similar to linker
script.

Unfortunately there is not way to specify link_depends for all targets,
so we have to specify it for all of them.

Thanks to @xclaesse for his patience and idea with link_depends.

Fixes: https://github.com/haasn/libplacebo/issues/191