Adds Metal Shader Generation and Rendering Support

.github/workflows/main.yml

@@ -81,6 +81,7 @@ jobs:
           compiler: xcode
           compiler_version: "13.3"
           python: 3.9
+          test_shaders: ON
 
         - name: Windows_VS2019_Win32_Python27
           os: windows-2019
@@ -202,16 +203,22 @@ jobs:
         python Scripts/generateshader.py ../resources/Materials/Examples/StandardSurface --path .. --target glsl
         python Scripts/generateshader.py ../resources/Materials/Examples/StandardSurface --path .. --target osl
         python Scripts/generateshader.py ../resources/Materials/Examples/StandardSurface --path .. --target mdl
+        python Scripts/generateshader.py ../resources/Materials/Examples/StandardSurface --path .. --target msl
       working-directory: python
 
-    - name: Shader Validation Tests
+    - name: Shader Validation Tests (Windows)
       if: matrix.test_shaders == 'ON' && runner.os == 'Windows'
       run: |
         vcpkg/vcpkg install glslang --triplet=x64-windows
         glslangValidator.exe -v
         python python/Scripts/generateshader.py resources/Materials/Examples/StandardSurface --path . --target glsl --validator glslangValidator.exe --vulkanGlsl True --validatorArgs="-V --aml"
         python python/Scripts/generateshader.py resources/Materials/Examples/StandardSurface --path . --target essl --validator glslangValidator.exe
 
+    - name: Shader Validation Tests (MacOS)
+      if: matrix.test_shaders == 'ON' && runner.os == 'macOS'
+      run: |
+        python python/Scripts/generateshader.py resources/Materials/Examples/StandardSurface --path . --target msl --validator "xcrun metal --language=metal" --validatorArgs="-w"
+
     - name: Static Analysis Tests
       if: matrix.static_analysis == 'ON' && runner.os == 'Linux'
       run: |

CMakeLists.txt

@@ -41,6 +41,7 @@ option(MATERIALX_BUILD_DOCS "Create HTML documentation using Doxygen. Requires t
 option(MATERIALX_BUILD_GEN_GLSL "Build the GLSL shader generator back-end." ON)
 option(MATERIALX_BUILD_GEN_OSL "Build the OSL shader generator back-end." ON)
 option(MATERIALX_BUILD_GEN_MDL "Build the MDL shader generator back-end." ON)
+option(MATERIALX_BUILD_GEN_MSL "Build the MSL shader generator back-end." ON)
 option(MATERIALX_BUILD_RENDER "Build the MaterialX Render modules." ON)
 option(MATERIALX_BUILD_OIIO "Build OpenImageIO support for MaterialXRender." OFF)
 option(MATERIALX_BUILD_TESTS "Build unit tests." ON)
@@ -105,6 +106,7 @@ mark_as_advanced(MATERIALX_BUILD_DOCS)
 mark_as_advanced(MATERIALX_BUILD_GEN_GLSL)
 mark_as_advanced(MATERIALX_BUILD_GEN_OSL)
 mark_as_advanced(MATERIALX_BUILD_GEN_MDL)
+mark_as_advanced(MATERIALX_BUILD_GEN_MSL)
 mark_as_advanced(MATERIALX_BUILD_RENDER)
 mark_as_advanced(MATERIALX_BUILD_OIIO)
 mark_as_advanced(MATERIALX_BUILD_TESTS)
@@ -235,7 +237,7 @@ add_subdirectory(source/MaterialXFormat)
 
 # Add shader generation subdirectories
 add_subdirectory(source/MaterialXGenShader)
-if(MATERIALX_BUILD_GEN_GLSL OR MATERIALX_BUILD_GEN_OSL OR MATERIALX_BUILD_GEN_MDL)
+if(MATERIALX_BUILD_GEN_GLSL OR MATERIALX_BUILD_GEN_OSL OR MATERIALX_BUILD_GEN_MDL OR MATERIALX_BUILD_GEN_MSL)
     if (MATERIALX_BUILD_GEN_GLSL)
         add_definitions(-DMATERIALX_BUILD_GEN_GLSL)
         add_subdirectory(source/MaterialXGenGlsl)
@@ -248,6 +250,10 @@ if(MATERIALX_BUILD_GEN_GLSL OR MATERIALX_BUILD_GEN_OSL OR MATERIALX_BUILD_GEN_MD
         add_definitions(-DMATERIALX_BUILD_GEN_MDL)
         add_subdirectory(source/MaterialXGenMdl)
     endif()
+    if (MATERIALX_BUILD_GEN_MSL)
+        add_definitions(-DMATERIALX_BUILD_GEN_MSL)
+        add_subdirectory(source/MaterialXGenMsl)
+    endif()
     add_subdirectory(libraries)
 endif()
 
@@ -258,6 +264,9 @@ if(MATERIALX_BUILD_RENDER)
     if (MATERIALX_BUILD_GEN_GLSL)
         add_subdirectory(source/MaterialXRenderGlsl)
     endif()
+    if (APPLE AND MATERIALX_BUILD_GEN_MSL)
+        add_subdirectory(source/MaterialXRenderMsl)
+    endif()
     if (MATERIALX_BUILD_GEN_OSL)
         add_subdirectory(source/MaterialXRenderOsl)
     endif()

libraries/lights/genmsl/lights_genmsl_impl.mtlx

@@ -0,0 +1,13 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<materialx version="1.37">
+
+  <!-- <point_light> -->
+  <implementation name="IM_point_light_genmsl" nodedef="ND_point_light" file="mx_point_light.metal" function="mx_point_light" target="genmsl"/>
+
+  <!-- <directional_light> -->
+  <implementation name="IM_directional_light_genmsl" nodedef="ND_directional_light" file="mx_directional_light.metal" function="mx_directional_light" target="genmsl"/>
+
+  <!-- <spot_light> -->
+  <implementation name="IM_spot_light_genmsl" nodedef="ND_spot_light" file="mx_spot_light.metal" function="mx_spot_light" target="genmsl"/>
+
+</materialx>

libraries/lights/genmsl/mx_directional_light.metal

@@ -0,0 +1,5 @@
+void mx_directional_light(LightData light, float3 position, thread lightshader& result)
+{
+    result.direction = -light.direction;
+    result.intensity = light.color * light.intensity;
+}

libraries/lights/genmsl/mx_point_light.metal

@@ -0,0 +1,8 @@
+void mx_point_light(LightData light, float3 position, thread lightshader& result)
+{
+    result.direction = light.position - position;
+    float distance = length(result.direction) + M_FLOAT_EPS;
+    float attenuation = pow(distance + 1.0, light.decay_rate + M_FLOAT_EPS);
+    result.intensity = light.color * light.intensity / attenuation;
+    result.direction /= distance;
+}

libraries/lights/genmsl/mx_spot_light.metal

@@ -0,0 +1,13 @@
+void mx_spot_light(LightData light, float3 position, thread lightshader& result)
+{
+    result.direction = light.position - position;
+    float distance = length(result.direction) + M_FLOAT_EPS;
+    float attenuation = pow(distance + 1.0, light.decay_rate + M_FLOAT_EPS);
+    result.intensity = light.color * light.intensity / attenuation;
+    result.direction /= distance;
+    float low = min(light.inner_angle, light.outer_angle);
+    float high = light.inner_angle;
+    float cosDir = dot(result.direction, -light.direction);
+    float spotAttenuation = smoothstep(low, high, cosDir);
+    result.intensity *= spotAttenuation;
+}

libraries/pbrlib/genglsl/lib/mx_microfacet_specular.glsl

@@ -30,6 +30,26 @@ struct FresnelData
 
     // Refraction
     bool refraction;
+
+#ifdef __METAL__ 
+FresnelData(int   _model        = 0, 
+            vec3  _ior          = vec3(0.0f),
+            vec3  _extinction   = vec3(0.0f),
+            vec3  _F0           = vec3(0.0f),
+            vec3  _F90          = vec3(0.0f),
+            float _exponent     = 0.0f,
+            float _tf_thickness = 0.0f,
+            float _tf_ior       = 0.0f,
+            bool  _refraction   = false) : 
+                model(_model),
+                ior(_ior),
+                extinction(_extinction),
+                F0(_F0), F90(_F90), exponent(_exponent),
+                tf_thickness(_tf_thickness),
+                tf_ior(_tf_ior),
+                refraction(_refraction) {}
+#endif
+
 };
 
 // https://media.disneyanimation.com/uploads/production/publication_asset/48/asset/s2012_pbs_disney_brdf_notes_v3.pdf
@@ -333,11 +353,12 @@ void mx_fresnel_dielectric_phase_polarized(float cosTheta, float eta1, float eta
 // Phase shift due to a conducting material
 void mx_fresnel_conductor_phase_polarized(float cosTheta, float eta1, vec3 eta2, vec3 kappa2, out vec3 phiP, out vec3 phiS)
 {
-    if (kappa2 == vec3(0, 0, 0) && eta2.x == eta2.y && eta2.y == eta2.z) {
+    if (dot(kappa2, kappa2) == 0.0 && eta2.x == eta2.y && eta2.y == eta2.z) {
         // Use dielectric formula to increase performance
-        mx_fresnel_dielectric_phase_polarized(cosTheta, eta1, eta2.x, phiP.x, phiS.x);
-        phiP = phiP.xxx;
-        phiS = phiS.xxx;
+        float phiPx, phiSx;
+        mx_fresnel_dielectric_phase_polarized(cosTheta, eta1, eta2.x, phiPx, phiSx);
+        phiP = vec3(phiPx, phiPx, phiPx);
+        phiS = vec3(phiSx, phiSx, phiSx);
         return;
     }
     vec3 k2 = kappa2 / eta2;

libraries/pbrlib/genmsl/pbrlib_genglsl_impl.mtlx

@@ -0,0 +1,74 @@
+<?xml version="1.0"?>
+<materialx version="1.38">
+
+  <!-- <oren_nayar_diffuse_bsdf> -->
+  <implementation name="IM_oren_nayar_diffuse_bsdf_genmsl" nodedef="ND_oren_nayar_diffuse_bsdf" file="../genglsl/mx_oren_nayar_diffuse_bsdf.glsl" function="mx_oren_nayar_diffuse_bsdf" target="genmsl" />
+
+  <!-- <burley_diffuse_bsdf> -->
+  <implementation name="IM_burley_diffuse_bsdf_genmsl" nodedef="ND_burley_diffuse_bsdf" file="../genglsl/mx_burley_diffuse_bsdf.glsl" function="mx_burley_diffuse_bsdf" target="genmsl" />
+
+  <!-- <translucent_bsdf> -->
+  <implementation name="IM_translucent_bsdf_genmsl" nodedef="ND_translucent_bsdf" file="../genglsl/mx_translucent_bsdf.glsl" function="mx_translucent_bsdf" target="genmsl" />
+
+  <!-- <dielectric_bsdf> -->
+  <implementation name="IM_dielectric_bsdf_genmsl" nodedef="ND_dielectric_bsdf" file="../genglsl/mx_dielectric_bsdf.glsl" function="mx_dielectric_bsdf" target="genmsl" />
+
+  <!-- <conductor_bsdf> -->
+  <implementation name="IM_conductor_bsdf_genmsl" nodedef="ND_conductor_bsdf" file="../genglsl/mx_conductor_bsdf.glsl" function="mx_conductor_bsdf" target="genmsl" />
+
+  <!-- <generalized_schlick_bsdf> -->
+  <implementation name="IM_generalized_schlick_bsdf_genmsl" nodedef="ND_generalized_schlick_bsdf" file="../genglsl/mx_generalized_schlick_bsdf.glsl" function="mx_generalized_schlick_bsdf" target="genmsl" />
+
+  <!-- <subsurface_bsdf> -->
+  <implementation name="IM_subsurface_bsdf_genmsl" nodedef="ND_subsurface_bsdf" file="../genglsl/mx_subsurface_bsdf.glsl" function="mx_subsurface_bsdf" target="genmsl" />
+
+  <!-- <sheen_bsdf> -->
+  <implementation name="IM_sheen_bsdf_genmsl" nodedef="ND_sheen_bsdf" file="../genglsl/mx_sheen_bsdf.glsl" function="mx_sheen_bsdf" target="genmsl" />
+
+  <!-- <anisotropic_vdf> -->
+  <implementation name="IM_anisotropic_vdf_genmsl" nodedef="ND_anisotropic_vdf" file="../genglsl/mx_anisotropic_vdf.glsl" function="mx_anisotropic_vdf" target="genmsl" />
+
+  <!-- <thin_film_bsdf> -->
+  <implementation name="IM_thin_film_bsdf_genmsl" nodedef="ND_thin_film_bsdf" target="genmsl" />
+
+  <!-- <layer> -->
+  <implementation name="IM_layer_bsdf_genmsl" nodedef="ND_layer_bsdf" target="genmsl" />
+  <implementation name="IM_layer_vdf_genmsl" nodedef="ND_layer_vdf" target="genmsl" />
+
+  <!-- <mix> -->
+  <implementation name="IM_mix_bsdf_genmsl" nodedef="ND_mix_bsdf" target="genmsl" />
+  <implementation name="IM_mix_edf_genmsl" nodedef="ND_mix_edf" target="genmsl" />
+
+  <!-- <add> -->
+  <implementation name="IM_add_bsdf_genmsl" nodedef="ND_add_bsdf" target="genmsl" />
+  <implementation name="IM_add_edf_genmsl" nodedef="ND_add_edf" target="genmsl" />
+
+  <!-- <multiply> -->
+  <implementation name="IM_multiply_bsdfC_genmsl" nodedef="ND_multiply_bsdfC" target="genmsl" />
+  <implementation name="IM_multiply_bsdfF_genmsl" nodedef="ND_multiply_bsdfF" target="genmsl" />
+  <implementation name="IM_multiply_edfC_genmsl" nodedef="ND_multiply_edfC" target="genmsl" />
+  <implementation name="IM_multiply_edfF_genmsl" nodedef="ND_multiply_edfF" target="genmsl" />
+
+  <!-- <uniform_edf> -->
+  <implementation name="IM_uniform_edf_genmsl" nodedef="ND_uniform_edf" file="../genglsl/mx_uniform_edf.glsl" function="mx_uniform_edf" target="genmsl" />
+
+  <!-- <surface> -->
+  <implementation name="IM_surface_genmsl" nodedef="ND_surface" target="genmsl" />
+
+  <!-- <displacement> -->
+  <implementation name="IM_displacement_float_genmsl" nodedef="ND_displacement_float" file="../genglsl/mx_displacement_float.glsl" function="mx_displacement_float" target="genmsl" />
+  <implementation name="IM_displacement_vector3_genmsl" nodedef="ND_displacement_vector3" file="../genglsl/mx_displacement_vector3.glsl" function="mx_displacement_vector3" target="genmsl" />
+
+  <!-- <light> -->
+  <implementation name="IM_light_genmsl" nodedef="ND_light" target="genmsl" />
+
+  <!-- <roughness_anisotropy> -->
+  <implementation name="IM_roughness_anisotropy_genmsl" nodedef="ND_roughness_anisotropy" file="../genglsl/mx_roughness_anisotropy.glsl" function="mx_roughness_anisotropy" target="genmsl" />
+
+  <!-- <roughness_dual> -->
+  <implementation name="IM_roughness_dual_genmsl" nodedef="ND_roughness_dual" file="../genglsl/mx_roughness_dual.glsl" function="mx_roughness_dual" target="genmsl" />
+
+  <!-- <artistic_ior> -->
+  <implementation name="IM_artistic_ior_genmsl" nodedef="ND_artistic_ior" file="../genglsl/mx_artistic_ior.glsl" function="mx_artistic_ior" target="genmsl" />
+
+</materialx>

libraries/stdlib/genglsl/lib/mx_transform_color.glsl

@@ -13,5 +13,5 @@ vec3 mx_srgb_texture_to_lin_rec709(vec3 color)
     bvec3 isAbove = greaterThan(color, vec3(0.04045));
     vec3 linSeg = color / 12.92;
     vec3 powSeg = pow(max(color + vec3(0.055), vec3(0.0)) / 1.055, vec3(2.4));
-    return mix(linSeg, powSeg, isAbove);
+    return mix(linSeg, powSeg, vec3(isAbove));
 }

libraries/stdlib/genmsl/lib/mx_math.metal

@@ -0,0 +1,107 @@
+#define M_FLOAT_EPS 1e-8
+
+float mx_square(float x)
+{
+    return x*x;
+}
+
+vec2 mx_square(vec2 x)
+{
+    return x*x;
+}
+
+vec3 mx_square(vec3 x)
+{
+    return x*x;
+}
+
+#ifdef __DECL_GL_MATH_FUNCTIONS__
+
+float radians(float degree) { return (degree * M_PI_F / 180.0f); }
+
+float3x3 inverse(float3x3 m)
+{
+    float n11 = m[0][0], n12 = m[1][0], n13 = m[2][0];
+    float n21 = m[0][1], n22 = m[1][1], n23 = m[2][1];
+    float n31 = m[0][2], n32 = m[1][2], n33 = m[2][2];
+
+    float det = determinant(m);
+    float idet = 1.0f / det;
+
+    float3x3 ret;
+
+    ret[0][0] = idet * (n22 * n33 - n32 * n23);
+    ret[1][0] = idet * (n32 * n13 - n12 * n33);
+    ret[2][0] = idet * (n12 * n23 - n22 * n13);
+
+    ret[0][1] = idet * (n31 * n23 - n21 * n33);
+    ret[1][1] = idet * (n11 * n33 - n31 * n13);
+    ret[2][1] = idet * (n21 * n13 - n11 * n23);
+
+    ret[0][2] = idet * (n21 * n32 - n31 * n22);
+    ret[1][2] = idet * (n31 * n12 - n11 * n32);
+    ret[2][2] = idet * (n11 * n22 - n21 * n12);
+
+    return ret;
+}
+
+float4x4 inverse(float4x4 m)
+{
+    float n11 = m[0][0], n12 = m[1][0], n13 = m[2][0], n14 = m[3][0];
+    float n21 = m[0][1], n22 = m[1][1], n23 = m[2][1], n24 = m[3][1];
+    float n31 = m[0][2], n32 = m[1][2], n33 = m[2][2], n34 = m[3][2];
+    float n41 = m[0][3], n42 = m[1][3], n43 = m[2][3], n44 = m[3][3];
+
+    float t11 = n23 * n34 * n42 - n24 * n33 * n42 + n24 * n32 * n43 - n22 * n34 * n43 - n23 * n32 * n44 + n22 * n33 * n44;
+    float t12 = n14 * n33 * n42 - n13 * n34 * n42 - n14 * n32 * n43 + n12 * n34 * n43 + n13 * n32 * n44 - n12 * n33 * n44;
+    float t13 = n13 * n24 * n42 - n14 * n23 * n42 + n14 * n22 * n43 - n12 * n24 * n43 - n13 * n22 * n44 + n12 * n23 * n44;
+    float t14 = n14 * n23 * n32 - n13 * n24 * n32 - n14 * n22 * n33 + n12 * n24 * n33 + n13 * n22 * n34 - n12 * n23 * n34;
+
+    float det = determinant(m);
+    float idet = 1.0f / det;
+
+    float4x4 ret;
+
+    ret[0][0] = t11 * idet;
+    ret[0][1] = (n24 * n33 * n41 - n23 * n34 * n41 - n24 * n31 * n43 + n21 * n34 * n43 + n23 * n31 * n44 - n21 * n33 * n44) * idet;
+    ret[0][2] = (n22 * n34 * n41 - n24 * n32 * n41 + n24 * n31 * n42 - n21 * n34 * n42 - n22 * n31 * n44 + n21 * n32 * n44) * idet;
+    ret[0][3] = (n23 * n32 * n41 - n22 * n33 * n41 - n23 * n31 * n42 + n21 * n33 * n42 + n22 * n31 * n43 - n21 * n32 * n43) * idet;
+
+    ret[1][0] = t12 * idet;
+    ret[1][1] = (n13 * n34 * n41 - n14 * n33 * n41 + n14 * n31 * n43 - n11 * n34 * n43 - n13 * n31 * n44 + n11 * n33 * n44) * idet;
+    ret[1][2] = (n14 * n32 * n41 - n12 * n34 * n41 - n14 * n31 * n42 + n11 * n34 * n42 + n12 * n31 * n44 - n11 * n32 * n44) * idet;
+    ret[1][3] = (n12 * n33 * n41 - n13 * n32 * n41 + n13 * n31 * n42 - n11 * n33 * n42 - n12 * n31 * n43 + n11 * n32 * n43) * idet;
+
+    ret[2][0] = t13 * idet;
+    ret[2][1] = (n14 * n23 * n41 - n13 * n24 * n41 - n14 * n21 * n43 + n11 * n24 * n43 + n13 * n21 * n44 - n11 * n23 * n44) * idet;
+    ret[2][2] = (n12 * n24 * n41 - n14 * n22 * n41 + n14 * n21 * n42 - n11 * n24 * n42 - n12 * n21 * n44 + n11 * n22 * n44) * idet;
+    ret[2][3] = (n13 * n22 * n41 - n12 * n23 * n41 - n13 * n21 * n42 + n11 * n23 * n42 + n12 * n21 * n43 - n11 * n22 * n43) * idet;
+
+    ret[3][0] = t14 * idet;
+    ret[3][1] = (n13 * n24 * n31 - n14 * n23 * n31 + n14 * n21 * n33 - n11 * n24 * n33 - n13 * n21 * n34 + n11 * n23 * n34) * idet;
+    ret[3][2] = (n14 * n22 * n31 - n12 * n24 * n31 - n14 * n21 * n32 + n11 * n24 * n32 + n12 * n21 * n34 - n11 * n22 * n34) * idet;
+    ret[3][3] = (n12 * n23 * n31 - n13 * n22 * n31 + n13 * n21 * n32 - n11 * n23 * n32 - n12 * n21 * n33 + n11 * n22 * n33) * idet;
+
+    return ret;
+}
+
+template<class T1, class T2>
+T1 mod(T1 x, T2 y)
+{
+    return x - y * floor(x/y);
+}
+
+template <typename T>
+T atan(T y_over_x) { return ::atan(y_over_x); }
+
+template <typename T>
+T atan(T y, T x) { return ::atan2(y, x); }
+
+#define lessThan(a, b) ((a) < (b))
+#define lessThanEqual(a, b) ((a) <= (b))
+#define greaterThan(a, b) ((a) > (b))
+#define greaterThanEqual(a, b) ((a) >= (b))
+#define equal(a, b) ((a) == (b))
+#define notEqual(a, b) ((a) != (b))
+
+#endif