Cleaned and Formatted Shaders.

README.md

@@ -4,39 +4,50 @@ A Vulkan based renderer created with Vulkan 1.3 in C++.
 
 ## Features (Current and Planned)
 
-- [ ] Forward Rendering
 - [ ] glTF 2.0 Support
-    - [ ] Load Vertex Data
+    - [X] Load Vertex Data
-    - [ ] Load Material Data
+    - [X] Load Material Data
     - [ ] Load Animation Data
     - [ ] Load Camera
     - [ ] Load Lights
     - [ ] Support Specular Materials
-- [ ] Bindless Descriptors
+- [X] Bindless Descriptors
-- [ ] PBR
+- [X] Simplified Descriptor Creation Pipeline
-- [ ] IBL
+- [ ] Debugging
+    - [ ] Tracy Integration
+    - [ ] Dear ImGui Integration
+- [ ] Transparency
+    - [ ] Sorted
+    - [ ] Order Independent
 - [ ] Shadows v1
     - [ ] Omnidirectional Cubemap Shadows
     - [ ] Spot Lights
     - [ ] Directional Shadows
     - [ ] Cascaded Shadows
     - [ ] PCF
-- [ ] Simplified Descriptor Creation Pipeline
+- [ ] Lighting / Shading Pipelines
-- [ ] Deferred Rendering
+    - [ ] Blinn-Phong
+    - [ ] PBR
+    - [ ] IBL
+- [ ] Rendering Techniques
+    - [ ] Forward Rendering
+    - [ ] Deferred Rendering
+    - [ ] Clustered-Forward Rendering
 - [ ] Ambient Occlusion
     - [ ] SSAO
     - [ ] HBAO
-    - [ ] VXAO/SDFAO
+    - [ ] VXAO
+    - [ ] SDFAO
     - [ ] RTX AO
 - [ ] Reflection
     - [ ] ScreenSpace Reflection (SSR)
     - [ ] Cubemap/Probe Reflection
-- [ ] Forward+ Rendering
+    - [ ] Ray-Traced Reflection
-    - [ ] Global Illumination
+- [ ] Global Illumination
-        - [ ] Precomputed Radiance Transfer
+    - [ ] Precomputed Radiance Transfer
-        - [ ] Voxel Cone Tracing
+    - [ ] Voxel Cone Tracing
-        - [ ] SDFGI
+    - [ ] SDFGI
-        - [ ] RTXGI
+    - [ ] RTXGI
 - [ ] Shadows v2
     - [ ] Omnidirectional Dual Paraboloid Shadows
     - [ ] Perspective Shadow Mapping

samples/00_util/CMakeLists.txt

@@ -6,7 +6,9 @@ add_library(util_helper STATIC
         helpers.h
         helpers.cpp
         frame.cpp
-        frame.h)
+        frame.h
+        gpu_resource_manager.cpp
+        gpu_resource_manager.h)
 
 target_link_libraries(util_helper PRIVATE aster_core)
 target_include_directories(util_helper PUBLIC ${CMAKE_CURRENT_SOURCE_DIR})

samples/00_util/gpu_resource_manager.cpp

@@ -1,9 +1,9 @@
 // =============================================
-//  Aster: pipeline_utils.h
+//  Aster: gpu_resource_manager.cpp
 //  Copyright (c) 2020-2024 Anish Bhobe
 // =============================================
 
-#include "render_resource_manager.h"
+#include "gpu_resource_manager.h"
 
 #include "buffer.h"
 #include "device.h"

samples/00_util/gpu_resource_manager.h

@@ -1,20 +1,18 @@
 // =============================================
-//  Aster: pipeline_utils.h
+//  Aster: gpu_resource_manager.h
 //  Copyright (c) 2020-2024 Anish Bhobe
 // =============================================
 
 #pragma once
 
 #include "global.h"
-#include "buffer.h"
-#include "image.h"
 
 #include <EASTL/deque.h>
 #include <EASTL/vector_map.h>
 
 struct Device;
 struct Texture;
-struct UniformStorageBuffer;
+struct StorageBuffer;
 
 struct GpuResourceHandle
 {

samples/03_model_render/CMakeLists.txt

@@ -8,11 +8,13 @@ find_path(TINYGLTF_INCLUDE_DIRS "tiny_gltf.h")
 add_executable(model_render model_render.cpp
         pipeline_utils.cpp
         pipeline_utils.h
-        render_resource_manager.cpp
-        render_resource_manager.h
         model_loader.cpp
         model_loader.h
-)
+        light_manager.cpp 
+        light_manager.h
+        nodes.cpp
+        nodes.h)
+
 add_shader(model_render shader/model.vs.hlsl)
 add_shader(model_render shader/model.ps.hlsl)
 

samples/03_model_render/light_manager.cpp

@@ -0,0 +1,277 @@
+// =============================================
+//  Aster: light_manager.cpp
+//  Copyright (c) 2020-2024 Anish Bhobe
+// =============================================
+
+#include "light_manager.h"
+
+#include "buffer.h"
+#include "glm/ext/matrix_transform.hpp"
+
+struct Light
+{
+    union {
+        vec3 um_Position;
+        vec3 um_Direction;
+    };
+    f32 m_Range;    // < 0.0 for invalid
+    u32 m_Color_;   // LSB is used for flags. (R G B Flags)
+    f32 m_Intensity;
+
+    constexpr static u32 MAX_GEN = 0x40;
+    constexpr static u32 GEN_MASK = MAX_GEN - 1;
+
+    constexpr static u32 TYPE_MASK = 0xC0;
+    constexpr static u32 TYPE_INVALID = 0x0;
+    constexpr static u32 TYPE_DIRECTIONAL = 1 << 6;
+    constexpr static u32 TYPE_POINT = 2 << 6;
+    constexpr static u32 TYPE_SPOT = 3 << 6; // Currently Unused
+
+    constexpr static u32 COLOR_MASK = ~(GEN_MASK | TYPE_MASK);
+};
+
+// Static Checks
+
+// Ensure layouts are exact.
+static_assert(offsetof(DirectionalLight, m_Direction) == offsetof(Light, um_Direction));
+static_assert(offsetof(DirectionalLight, m_Color_) == offsetof(Light, m_Color_));
+static_assert(offsetof(DirectionalLight, m_Intensity) == offsetof(Light, m_Intensity));
+static_assert(sizeof(DirectionalLight) <= sizeof(Light));
+
+// Ensure layouts are exact.
+static_assert(offsetof(PointLight, m_Position) == offsetof(Light, um_Position));
+static_assert(offsetof(PointLight, m_Range) == offsetof(Light, m_Range));
+static_assert(offsetof(PointLight, m_Color_) == offsetof(Light, m_Color_));
+static_assert(offsetof(PointLight, m_Intensity) == offsetof(Light, m_Intensity));
+static_assert(sizeof(PointLight) <= sizeof(Light));
+
+// Ensure bitmask are in the right place.
+static_assert((Light::TYPE_MASK & Light::TYPE_INVALID) == Light::TYPE_INVALID);
+static_assert((Light::TYPE_MASK & Light::TYPE_DIRECTIONAL) == Light::TYPE_DIRECTIONAL);
+static_assert((Light::TYPE_MASK & Light::TYPE_POINT) == Light::TYPE_POINT);
+static_assert((Light::TYPE_MASK & Light::TYPE_SPOT) == Light::TYPE_SPOT);
+static_assert(Light::COLOR_MASK == 0xFFFFFF00);
+
+inline u32
+ToColor32(const vec4 &col)
+{
+    const u32 r = Cast<u32>(eastl::min(col.r, 1.0f) * 255.99f);
+    const u32 g = Cast<u32>(eastl::min(col.g, 1.0f) * 255.99f);
+    const u32 b = Cast<u32>(eastl::min(col.b, 1.0f) * 255.99f);
+    const u32 a = Cast<u32>(eastl::min(col.a, 1.0f) * 255.99f);
+
+    return r << 24 | g << 16 | b << 8 | a;
+}
+
+inline u32
+ToColor32(const vec3 &col)
+{
+    const u32 r = Cast<u32>(eastl::min(col.r, 1.0f) * 255.99f);
+    const u32 g = Cast<u32>(eastl::min(col.g, 1.0f) * 255.99f);
+    const u32 b = Cast<u32>(eastl::min(col.b, 1.0f) * 255.99f);
+    constexpr u32 a = 255;
+
+    return r << 24 | g << 16 | b << 8 | a;
+}
+
+LightManager::LightManager(GpuResourceManager *resourceManager)
+    : m_ResourceManager{resourceManager}
+    , m_DirectionalLightCount{}
+    , m_PointLightCount{}
+    , m_MetaInfo{}
+    , m_GpuBufferCapacity_{0}
+{
+}
+
+LightManager::~LightManager()
+{
+    m_ResourceManager->Release(m_MetaInfo.m_LightBuffer);
+}
+
+LightManager::LightManager(LightManager &&other) noexcept
+    : m_ResourceManager(other.m_ResourceManager)
+    , m_Lights(std::move(other.m_Lights))
+    , m_DirectionalLightCount(other.m_DirectionalLightCount)
+    , m_PointLightCount(other.m_PointLightCount)
+    , m_MetaInfo(other.m_MetaInfo)
+    , m_GpuBufferCapacity_(other.m_GpuBufferCapacity_)
+{
+    other.m_MetaInfo.m_LightBuffer = {};
+}
+
+LightManager &
+LightManager::operator=(LightManager &&other) noexcept
+{
+    if (this == &other)
+        return *this;
+    m_ResourceManager = other.m_ResourceManager;
+    m_Lights = std::move(other.m_Lights);
+    m_DirectionalLightCount = other.m_DirectionalLightCount;
+    m_PointLightCount = other.m_PointLightCount;
+    m_MetaInfo = other.m_MetaInfo;
+    other.m_MetaInfo.m_LightBuffer = {};
+    m_GpuBufferCapacity_ = other.m_GpuBufferCapacity_;
+    return *this;
+}
+
+LightHandle
+LightManager::AddDirectional(const vec3 &direction, const vec3 &color)
+{
+    const vec3 normDirection = normalize(direction);
+    if (m_DirectionalLightCount < m_MetaInfo.m_DirectionalLightMaxCount)
+    {
+        u16 index = 0;
+        for (auto &light : m_Lights)
+        {
+            if (light.m_Range < 0)
+            {
+                const u8 gen = light.m_Color_ & Light::GEN_MASK;
+
+                light.m_Color_ = (ToColor32(color) & Light::COLOR_MASK) | Light::TYPE_DIRECTIONAL | gen;
+                light.m_Range = 1.0f;
+                light.um_Direction = normDirection;
+
+                m_GpuBufferCapacity_ |= UPDATE_REQUIRED_BIT;
+
+                return {Light::TYPE_DIRECTIONAL, gen, index};
+            }
+            ++index;
+            assert(index < m_MetaInfo.m_DirectionalLightMaxCount); // Gap not found. But must exist
+        }
+    }
+
+    // In case we will end up intersecting, we move point lights away. (2 at a time)
+    if (m_DirectionalLightCount == m_MetaInfo.m_DirectionalLightMaxCount &&
+        m_MetaInfo.m_DirectionalLightMaxCount == m_MetaInfo.m_PointLightOffset)
+    {
+        const u16 oldPointLightOffset = m_MetaInfo.m_PointLightOffset;
+        const u32 pointLightMaxCount = m_MetaInfo.m_PointLightMaxCount;
+        // Might cause a capacity increase, but I want to use that for my gpu buffer resize.
+        m_Lights.push_back();
+        m_Lights.push_back();
+
+        if (m_MetaInfo.m_PointLightMaxCount > 0)    // Edge Case: nullptr at size 0
+        {
+            Light *oldPointStart = m_Lights.data() + oldPointLightOffset;
+            Light *newPointStart = oldPointStart + 2;
+
+            static_assert(std::is_trivially_copyable_v<Light>);
+            memcpy(newPointStart, oldPointStart, pointLightMaxCount * sizeof *newPointStart);
+        }
+
+        m_MetaInfo.m_PointLightOffset += 2;
+    }
+
+    constexpr u8 gen = 0; // New light
+    m_Lights[m_DirectionalLightCount].m_Color_ = (ToColor32(color) & Light::COLOR_MASK) | Light::TYPE_DIRECTIONAL | gen;
+    m_Lights[m_DirectionalLightCount].m_Range = 1.0f;
+    m_Lights[m_DirectionalLightCount].um_Direction = normDirection;
+    const u16 index = m_DirectionalLightCount;
+
+    ++m_DirectionalLightCount;
+    ++m_MetaInfo.m_DirectionalLightMaxCount;
+
+    return {Light::TYPE_DIRECTIONAL, gen, index};
+}
+
+LightHandle
+LightManager::AddPoint(const vec3 & position, const vec3 &color, const f32 radius)
+{
+    assert(m_PointLightCount <= m_MetaInfo.m_PointLightMaxCount);
+    assert(radius >= 0.0f);
+    if (m_PointLightCount < m_MetaInfo.m_PointLightMaxCount)
+    {
+        Light *light = m_Lights.data() + m_MetaInfo.m_PointLightOffset;
+        for (u32 index = 0; index < m_MetaInfo.m_PointLightMaxCount; ++index)
+        {
+            if (light->m_Range < 0)
+            {
+                const u8 gen = light->m_Color_ & Light::GEN_MASK;
+
+                light->m_Color_ = (ToColor32(color) & Light::COLOR_MASK) | Light::TYPE_POINT | gen;
+                light->m_Range = radius;
+                light->um_Position = position;
+
+                m_GpuBufferCapacity_ |= UPDATE_REQUIRED_BIT;
+
+                return {Light::TYPE_POINT, gen, Cast<u16>(index)};
+            }
+            ++light;
+        }
+        assert(false); // gap must exists.
+        return {};
+    }
+
+    m_Lights.push_back();
+    const u16 index = m_PointLightCount;
+
+    Light *light = &m_Lights[index + m_MetaInfo.m_PointLightOffset];
+    constexpr u8 gen = 0;   // New light
+
+    light->m_Color_ = (ToColor32(color) & Light::COLOR_MASK) | Light::TYPE_POINT | gen;
+    light->m_Range = radius;
+    light->um_Position = position;
+
+    ++m_PointLightCount;
+    ++m_MetaInfo.m_PointLightMaxCount;
+
+    m_GpuBufferCapacity_ |= UPDATE_REQUIRED_BIT;
+
+    return {Light::TYPE_POINT, gen, index};
+}
+
+void
+LightManager::Update()
+{
+    const u16 requiredBufferCapacity = eastl::min(Cast<u16>(m_Lights.capacity()), MAX_LIGHTS);
+    if ((m_GpuBufferCapacity_ & CAPACITY_MASK) < requiredBufferCapacity)
+    {
+        StorageBuffer newBuffer;
+        newBuffer.Init(m_ResourceManager->m_Device, requiredBufferCapacity * sizeof m_Lights[0], true, "Light Buffer");
+        m_GpuBufferCapacity_ = requiredBufferCapacity | UPDATE_REQUIRED_BIT;
+
+        m_ResourceManager->Release(m_MetaInfo.m_LightBuffer);
+        m_MetaInfo.m_LightBuffer = m_ResourceManager->Commit(&newBuffer);
+    }
+    if (m_GpuBufferCapacity_ & UPDATE_REQUIRED_BIT)
+    {
+        m_ResourceManager->Write(m_MetaInfo.m_LightBuffer, 0, m_Lights.size() * sizeof m_Lights[0], m_Lights.data());
+    }
+}
+
+void
+LightManager::RemoveLight(const LightHandle handle)
+{
+    const u8 handleGen = handle.m_Generation;
+
+
+    if (handle.m_Type == Light::TYPE_DIRECTIONAL)
+    {
+        Light *lightSlot = &m_Lights[handle.m_Index];
+        const u8 slotGen = lightSlot->m_Color_ & Light::GEN_MASK;
+        if (slotGen > handleGen)
+        {
+            WARN("Invalid handle gen: {} being freed. (slot gen: {})", handleGen, slotGen);
+            return;
+        }
+
+        lightSlot->m_Range = -1.0f;
+        lightSlot->m_Color_ = Light::TYPE_INVALID | (slotGen + 1) % Light::MAX_GEN;
+        --m_DirectionalLightCount;
+    }
+
+    if (handle.m_Type == Light::TYPE_POINT)
+    {
+        Light *lightSlot = &m_Lights[handle.m_Index + m_MetaInfo.m_PointLightOffset];
+        const u8 slotGen = lightSlot->m_Color_ & Light::GEN_MASK;
+        if (slotGen > handleGen)
+        {
+            WARN("Invalid handle gen: {} being freed. (slot gen: {})", handleGen, slotGen);
+            return;
+        }
+
+        lightSlot->m_Range = -1.0f;
+        lightSlot->m_Color_ = Light::TYPE_INVALID | (slotGen + 1) % Light::MAX_GEN;
+        --m_PointLightCount;
+    }
+}

samples/03_model_render/light_manager.h

@@ -0,0 +1,83 @@
+// =============================================
+//  Aster: light_manager.h
+//  Copyright (c) 2020-2024 Anish Bhobe
+// =============================================
+
+#pragma once
+
+#include "global.h"
+
+// TODO: Separate files so you only import handles.
+#include "gpu_resource_manager.h"
+
+struct DirectionalLight
+{
+    vec3 m_Direction;
+    u32 m_UnusedPadding0_;
+    u32 m_Color_; // LSB is used for flags. (R G B Flags)
+    f32 m_Intensity;
+};
+
+struct PointLight
+{
+    vec3 m_Position;
+    f32 m_Range;
+    u32 m_Color_; // LSB is used for flags. (R G B Flags)
+    f32 m_Intensity;
+};
+
+struct LightHandle
+{
+    u8 m_Type;
+    u8 m_Generation;
+    u16 m_Index;
+};
+
+struct Light;
+
+struct LightManager
+{
+    constexpr static u16 MAX_LIGHTS = MaxValue<u16>;
+    struct LightMetaInfo
+    {
+        // The number of directional lights is relatively low (1 - 2) and will almost never change in a scene.
+        // We can use that with Offset = 0, and point light at further offsets.
+        // This way we don't need to move point lights often.
+
+        BufferHandle m_LightBuffer;         // 04   04
+        u16 m_PointLightMaxCount;           // 02   06
+        u16 m_PointLightOffset;             // 02   08
+        u16 m_DirectionalLightMaxCount;     // 02   10
+        u16 m_UnusedPadding0 = 0;           // 02   12
+    };
+
+    GpuResourceManager *m_ResourceManager;
+    eastl::vector<Light> m_Lights;
+
+    // We don't need a Directional Light free list. We will just brute force iterate.
+    u16 m_DirectionalLightCount;
+    // TODO: A point light free list. We will brute force until we have a lot (100+) of point lights.
+    u16 m_PointLightCount;
+
+    LightMetaInfo m_MetaInfo;
+    // Using lower bit for flags. Use CAPACITY_MASK for value.
+    u16 m_GpuBufferCapacity_;
+
+    // Using lower bit. Capacity can be directly a multiple of 2
+    // Thus, range is up to MaxValue<u16>
+    constexpr static u16 UPDATE_REQUIRED_BIT = 1;
+    constexpr static u16 CAPACITY_MASK = ~(UPDATE_REQUIRED_BIT);
+
+    LightHandle AddDirectional(const vec3 &direction, const vec3 &color);
+    LightHandle AddPoint(const vec3& position, const vec3 &color, f32 radius);
+    void Update();
+    void RemoveLight(LightHandle handle);
+
+    explicit LightManager(GpuResourceManager *resourceManager);
+    ~LightManager();
+
+    LightManager(LightManager &&other) noexcept;
+    LightManager &operator=(LightManager &&other) noexcept;
+
+    DISALLOW_COPY_AND_ASSIGN(LightManager);
+};

samples/03_model_render/model_loader.cpp

@@ -16,7 +16,7 @@
 #include "device.h"
 #include "helpers.h"
 #include "image.h"
-#include "render_resource_manager.h"
+#include "gpu_resource_manager.h"
 
 #include <glm/gtc/type_ptr.hpp>
 #include <EASTL/array.h>
@@ -349,10 +349,9 @@ ModelLoader::LoadModel(cstr path, cstr name, bool batched)
                 }
             }
 #pragma endregion
-
+#pragma region Vertex Data
             vertexData.resize(vertexPositions.size());
 
-            #pragma region Normal
             // Normal Coords
             if (prim.attributes.contains(ANormal))
             {
@@ -395,8 +394,7 @@ ModelLoader::LoadModel(cstr path, cstr name, bool batched)
                     }
                 }
             }
-#pragma endregion
+
-#pragma region UV0
             // UV0
             if (prim.attributes.contains(ATexCoord0))
             {
@@ -421,8 +419,7 @@ ModelLoader::LoadModel(cstr path, cstr name, bool batched)
                     }
                 }
             }
-#pragma endregion
+
-#pragma region Color
             if (prim.attributes.contains(AColor0))
             {
                 tinygltf::Accessor *colorAccessor = &model.accessors[prim.attributes[AColor0]];
@@ -673,115 +670,6 @@ Model::operator=(Model &&other) noexcept
     return *this;
 }
 
-u32
-Nodes::Add(const mat4 &transform, const i32 parent)
-{
-    m_Dirty = true;
-    const u32 index = Count();
-    m_Transforms.push_back(transform);
-    m_GlobalTransforms.push_back(transform);
-    const u32 parentVal = (parent < 0 ? ROOT_BIT : parent & PARENT_MASK) | DIRTY_BIT;
-    m_Parents_.push_back(parentVal);
-
-    return index;
-}
-
-const mat4 &
-Nodes::Get(const u32 index) const
-{
-    return m_Transforms[index];
-}
-
-void
-Nodes::Set(const u32 index, const mat4 &transform)
-{
-    m_Dirty = true;
-    m_Transforms[index] = transform;
-    m_Parents_[index] |= DIRTY_BIT;
-}
-
-const mat4 &
-Nodes::operator[](const u32 index) const
-{
-    return m_Transforms[index]; 
-}
-
-mat4 &
-Nodes::operator[](const u32 index)
-{
-    m_Dirty = true;
-    m_Parents_[index] |= DIRTY_BIT;
-    return m_Transforms[index];
-}
-
-u32
-Nodes::Count() const
-{
-    return Cast<u32>(m_Transforms.size());
-}
-
-usize
-Nodes::GetGlobalTransformByteSize() const
-{
-    return m_GlobalTransforms.size() * sizeof m_GlobalTransforms[0];
-}
-
-const mat4 *
-Nodes::GetGlobalTransformPtr() const
-{
-    return m_GlobalTransforms.data();
-}
-
-bool
-Nodes::Update()
-{
-    if (!m_Dirty)
-        return false;
-
-    auto transformIter = m_Transforms.begin();
-    auto globalTransformIter = m_GlobalTransforms.begin();
-    auto parentIter = m_Parents_.begin();
-    const auto parentEnd = m_Parents_.end();
-
-    while (parentIter != parentEnd)
-    {
-        const bool isRoot = *parentIter & ROOT_BIT;
-        const bool isDirty = *parentIter & DIRTY_BIT;
-
-        if (isRoot)
-        {
-            if (isDirty)
-            {
-                // Copy-update if the root is dirty.
-                *globalTransformIter = *transformIter;
-            }
-        }
-        else
-        {
-            const u32 parentIdx = *parentIter & PARENT_MASK;
-            const bool isParentDirty = m_Parents_[parentIdx] & DIRTY_BIT;
-            if (isDirty || isParentDirty)
-            {
-                // Update w.r.t parent if either local or parent transforms updated.
-                *globalTransformIter = m_GlobalTransforms[parentIdx] * *transformIter;
-                m_Parents_[parentIdx] |= DIRTY_BIT;       // Set dirty to propagate the update.
-            }
-        }
-
-        ++parentIter;
-        ++globalTransformIter;
-        ++transformIter;
-    }
-
-    for (u32 &parentValue : m_Parents_)
-    {
-        parentValue &= ~DIRTY_BIT;  // Unset dirty.
-    }
-
-    m_Dirty = false;
-    return true;
-}
-
 const mat4 &
 Model::GetModelTransform() const
 {

samples/03_model_render/model_loader.h

@@ -8,7 +8,8 @@
 #include "buffer.h"
 #include "global.h"
 
-#include "render_resource_manager.h"
+#include "gpu_resource_manager.h"
+#include "nodes.h"
 
 #include <tiny_gltf.h>
 
@@ -27,32 +28,6 @@ struct MeshPrimitive
     i32 m_TransformIdx;
 };
 
-struct Nodes
-{
-    eastl::vector<mat4> m_Transforms;
-    eastl::vector<mat4> m_GlobalTransforms;
-    /// Parents are also used for bookkeeping
-    eastl::vector<u32> m_Parents_;
-    bool m_Dirty = true;
-
-    constexpr static u32 ROOT_BIT = 1u << 31;
-    constexpr static u32 DIRTY_BIT = 1u << 30;
-    constexpr static u32 PARENT_MASK = ~(ROOT_BIT | DIRTY_BIT);
-
-    u32 Add(const mat4 &transform, const i32 parent = -1);
-    [[nodiscard]] const mat4 &Get(const u32 index) const;
-    void Set(const u32 index, const mat4 &transform);
-    [[nodiscard]] u32 Count() const;
-
-    [[nodiscard]] const mat4 &operator[](const u32 index) const;
-    [[nodiscard]] mat4 &operator[](const u32 index);
-
-    [[nodiscard]] usize GetGlobalTransformByteSize() const;
-    [[nodiscard]] const mat4 *GetGlobalTransformPtr() const;
-
-    bool Update();
-};
-
 struct Material
 {
     vec4 m_AlbedoFactor;            // 16   16

samples/03_model_render/model_render.cpp

@@ -16,10 +16,11 @@
 
 #include "frame.h"
 #include "helpers.h"
+#include "light_manager.h"
 
 #include "model_loader.h"
 #include "pipeline_utils.h"
-#include "render_resource_manager.h"
+#include "gpu_resource_manager.h"
 
 #include <EASTL/array.h>
 #include <stb_image.h>
@@ -27,27 +28,13 @@
 #include <filesystem>
 
 constexpr u32 MAX_FRAMES_IN_FLIGHT = 3;
-constexpr auto MODEL_FILE = "model/OrientationTest.glb";
+constexpr auto MODEL_FILE = "model/DamagedHelmet.glb";
-
-struct ImageFile
-{
-    u8 *m_Data = nullptr;
-    u32 m_Width = 0;
-    u32 m_Height = 0;
-    u32 m_NumChannels = 0;
-    bool m_Constant = false;
-
-    bool Load(cstr fileName);
-    bool Load(vec4 color);
-    [[nodiscard]] usize GetSize() const;
-    ~ImageFile();
-};
 
 struct Camera
 {
-    mat4 m_Model;
     mat4 m_View;
     mat4 m_Perspective;
+    vec4 m_Position;
 };
 
 int
@@ -88,18 +75,25 @@ main(int, char **)
     GpuResourceManager resourceManager = {&device, 1000};
 
     ModelLoader modelLoader = {&resourceManager, commandQueue, queueAllocation.m_Family, queueAllocation.m_Family};
+    LightManager lightManager = LightManager{&resourceManager};
 
-    auto model = modelLoader.LoadModel(MODEL_FILE);
+    Model model = modelLoader.LoadModel(MODEL_FILE);
 
     Pipeline pipeline = CreatePipeline(&device, &swapchain, &resourceManager);
 
     Camera camera = {
-        .m_Model = {1.0f},
+        .m_View = glm::lookAt(vec3(0.0f, 2.0f, 2.0f), vec3(0.0f), vec3(0.0f, 1.0f, 0.0f)),
-        .m_View = glm::lookAt(vec3(0.0f, 12.0f, 12.0f), vec3(0.0f), vec3(0.0f, 1.0f, 0.0f)),
         .m_Perspective = glm::perspective(
             70_deg, Cast<f32>(swapchain.m_Extent.width) / Cast<f32>(swapchain.m_Extent.height), 0.1f, 100.0f),
+        .m_Position = vec4{0.0f, 2.0f, 2.0f, 1.0f},
     };
 
+    lightManager.AddDirectional(vec3(0.0f, -1.0f, 0.0f), {0.0f, 1.0f, 0.0f});
+    lightManager.AddPoint(vec3{2.0f, 1.0f, 0.0f}, {1.0f, 0.0f, 0.0f}, 15.0f);
+    lightManager.AddPoint(vec3{-2.0f, 1.0f, 0.0f}, {0.0f, 0.0f, 1.0f}, 15.0f);
+
+    lightManager.Update();
+
     vk::DescriptorPool descriptorPool;
     vk::DescriptorSet descriptorSet;
 
@@ -280,16 +274,23 @@ main(int, char **)
         cmd.bindDescriptorSets(vk::PipelineBindPoint::eGraphics, pipeline.m_Layout, 1, 1, &descriptorSet, 0, nullptr);
 
         cmd.bindIndexBuffer(model.m_IndexBuffer.m_Buffer, 0, vk::IndexType::eUint32);
-        cmd.pushConstants(pipeline.m_Layout, vk::ShaderStageFlagBits::eAll, 0, sizeof model.m_Handles,
+
+        u32 pcbOffset = 0;
+        cmd.pushConstants(pipeline.m_Layout, vk::ShaderStageFlagBits::eAll, pcbOffset, sizeof model.m_Handles,
                           &model.m_Handles);
+        pcbOffset += sizeof model.m_Handles;
+        cmd.pushConstants(pipeline.m_Layout, vk::ShaderStageFlagBits::eAll, pcbOffset, sizeof lightManager.m_MetaInfo,
+                          &lightManager.m_MetaInfo);
+        pcbOffset += sizeof lightManager.m_MetaInfo;
 
         for (auto &prim : model.m_MeshPrimitives)
         {
-            cmd.pushConstants(pipeline.m_Layout, vk::ShaderStageFlagBits::eAll, sizeof model.m_Handles,
+            cmd.pushConstants(pipeline.m_Layout, vk::ShaderStageFlagBits::eAll, pcbOffset,
                               sizeof prim.m_MaterialIdx, &prim.m_MaterialIdx);
-            cmd.pushConstants(pipeline.m_Layout, vk::ShaderStageFlagBits::eAll,
+            pcbOffset += sizeof prim.m_MaterialIdx;
-                              sizeof model.m_Handles + sizeof prim.m_MaterialIdx,
+            cmd.pushConstants(pipeline.m_Layout, vk::ShaderStageFlagBits::eAll, pcbOffset,
                               sizeof prim.m_TransformIdx, &prim.m_TransformIdx);
+            pcbOffset += sizeof prim.m_TransformIdx;
             cmd.drawIndexed(prim.m_IndexCount, 1, prim.m_FirstIndex, Cast<i32>(prim.m_VertexOffset), 0);
         }
 
@@ -326,65 +327,3 @@ main(int, char **)
 
     return 0;
 }
-
-bool
-ImageFile::Load(vec4 color)
-{
-    constexpr usize size = 512llu * 512llu * 4llu;
-    u8 *pData = new u8[size];
-
-    const vec4 color255 = 255.999f * color;
-    const glm::vec<4, u8> color8 = color255;
-
-    for (usize i = 0; i < size; i += 4)
-    {
-        memcpy(pData + i, &color8, sizeof color8);
-    }
-
-    m_Data = pData;
-    m_Constant = true;
-    m_Height = 512;
-    m_Width = 512;
-    m_NumChannels = 4;
-
-    return true;
-}
-
-bool
-ImageFile::Load(cstr fileName)
-{
-    int width, height, nrChannels;
-    m_Data = stbi_load(fileName, &width, &height, &nrChannels, 4);
-    ERROR_IF(!m_Data, "Could not load {}", fileName);
-
-    if (!m_Data)
-    {
-        return false;
-    }
-
-    m_Width = width;
-    m_Height = height;
-    m_NumChannels = 4;
-    m_Constant = false;
-
-    return true;
-}
-
-usize
-ImageFile::GetSize() const
-{
-    return Cast<usize>(m_Width) * m_Height * m_NumChannels;
-}
-
-ImageFile::~ImageFile()
-{
-    if (m_Constant)
-    {
-        delete[] Cast<u8 *>(m_Data);
-    }
-    else
-    {
-        stbi_image_free(m_Data);
-    }
-    m_Data = nullptr;
-}

samples/03_model_render/nodes.cpp

@@ -0,0 +1,115 @@
+// =============================================
+//  Aster: nodes.cpp
+//  Copyright (c) 2020-2024 Anish Bhobe
+// =============================================
+
+#include "nodes.h"
+
+u32
+Nodes::Add(const mat4 &transform, const i32 parent)
+{
+    m_Dirty = true;
+    const u32 index = Count();
+    m_Transforms.push_back(transform);
+    m_GlobalTransforms.emplace_back(transform);
+    const u32 parentVal = (parent < 0 ? ROOT_BIT : parent & PARENT_MASK) | DIRTY_BIT;
+    m_Parents_.push_back(parentVal);
+
+    return index;
+}
+
+const mat4 &
+Nodes::Get(const u32 index) const
+{
+    return m_Transforms[index];
+}
+
+void
+Nodes::Set(const u32 index, const mat4 &transform)
+{
+    m_Dirty = true;
+    m_Transforms[index] = transform;
+    m_Parents_[index] |= DIRTY_BIT;
+}
+
+const mat4 &
+Nodes::operator[](const u32 index) const
+{
+    return m_Transforms[index];
+}
+
+mat4 &
+Nodes::operator[](const u32 index)
+{
+    m_Dirty = true;
+    m_Parents_[index] |= DIRTY_BIT;
+    return m_Transforms[index];
+}
+
+u32
+Nodes::Count() const
+{
+    return Cast<u32>(m_Transforms.size());
+}
+
+usize
+Nodes::GetGlobalTransformByteSize() const
+{
+    return m_GlobalTransforms.size() * sizeof m_GlobalTransforms[0];
+}
+
+const Nodes::Transform *
+Nodes::GetGlobalTransformPtr() const
+{
+    return m_GlobalTransforms.data();
+}
+
+bool
+Nodes::Update()
+{
+    if (!m_Dirty)
+        return false;
+
+    auto transformIter = m_Transforms.begin();
+    auto globalTransformIter = m_GlobalTransforms.begin();
+    auto parentIter = m_Parents_.begin();
+    const auto parentEnd = m_Parents_.end();
+
+    while (parentIter != parentEnd)
+    {
+        const bool isRoot = *parentIter & ROOT_BIT;
+        const bool isDirty = *parentIter & DIRTY_BIT;
+
+        if (isRoot)
+        {
+            if (isDirty)
+            {
+                // Copy-update if the root is dirty.
+                *globalTransformIter = *transformIter;
+            }
+        }
+        else
+        {
+            const u32 parentIdx = *parentIter & PARENT_MASK;
+            const bool isParentDirty = m_Parents_[parentIdx] & DIRTY_BIT;
+            if (isDirty || isParentDirty)
+            {
+                // Update w.r.t parent if either local or parent transforms updated.
+                *globalTransformIter = m_GlobalTransforms[parentIdx].m_GlobalTransforms * *transformIter;
+                m_Parents_[parentIdx] |= DIRTY_BIT; // Set dirty to propagate the update.
+            }
+        }
+
+        ++parentIter;
+        ++globalTransformIter;
+        ++transformIter;
+    }
+
+    for (u32 &parentValue : m_Parents_)
+    {
+        parentValue &= ~DIRTY_BIT; // Unset dirty.
+    }
+
+    m_Dirty = false;
+    return true;
+}

samples/03_model_render/nodes.h

@@ -0,0 +1,56 @@
+// =============================================
+//  Aster: nodes.h
+//  Copyright (c) 2020-2024 Anish Bhobe
+// =============================================
+
+#pragma once
+
+#include "global.h"
+
+#include "EASTL/vector.h"
+
+struct Nodes
+{
+    struct Transform
+    {
+        mat4 m_GlobalTransforms;
+        mat4 m_NormalTransforms;
+
+        explicit Transform(const mat4 &transform)
+            : m_GlobalTransforms(transform)
+            , m_NormalTransforms(transpose(inverse(mat3{transform})))
+        {
+        }
+
+        Transform &
+        operator=(const mat4 &transform)
+        {
+            m_GlobalTransforms = transform;
+            m_NormalTransforms = transpose(inverse(mat3{transform}));
+            return *this;
+        }
+    };
+
+    eastl::vector<mat4> m_Transforms;
+    eastl::vector<Transform> m_GlobalTransforms;
+    /// Parents are also used for bookkeeping
+    eastl::vector<u32> m_Parents_;
+    bool m_Dirty = true;
+
+    constexpr static u32 ROOT_BIT = 1u << 31;
+    constexpr static u32 DIRTY_BIT = 1u << 30;
+    constexpr static u32 PARENT_MASK = ~(ROOT_BIT | DIRTY_BIT);
+
+    u32 Add(const mat4 &transform, const i32 parent = -1);
+    [[nodiscard]] const mat4 &Get(const u32 index) const;
+    void Set(const u32 index, const mat4 &transform);
+    [[nodiscard]] u32 Count() const;
+
+    [[nodiscard]] const mat4 &operator[](const u32 index) const;
+    [[nodiscard]] mat4 &operator[](const u32 index);
+
+    [[nodiscard]] usize GetGlobalTransformByteSize() const;
+    [[nodiscard]] const Transform *GetGlobalTransformPtr() const;
+
+    bool Update();
+};

samples/03_model_render/pipeline_utils.cpp

@@ -7,7 +7,7 @@
 
 #include "device.h"
 #include "helpers.h"
-#include "render_resource_manager.h"
+#include "gpu_resource_manager.h"
 #include "swapchain.h"
 
 #include <EASTL/array.h>
@@ -58,7 +58,7 @@ CreatePipeline(const Device *device, const Swapchain *swapchain, const GpuResour
     vk::PushConstantRange pushConstantRange = {
         .stageFlags = vk::ShaderStageFlagBits::eAll,
         .offset = 0,
-        .size = 24,
+        .size = 36,
     };
 
     vk::PipelineLayoutCreateInfo pipelineLayoutCreateInfo = {

samples/03_model_render/shader/bindless_structs.hlsli

@@ -1,63 +1,75 @@
 
-typedef float4 PositionData;
-typedef float2 UVData;
-typedef float4 NormalData;
-typedef float4 ColorData;
-
 struct VertexData
 {
-    float4 m_Normal;
+    float4 Normal;
-    float2 m_TexCoord0;
+    float2 TexCoord0;
-    float2 m_TexCoord1;
+    float2 TexCoord1;
-    float4 m_Color0;
+    float4 Color0;
 };
 
 struct TransformData
 {
-    float4x4 transform;
+    float4x4 Transform;
+    float4x4 NormalTransform;
 };
 
 struct MaterialData
 {
-    float m_AlbedoFactor[4];
+    float AlbedoFactor[4];
-    float m_EmissionFactor[3];
+    float EmissionFactor[3];
-    float m_MetalFactor;
+    float MetalFactor;
-    float m_RoughFactor;
+    float RoughFactor;
-    uint m_AlbedoTex;
+    uint AlbedoTex;
-    uint m_NormalTex;
+    uint NormalTex;
-    uint m_MetalRoughTex;
+    uint MetalRoughTex;
-    uint m_OcclusionTex;
+    uint OcclusionTex;
-    uint m_EmissionTex;
+    uint EmissionTex;
 };
 
 struct Block
 {
-    uint vertexBufferHandle;
+    uint VertexBufferHandle;
-    uint vertexDataHandle;
+    uint VertexDataHandle;
-    uint materialBufferHandle;
+    uint MaterialBufferHandle;
-    uint nodeBufferHandle;
+    uint NodeBufferHandle;
-    int m_MaterialIdx;
+    uint LightHandle;
-    uint m_NodeIdx;
+    uint PointLightIndexer;
+    uint DirectionLightIndexer;
+    int MaterialIdx;
+    uint NodeIdx;
 };
 
-struct Camera
+struct Light
 {
-    float4x4 model;
+    float Position[3];
-    float4x4 view;
+    float Range;
-    float4x4 proj;
+    uint Color;
+    float Intensity;
 };
 
+struct CameraData
+{
+    float4x4 View;
+    float4x4 Projection;
+    float4 Position;
+};
+
+// Little Endian storage. First short is least significant.
+#define IndexerCount(Indexer) (Indexer & 0xFFFF)            
+#define IndexerOffset(Indexer) ((Indexer & 0xFFFF0000) >> 16);
+
 #define INVALID_HANDLE 0xFFFFFFFF
 
-[[vk::binding(0, 0)]] StructuredBuffer<PositionData> vertexBuffer[];
+[[vk::binding(0, 0)]] StructuredBuffer<float4> VertexBuffer[];
-[[vk::binding(0, 0)]] StructuredBuffer<VertexData> vertexDataBuffer[];
+[[vk::binding(0, 0)]] StructuredBuffer<VertexData> VertexDataBuffer[];
-[[vk::binding(0, 0)]] StructuredBuffer<MaterialData> materialsBuffer[];
+[[vk::binding(0, 0)]] StructuredBuffer<MaterialData> MaterialsBuffer[];
-[[vk::binding(0, 0)]] StructuredBuffer<TransformData> nodeBuffer[];
+[[vk::binding(0, 0)]] StructuredBuffer<TransformData> NodeBuffer[];
+[[vk::binding(0, 0)]] StructuredBuffer<Light> LightBuffer[];
 
-[[vk::binding(1, 0)]] Texture2D<float4> textures[];
+[[vk::binding(1, 0)]] Texture2D<float4> Textures[];
-[[vk::binding(1, 0)]] SamplerState immutableSamplers[];
+[[vk::binding(1, 0)]] SamplerState ImmutableSamplers[];
 
-[[vk::binding(0, 1)]] ConstantBuffer<Camera> camera;
+[[vk::binding(0, 1)]] ConstantBuffer<CameraData> Camera;
 
 [[vk::push_constant]]
-Block pcb;
+Block PushConstant;

samples/03_model_render/shader/model.ps.hlsl

@@ -2,37 +2,99 @@
 
 struct FS_Input
 {
-    float4 inPosition : POSITION;
+    float4 InPosition   : POSITION;
-    float4 inColor : COLOR0;
+    float4 InNormal     : NORMAL;
-    float2 inUV : TEXCOORD0;
+    float4 InColor      : COLOR0;
+    float2 InUV0        : TEXCOORD0;
 };
 
 struct FS_Output
 {
-    float4 outColor : SV_Target0;
+    float4 ColorTarget : SV_Target0;
 };
 
-float4 ArrayToVector(float arr[4])
+float4 GetAlbedo(int MaterialIdx, float2 UV)
 {
-    return float4(arr[0], arr[1], arr[2], arr[3]);
+    uint albedoTexId = MaterialsBuffer[PushConstant.MaterialBufferHandle][MaterialIdx].AlbedoTex;
-}
-
-float4 GetAlbedo(uint materialBufferId, int materialId, float2 uv)
-{
-    uint albedoTexId = materialsBuffer[materialBufferId][materialId].m_AlbedoTex;
     if (albedoTexId == INVALID_HANDLE)
     {
-        return ArrayToVector(materialsBuffer[materialBufferId][materialId].m_AlbedoFactor);
+        return (float4) MaterialsBuffer[PushConstant.MaterialBufferHandle][MaterialIdx].AlbedoFactor;
     }
     else
     {
-        return textures[albedoTexId].Sample(immutableSamplers[albedoTexId], uv);
+        return Textures[albedoTexId].Sample(ImmutableSamplers[albedoTexId], UV);
     }
 }
 
-FS_Output main(FS_Input stage_input)
+float3 GetDirectionalLightInfluence(float3 Normal)
 {
-    FS_Output output;
+    if (PushConstant.LightHandle == INVALID_HANDLE) 
-    output.outColor = pcb.m_MaterialIdx < 0 ? stage_input.inColor : GetAlbedo(pcb.materialBufferHandle, pcb.m_MaterialIdx, stage_input.inUV);
+        return float3(0.0f, 0.0f, 0.0f);
-    return output;
+    
+    uint Count = IndexerCount(PushConstant.DirectionLightIndexer);
+    
+    float3 Contrib = 0.0f;
+    for (uint i = 0; i < Count; ++i)
+    {
+        Light Light = LightBuffer[PushConstant.LightHandle][i];
+        float3 LightDir = - (float3) Light.Position; // Position is actually direction for directionalLight; LightDir is Direction towards the light (-direction)
+        float DiffuseFactor = max(dot(Normal, LightDir), 0.0f);
+    
+        float R = (Light.Color & 0xFF000000) >> 24;
+        float G = (Light.Color & 0x00FF0000) >> 16;
+        float B = (Light.Color & 0x0000FF00) >> 8;
+        
+        float3 Color = float3(R, G, B) * 0.00392156862f; // 0.00392156862 = 1/255
+        
+        float3 Diffuse = DiffuseFactor * Color;
+        
+        Contrib += (Light.Range < 0 ? float3(0.0f, 0.0f, 0.0f) : Diffuse);
+    }
+    
+    return Contrib;
+}
+
+float3 GetPointLightInfluence(float3 Position, float3 Normal)
+{
+    if (PushConstant.LightHandle == INVALID_HANDLE) 
+        return float3(0.0f, 0.0f, 0.0f);
+    
+    uint Offset = IndexerOffset(PushConstant.PointLightIndexer);
+    uint Count = IndexerCount(PushConstant.PointLightIndexer);
+    
+    float3 Contrib = 0.0f;
+    for (uint i = 0; i < Count; ++i)
+    {
+        Light Light = LightBuffer[PushConstant.LightHandle][i + Offset];
+        float3 LightDir = normalize(((float3)Light.Position) - Position);
+        float DiffuseFactor = max(dot(Normal, LightDir), 0.0f);
+    
+        float R = (Light.Color & 0xFF000000) >> 24;
+        float G = (Light.Color & 0x00FF0000) >> 16;
+        float B = (Light.Color & 0x0000FF00) >> 8;
+        
+        float3 Color = float3(R, G, B) * 0.00392156862f; // 0.00392156862 = 1/255
+        
+        float3 Diffuse = DiffuseFactor * Color;
+        
+        Contrib += (Light.Range < 0 ? float3(0.0f, 0.0f, 0.0f) : Diffuse);
+    }
+    
+    return Contrib;
+}
+
+FS_Output main(FS_Input StageInput)
+{
+    float3 Ambient = float3(0.02f, 0.02f, 0.02f);
+    
+    float3 Normal = normalize(StageInput.InNormal.xyz);
+    float3 Position = StageInput.InPosition.xyz;
+    
+    float4 ObjColor = PushConstant.MaterialIdx < 0 ? StageInput.InColor : GetAlbedo(PushConstant.MaterialIdx, StageInput.InUV0);
+    
+    float3 Diffuse = GetDirectionalLightInfluence(Normal) + GetPointLightInfluence(Position, Normal);
+    
+    FS_Output Output;
+    Output.ColorTarget = float4(ObjColor.rgb * (Diffuse + Ambient), ObjColor.a);
+    return Output;
 }

samples/03_model_render/shader/model.vs.hlsl

@@ -2,46 +2,66 @@
 
 struct VS_Input
 {
-    uint vertexIndex : SV_VertexID;
+    uint VertexIndex : SV_VertexID;
 };
 
 struct VS_Output
 {
-    float4 outPosition : POSITION;
+    float4 WorldPosition    : POSITION;
-    float4 outColor : COLOR0;
+    float4 WorldNormal      : NORMAL;
-    float2 outUV : TEXCOORD0;
+    float4 VertexColor      : COLOR0;
-    float4 position : SV_Position;
+    float2 UV0              : TEXCOORD0;
+    float4 VertexPosition   : SV_Position;
 };
 
-float2 GetUV(uint bufferId, uint vertexIdx)
+float2 GetUV(uint VertexIdx)
 {
-    return (bufferId == INVALID_HANDLE) ? 0.0f.xx : vertexDataBuffer[bufferId][vertexIdx].m_TexCoord0.xy;
+    uint BufferId = PushConstant.VertexDataHandle;
+    return VertexDataBuffer[BufferId][VertexIdx].TexCoord0.xy;
 }
 
-float4 GetPosition(uint bufferId, uint vertexIdx)
+float4 GetPosition(uint VertexIdx)
 {
-    return float4(vertexBuffer[bufferId][vertexIdx].xyz, 1.0f);
+    uint BufferId = PushConstant.VertexBufferHandle;
+    return float4(VertexBuffer[BufferId][VertexIdx].xyz, 1.0f);
 }
 
-float4 GetColor(uint bufferId, uint vertexIdx)
+float4 GetNormal(uint VertexIdx)
 {
-    return (bufferId == INVALID_HANDLE) ? float4(1.0f, 0.0f, 1.0f, 1.0f) : vertexDataBuffer[bufferId][vertexIdx].m_Color0;
+    uint BufferId = PushConstant.VertexDataHandle;
+    return VertexDataBuffer[BufferId][VertexIdx].Normal;
 }
 
-float4x4 GetModel(uint bufferId, uint index)
+float4 GetColor(uint VertexIdx)
 {
-    return nodeBuffer[bufferId][index].transform;
+    uint BufferId = PushConstant.VertexDataHandle;
+    return VertexDataBuffer[BufferId][VertexIdx].Color0;
 }
 
-VS_Output main(VS_Input stage_input)
+float4x4 GetNodeTransform(uint NodeIndex)
 {
-    VS_Output output;
+    uint BufferId = PushConstant.NodeBufferHandle;
-    output.outPosition = GetPosition(pcb.vertexBufferHandle, stage_input.vertexIndex);
+    return NodeBuffer[BufferId][NodeIndex].Transform;
-    output.outUV = GetUV(pcb.vertexDataHandle, stage_input.vertexIndex);
+}
-    output.outColor = GetColor(pcb.vertexDataHandle, stage_input.vertexIndex);
+
+float4x4 GetNormalTransform(uint NodeIndex)
+{
+    uint BufferId = PushConstant.NodeBufferHandle;
+    return NodeBuffer[BufferId][NodeIndex].NormalTransform;
+}
+
+VS_Output main(VS_Input StageInput)
+{
+    VS_Output Output;
     
-    float4 globalPosition = mul(camera.model, mul(GetModel(pcb.nodeBufferHandle, pcb.m_NodeIdx), GetPosition(pcb.vertexBufferHandle, stage_input.vertexIndex)));
+    float4 GlobalPosition = mul(GetNodeTransform(PushConstant.NodeIdx), GetPosition(StageInput.VertexIndex));
-    float4 clipSpace = mul(camera.view, globalPosition);
+    float4 ClipSpace = mul(Camera.View, GlobalPosition);
-    output.position = mul(camera.proj, clipSpace);
+    
-    return output;
+    Output.VertexPosition = mul(Camera.Projection, ClipSpace);
+    Output.WorldPosition = GlobalPosition;
+    Output.UV0 = GetUV(StageInput.VertexIndex);
+    Output.VertexColor = GetColor(StageInput.VertexIndex);
+    
+    Output.WorldNormal = mul(GetNormalTransform(PushConstant.NodeIdx), GetNormal(StageInput.VertexIndex));
+    return Output;
 }