Complete bxdf tests #236

devshgraphicsprogramming · 2025-12-16T07:34:27Z

No description provided.

…ng/Nabla-Examples-and-Tests into bxdf_unit_tests

devshgraphicsprogramming · 2025-12-19T12:24:39Z

66_HLSLBxDFTests/main.cpp

+#include "nlohmann/json.hpp"
+
+using json = nlohmann::json;


@AnastaZIuk is this ok ?

yes, its ok

devshgraphicsprogramming · 2025-12-19T12:26:06Z

66_HLSLBxDFTests/main.cpp

 int main(int argc, char** argv)
 {
    std::cout << std::fixed << std::setprecision(4);
+    IApplicationFramework::GlobalsInit();
+    auto system = IApplicationFramework::createSystem();
+    const system::path CWD = path(argv[0]).parent_path().generic_string() + "/";
+    smart_refctd_ptr<ILogger> logger = make_smart_refctd_ptr<CColoredStdoutLoggerANSI>(ILogger::DefaultLogMask());
+    logger->log("Logger Created!", ILogger::ELL_INFO);
+    auto assetManager = make_smart_refctd_ptr<IAssetManager>(smart_refctd_ptr(system));


no, if you use the application framework you shouldn't have an int main anymore

devshgraphicsprogramming · 2025-12-19T12:46:11Z

66_HLSLBxDFTests/main.cpp

-        fprintf(stderr, "[ERROR] could not open config file\n");
+        logger->log("could not open config file\n", ILogger::ELL_ERROR);


Unformatted String Vulnerability, see https://en.wikipedia.org/wiki/Uncontrolled_format_string

devshgraphicsprogramming · 2025-12-19T12:58:05Z

66_HLSLBxDFTests/tests.h

+
+    float bin(float a)
+    {
+        float diff = std::fmod(a, stride);


nitpick, use the hlsl namespace's tgmath (or intrinsics, I'm sure that @Przemog1 added such a function) fmod function for this

I don't think there's a fmod function. tgmath only has modf, which is different I think.

devshgraphicsprogramming · 2025-12-19T12:59:26Z

66_HLSLBxDFTests/tests.h

+            if (!s.isValid())
+                continue;


it should be an error to not generate a valid sample if 100% VNDF is guaranteed by the NDF trait

#236 (comment)
I guess I'll work on differentiating it in the sample_type::makeInvalid() function

devshgraphicsprogramming · 2025-12-19T13:18:28Z

66_HLSLBxDFTests/tests.h

+            if (pdf.pdf == bit_cast<float>(numeric_limits<float>::infinity))
+                buckets[bucket] += 1;


explain this logic ?

Why are only infinite PDF samples added to a bucket !?

devshgraphicsprogramming · 2025-12-19T13:18:51Z

66_HLSLBxDFTests/tests.h

+
+        sample_t s;
+        quotient_pdf_t pdf;
+        float32_t3 bsdf;


unused variable ?

devshgraphicsprogramming · 2025-12-19T13:19:21Z

66_HLSLBxDFTests/tests.h

+                buckets[bucket] += 1;
+        }
+
+#ifndef __HLSL_VERSION


remove the ifndef, you're in C++ header now

devshgraphicsprogramming · 2025-12-19T13:21:21Z

66_HLSLBxDFTests/tests.h

+            if (!selective || b.second > 0)
+            {
+                math::Polar<float> polarCoords;
+                polarCoords.theta = b.first.x * numbers::pi<float>;
+                polarCoords.phi = b.first.y * 2.f * numbers::pi<float>;
+                const float32_t3 v = polarCoords.getCartesian();
+                base_t::errMsg += std::format("({:.3f},{:.3f},{:.3f}): {}\n", v.x, v.y, v.z, b.second);
+            }


We need some documentation about this test.

I don't get why you're only counting inf PDF samples, and why your only error is an inf PDF.

Also why the error message build is done here, and the back buckets are not appended to a list to format and print later closer to cb.__call

devshgraphicsprogramming · 2025-12-19T13:28:40Z

66_HLSLBxDFTests/tests.h

+inline float adaptiveSimpson(const std::function<float(float)>& f, float x0, float x1, float eps = 1e-6, int depth = 6)
+{
+    int count = 0;
+    std::function<float(float, float, float, float, float, float, float, float, int)> integrate = 
+    [&](float a, float b, float c, float fa, float fb, float fc, float I, float eps, int depth)
+    {
+        float d = 0.5f * (a + b);
+        float e = 0.5f * (b + c);
+        float fd = f(d);
+        float fe = f(e);
+
+        float h = c - a;
+        float I0 = (1.0f / 12.0f) * h * (fa + 4 * fd + fb);
+        float I1 = (1.0f / 12.0f) * h * (fb + 4 * fe + fc);
+        float Ip = I0 + I1;
+        count++;
+
+        if (depth <= 0 || std::abs(Ip - I) < 15 * eps)
+            return Ip + (1.0f / 15.0f) * (Ip - I);
+
+        return integrate(a, d, b, fa, fd, fb, I0, .5f * eps, depth - 1) +
+                integrate(b, e, c, fb, fe, fc, I1, .5f * eps, depth - 1);
+    };
+
+    float a = x0;
+    float b = 0.5f * (x0 + x1);
+    float c = x1;
+    float fa = f(a);
+    float fb = f(b);
+    float fc = f(c);
+    float I = (c - a) * (1.0f / 6.0f) * (fa + 4.f * fb + fc);
+    return integrate(a, b, c, fa, fb, fc, I, eps, depth);
+}
+
+inline float adaptiveSimpson2D(const std::function<float(float, float)>& f, float32_t2 x0, float32_t2 x1, float eps = 1e-6, int depth = 6)
+{
+    const auto integrate = [&](float y) -> float
+    {
+        return adaptiveSimpson(std::bind(f, std::placeholders::_1, y), x0.x, x1.x, eps, depth);
+    };
+    return adaptiveSimpson(integrate, x0.y, x1.y, eps, depth);
+}


replace std::function with a generic template functor, right now you're doing virtual calls on every invocation.

Also it would make sense to move this to Nabla somwhere in https://github.com/Devsh-Graphics-Programming/Nabla/tree/master/include/nbl/builtin/hlsl/math/quadrature

because we have a similar function to compute integrals https://github.com/Devsh-Graphics-Programming/Nabla/blob/3f5b2c33e07d43c6bfd78193a8c75ffa11842ecf/include/nbl/builtin/hlsl/math/quadrature/gauss_legendre/gauss_legendre.hlsl#L22

You can do recursion in HLSL a slightly cursed way -> https://godbolt.org/z/6rvE8rEqW but please do compile-time depth so we don't kill register occupancy

@Erfan-Ahmadi look at the new toys you're getting to play with

devshgraphicsprogramming · 2025-12-19T13:34:31Z

66_HLSLBxDFTests/tests.h

+    double RLGamma(double a, double x) {
+        const double epsilon = 0.000000000000001;
+        const double big = 4503599627370496.0;
+        const double bigInv = 2.22044604925031308085e-16;
+        assert(a >= 0 && x >= 0);
+
+        if (x == 0)
+            return 0.0f;
+
+        double ax = (a * std::log(x)) - x - std::lgamma(a);
+        if (ax < -709.78271289338399)
+            return a < x ? 1.0 : 0.0;
+
+        if (x <= 1 || x <= a)
+        {
+            double r2 = a;
+            double c2 = 1;
+            double ans2 = 1;
+
+            do {
+                r2 = r2 + 1;
+                c2 = c2 * x / r2;
+                ans2 += c2;
+            } while ((c2 / ans2) > epsilon);
+
+            return std::exp(ax) * ans2 / a;
+        }
+
+        int c = 0;
+        double y = 1 - a;
+        double z = x + y + 1;
+        double p3 = 1;
+        double q3 = x;
+        double p2 = x + 1;
+        double q2 = z * x;
+        double ans = p2 / q2;
+        double error;
+
+        do {
+            c++;
+            y += 1;
+            z += 2;
+            double yc = y * c;
+            double p = (p2 * z) - (p3 * yc);
+            double q = (q2 * z) - (q3 * yc);
+
+            if (q != 0)
+            {
+                double nextans = p / q;
+                error = std::abs((ans - nextans) / nextans);
+                ans = nextans;
+            }
+            else
+            {
+                error = 1;
+            }
+
+            p3 = p2;
+            p2 = p;
+            q3 = q2;
+            q2 = q;
+
+            if (std::abs(p) > big)
+            {
+                p3 *= bigInv;
+                p2 *= bigInv;
+                q3 *= bigInv;
+                q2 *= bigInv;
+            }
+        } while (error > epsilon);
+
+        return 1.0 - (std::exp(ax) * ans);
+    }


This is https://en.wikipedia.org/wiki/Incomplete_gamma_function

Which in Boost is in this file https://www.boost.org/doc/libs/latest/libs/math/doc/html/math_toolkit/sf_gamma/igamma.html

C++ doesn't have incomplete gamma functions, but other "special functions" like Bessel, would live in tgmath (cmath, but we always template everything for good measure)

For now you can constrain the implementation to a scalar.

devshgraphicsprogramming · 2025-12-19T13:36:43Z

66_HLSLBxDFTests/tests.h

+    smart_refctd_ptr<ICPUImage> writeToCPUImage()
+    {


nitpick, instead of burying this function a test, I'd rather have it in main.cpp as a function that does

smart_refctd_ptr<ICPUImage> writeAsCPUImage(const TestChi2<BxDF,aniso>&)

devshgraphicsprogramming · 2025-12-19T13:37:23Z

66_HLSLBxDFTests/tests.h

+        const auto format = E_FORMAT::EF_R32G32B32A32_SFLOAT;
+
+        IImage::SCreationParams imageParams = {};
+        imageParams.flags = static_cast<asset::IImage::E_CREATE_FLAGS>(asset::IImage::ECF_MUTABLE_FORMAT_BIT | asset::IImage::ECF_EXTENDED_USAGE_BIT);


nitpick, you shouldn't care about this when you're not making GPU images from the CPU image via asset converter or multiple views with different formats

devshgraphicsprogramming · 2025-12-19T13:39:09Z

66_HLSLBxDFTests/tests.h

+        // write sample count from generate, top half
+        for (uint64_t j = 0u; j < thetaSplits; ++j)
+            for (uint64_t i = 0; i < phiSplits; ++i)
+            {
+                float32_t3 pixelColor = mapColor(countFreq[j * phiSplits + i], 0.f, maxCountFreq);
+                double decodedPixel[4] = { pixelColor[0], pixelColor[1], pixelColor[2], 1 };
+
+                const uint64_t pixelIndex = j * phiSplits + i;
+                asset::encodePixelsRuntime(format, bytePtr + pixelIndex * asset::getTexelOrBlockBytesize(format), decodedPixel);
+            }
+
+        // write values of pdf, bottom half
+        for (uint64_t j = 0u; j < thetaSplits; ++j)
+            for (uint64_t i = 0; i < phiSplits; ++i)
+            {
+                float32_t3 pixelColor = mapColor(integrateFreq[j * phiSplits + i], 0.f, maxIntFreq);
+                double decodedPixel[4] = { pixelColor[0], pixelColor[1], pixelColor[2], 1 };
+
+                const uint64_t pixelIndex = (thetaSplits + j) * phiSplits + i;
+                asset::encodePixelsRuntime(format, bytePtr + pixelIndex * asset::getTexelOrBlockBytesize(format), decodedPixel);
+            }


I'd just write first image into RGB, and second into Alpha.

Since you're paying for RGBA full fp32 eacvh channel, and one can easily turn off certain channels or view just alpha in Renderdoc and GIMP (Can leave this as a comment)

Both scalar values are mapped to rgb though, where blue is low freq and going towards red is increasing freq. Like this

I'm not sure if compressing range down to just 1 channel will be clear to see, but could try

devshgraphicsprogramming · 2025-12-19T13:44:35Z

66_HLSLBxDFTests/tests.h

+    float32_t3 mapColor(float v, float vmin, float vmax)
+    {
+        float32_t3 c = float32_t3(1, 1, 1);
+        float diff = vmax - vmin;
+        v = clamp<float>(v, vmin, vmax);
+
+        if (v < (vmin + 0.25f * diff))
+        {
+            c.r = 0;
+            c.g = 4.f * (v - vmin) / diff;
+        }
+        else if (v < (vmin + 0.5f * diff))
+        {
+            c.r = 0;
+            c.b = 1.f + 4.f * (vmin + 0.25f * diff - v) / diff;
+        }
+        else if (v < (vmin + 0.75f * diff))
+        {
+            c.r = 4.f * (v - vmin - 0.5f * diff) / diff;
+            c.b = 0;
+        }
+        else
+        {
+            c.g = 1.f + 4.f * (vmin + 0.75f * diff - v) / diff;
+            c.b = 0;
+        }
+
+        return c;
+    }


I'd add this to nbl::hlsl::visualization::false_color and keep our Sci-viz gradient there

@AnastaZIuk same thing for your IESes

Some color mappings have particular names in scientific literature https://docs.paraview.org/en/latest/ReferenceManual/colorMapping.html

https://pmc.ncbi.nlm.nih.gov/articles/PMC4959790/

https://www.youtube.com/watch?v=tJ6aEIB61hY & https://www.fabiocrameri.ch/colourmaps/

yea we could have nbl::hlsl::visualization:: stuff

mine is blue-magenta, I don't think there is a canonical name for it (cool-like in scientific literature)

Nabla-Examples-and-Tests/50.IESViewer/app_resources/ies.unified.hlsl

Line 69 in d0650a3

float32_t3 falseColor(float32_t v)

and almost monotonic in luminance (mapColor does not appear to be https://www.kennethmoreland.com/color-advice/)

It corresponds to jet colormap in matplotlib
https://www.mathworks.com/help/matlab/ref/jet.html

devshgraphicsprogramming · 2025-12-19T13:47:51Z

66_HLSLBxDFTests/tests.h

+            if (!s.isValid())
+                continue;
+


same comment as always about VNDF

devshgraphicsprogramming · 2025-12-19T13:52:22Z

66_HLSLBxDFTests/tests.h

+        if (write_frequencies)
+            writeToEXR(assetManager);


we're liklely to drown in temp files, maybe have an option to only write failed tests' frequencies instead of always ?

basically turning write_frequencies into an enum

Also if you want the tests to be thread safe and parallizable, you'd need to somehow document that you expect that all base_t::rc.halfSeed to be unique (with a wide enough seed, good enough rng you shouldn't get collisions in rng output until veeeeeery large test instance counts)

halfSeed right now is the iteration number. Goes into a PCG32 hash to produce uint32_t2 seed for Xoroshiro rng, which is what is used throughout the tests. Is that enough?

AnastaZIuk · 2025-12-20T14:02:11Z

66_HLSLBxDFTests/main.cpp

+#include "nlohmann/json.hpp"
+
+using json = nlohmann::json;


yes, its ok

AnastaZIuk · 2025-12-20T14:11:00Z

66_HLSLBxDFTests/tests.h

+#define GLM_ENABLE_EXPERIMENTAL
+#include <glm/gtx/hash.hpp>


no conflict, gtx stuff is opt-in and not included in nabla, as long as we are not planning to use this experimental feature in our public headers it should stay like this right there

if we ever do then we just move

#define GLM_ENABLE_EXPERIMENTAL #include <glm/gtx/hash.hpp>

to nbl/builtin/hlsl/cpp_compat/vector.hlsl

AnastaZIuk · 2025-12-20T14:31:14Z

66_HLSLBxDFTests/tests.h

+    float32_t3 mapColor(float v, float vmin, float vmax)
+    {
+        float32_t3 c = float32_t3(1, 1, 1);
+        float diff = vmax - vmin;
+        v = clamp<float>(v, vmin, vmax);
+
+        if (v < (vmin + 0.25f * diff))
+        {
+            c.r = 0;
+            c.g = 4.f * (v - vmin) / diff;
+        }
+        else if (v < (vmin + 0.5f * diff))
+        {
+            c.r = 0;
+            c.b = 1.f + 4.f * (vmin + 0.25f * diff - v) / diff;
+        }
+        else if (v < (vmin + 0.75f * diff))
+        {
+            c.r = 4.f * (v - vmin - 0.5f * diff) / diff;
+            c.b = 0;
+        }
+        else
+        {
+            c.g = 1.f + 4.f * (vmin + 0.75f * diff - v) / diff;
+            c.b = 0;
+        }
+
+        return c;
+    }


yea we could have nbl::hlsl::visualization:: stuff

mine is blue-magenta, I don't think there is a canonical name for it (cool-like in scientific literature)

Nabla-Examples-and-Tests/50.IESViewer/app_resources/ies.unified.hlsl

Line 69 in d0650a3

float32_t3 falseColor(float32_t v)

and almost monotonic in luminance (mapColor does not appear to be https://www.kennethmoreland.com/color-advice/)

AnastaZIuk · 2025-12-20T14:33:18Z

66_HLSLBxDFTests/main.cpp

+        options.preprocessorOptions.includeFinder = compiler->getDefaultIncludeFinder();
+        const IShaderCompiler::SMacroDefinition WorkgroupSizeDefine = { "WORKGROUP_SIZE", WorkgroupSizeAsStr };
+        options.preprocessorOptions.extraDefines = { &WorkgroupSizeDefine,&WorkgroupSizeDefine + 1 };
+        if (!(shader = compiler->compileToSPIRV((const char*)shaderSrc->getContent()->getPointer(), options)))


precompile https://github.com/Devsh-Graphics-Programming/Nabla/blob/master/docs/nsc-prebuilds.md

you are injecting WORKGROUP_SIZE but value you use is constexpr, you can move this one to your example's CMake. Normally you would just pass -D to NSC compile options (optionally to cpp compiler if required) but since we have a bug (I assigned @Przemog1 to fix) temporary you need to create "proxy" input in which you put your define and include your original input

set(OUTPUT_DIRECTORY "${CMAKE_CURRENT_BINARY_DIR}/auto-gen") file(MAKE_DIRECTORY "${OUTPUT_DIRECTORY}") # .. # read docs/nsc-prebuilds.md set(WORKGROUP_SIZE 256u) # you might turn into option to toggle in GUI set(PROXY_TEST_COMPILE_COMP_HLSL "${OUTPUT_DIRECTORY}/proxy_test_compile.comp.hlsl") set(TEST_COMPILE_VIEW [=[ #define WORKGROUP_SIZE @WORKGROUP_SIZE@ #include "app_resources/test_compile.comp.hlsl" ]=]) string(CONFIGURE "${TEST_COMPILE_VIEW}" TEST_COMPILE_CONTENT @ONLY) file(WRITE "${PROXY_TEST_COMPILE_COMP_HLSL}" "${TEST_COMPILE_CONTENT}") # .. # remember to add -I "${CMAKE_CURRENT_SOURCE_DIR}" to COMPILE_OPTIONS # ${PROXY_TEST_COMPILE_COMP_HLSL} is your input to precompile which you load at runtime

when NSC's bug is fixed you just add -D WORKGROUP_SIZE=${WORKGROUP_SIZE} to COMPILE_OPTIONS and don't write proxy file

AnastaZIuk · 2025-12-20T15:11:10Z

66_HLSLBxDFTests/main.cpp

+    BOOST_PP_REMOVE_PARENS(TEST_TYPE)::run(INIT_PARAMS, cb __VA_OPT__(,) __VA_ARGS__);\
+}\
+
 int main(int argc, char** argv)


you want to create class and inherit from application_templates:: + BuiltinResourcesApplication, check other examples for usage, no int main

AnastaZIuk

I left some comments, please address

…hould already be invalid

… on error

keptsecret and others added 30 commits January 15, 2025 11:50

check util functions

3668d99

reciprocity test

a66dc44

negative value checks

b5d9778

Merge branch 'bxdf_unit_tests' of github.com:Devsh-Graphics-Programmi…

48cc64c

…ng/Nabla-Examples-and-Tests into bxdf_unit_tests

fix test orders

5df264a

Merge branch 'bxdf_unit_tests' of github.com:Devsh-Graphics-Programmi…

983efc3

…ng/Nabla-Examples-and-Tests into bxdf_unit_tests

use new param structs for bxdf funcs

e6e434a

print more info on error

d494e3d

Merge branch 'master' into bxdf_unit_tests

3305441

fix reciprocity test, adjust comparison

3625ad2

fixed reciprocity test

839d42c

test buckets of inf

a40ba94

bucket by polar coords

d0d5d93

chi2 test

c8f8dae

Merge branch 'master' into bxdf_unit_tests

8ea1538

Merge branch 'master' into bxdf_unit_tests

dcc2927

fix chi2 test, beckmann still error

8781cb0

add visualization option for chi2 samples

724acd5

fixed exr visualization

e1cbc1a

use bxdf traits in test, better chi2 vis

9387298

test config with json

df07bfd

Merge branch 'master' into bxdf_unit_tests

de11fe2

fixed json config bugs

da1a4ad

smooth dielectric tests

d7f7a87

fix pcg usage

111c11b

Merge branch 'master' into bxdf_unit_tests

83d8a92

update with master

db74fad

refactor according to changes with bxdf

f3144f1

added a compile shader test

5a9e8b5

fix compile test

a133452

devshgraphicsprogramming commented Dec 19, 2025

View reviewed changes

devshgraphicsprogramming requested a review from AnastaZIuk December 19, 2025 12:56

devshgraphicsprogramming commented Dec 19, 2025

View reviewed changes

66_HLSLBxDFTests/tests.h Outdated

sample_t s;

quotient_pdf_t pdf;

float32_t3 bsdf;

Copy link

Member Author

devshgraphicsprogramming Dec 19, 2025

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

unused variable ?

devshgraphicsprogramming commented Dec 19, 2025

View reviewed changes

AnastaZIuk reviewed Dec 20, 2025

View reviewed changes

AnastaZIuk requested changes Dec 20, 2025

View reviewed changes

keptsecret added 10 commits December 22, 2025 16:20

renamed generated sample pdf = 0 test, eta/etak is rgb

c162839

rename test result enums to make clearer

403cd69

static assert concepts, indent inside for loop macros

249f943

added test for reflectance, adjusted ndf test

8b04878

vndf should not produce invalid samples for correct test config

1045c6a

Merge branch 'master' into iridescent_bxdf_tests

9834314

Merge branch 'master' into iridescent_bxdf_tests

8c5e41a

reorder some tests in jacobian, removed incorrect test where sample s…

565c2d3

…hould already be invalid

moved tests to using BuiltinResourcesApplication application framework

fec9948

changed chi2 test write frequencies to enum with option to only write…

444203f

… on error

		fprintf(stderr, "[ERROR] could not open config file\n");
		logger->log("could not open config file\n", ILogger::ELL_ERROR);

		if (pdf.pdf == bit_cast<float>(numeric_limits<float>::infinity))
		buckets[bucket] += 1;

Complete bxdf tests #236

Are you sure you want to change the base?

Complete bxdf tests #236

Uh oh!

Conversation

devshgraphicsprogramming commented Dec 16, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

devshgraphicsprogramming Dec 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

devshgraphicsprogramming Dec 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

devshgraphicsprogramming Dec 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

devshgraphicsprogramming Dec 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

AnastaZIuk left a comment

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

devshgraphicsprogramming Dec 19, 2025 •

edited

Loading

devshgraphicsprogramming Dec 19, 2025 •

edited

Loading

devshgraphicsprogramming Dec 19, 2025 •

edited

Loading

devshgraphicsprogramming Dec 19, 2025 •

edited

Loading