Created CPU and GPU data for SSAO

Renderpass for SSAO not yet done (that's next)

Assets/Shaders/SSAO_CS.glsl

@@ -0,0 +1,24 @@
+#version 450
+
+uint const NUM_SAMPLES = 64;
+uint const NUM_ROTATIONS = 16;
+
+layout(local_size_x = 16, local_size_y = 16) in;
+layout(set = 4, binding = 0, rgba32f) uniform image2D positions;
+layout(set = 4, binding = 1, rgba32f) uniform image2D normals;
+layout(set = 4, binding = 2, rgba32f) uniform image2D outputImage;
+
+
+// SSAO data
+layout(std430, set = 5, binding = 0) buffer SSAOData
+{
+  vec3 samples[NUM_SAMPLES];
+  vec3 rotations[NUM_ROTATIONS];
+
+} ssaoData;
+
+void main()
+{ 
+  // store result into result image
+  imageStore(outputImage, ivec2(gl_GlobalInvocationID.xy), vec4(fragColor, 1.0f));
+}

SHADE_Engine/src/Graphics/MiddleEnd/Interface/SHGraphicsSystem.cpp

@@ -253,6 +253,18 @@ namespace SHADE
 
     lightingSubSystem = resourceManager.Create<SHLightingSubSystem>();
     lightingSubSystem->Init(device, descPool);
+
+    ssaoStorage = resourceManager.Create<SHSSAO>();
+
+    ssaoTransferCmdBuffer = graphicsCmdPool->RequestCommandBuffer(SH_CMD_BUFFER_TYPE::PRIMARY);
+    ssaoTransferCmdBuffer->BeginRecording();
+    
+    ssaoStorage->Init(device, ssaoTransferCmdBuffer);
+
+    ssaoTransferCmdBuffer->EndRecording();
+    graphicsQueue->SubmitCommandBuffer({ssaoTransferCmdBuffer});
+
+    graphicsQueue->WaitIdle();
   }
 
 #ifdef SHEDITOR

SHADE_Engine/src/Graphics/MiddleEnd/Interface/SHGraphicsSystem.h

@@ -32,6 +32,7 @@ of DigiPen Institute of Technology is prohibited.
 #include "../Textures/SHVkSamplerCache.h"
 #include "Graphics/MiddleEnd/Interface/SHPostOffscreenRenderSystem.h"
 #include "Graphics/MiddleEnd/Lights/SHLightingSubSystem.h"
+#include "Graphics/MiddleEnd/PostProcessing/SHSSAO.h"
 
 namespace SHADE
 {
@@ -313,6 +314,7 @@ namespace SHADE
     Handle<SHVkDescriptorPool>                                descPool;
     Handle<SHVkCommandPool>                                   graphicsCmdPool;
     Handle<SHVkCommandBuffer>                                 transferCmdBuffer;
+    Handle<SHVkCommandBuffer>                                 ssaoTransferCmdBuffer;
     Handle<SHVkCommandBuffer>                                 graphicsTexCmdBuffer;
     SHRenderContext                                           renderContext;
     std::array<Handle<SHVkSemaphore>, 2>                      graphSemaphores; 
@@ -357,6 +359,7 @@ namespace SHADE
     Handle<SHMousePickSystem>                                 mousePickSystem;
     Handle<SHPostOffscreenRenderSystem>                       postOffscreenRender;
     Handle<SHLightingSubSystem>                               lightingSubSystem;
+    Handle<SHSSAO>                                            ssaoStorage;
 
     uint32_t                                                  resizeWidth = 1;
     uint32_t                                                  resizeHeight = 1;

SHADE_Engine/src/Graphics/MiddleEnd/PostProcessing/SHSSAO.cpp

@@ -1,7 +1,60 @@
 #include "SHpch.h"
 #include "SHSSAO.h"
+#include "Graphics/Devices/SHVkLogicalDevice.h"
+#include "Graphics/Buffers/SHVkBuffer.h"
+#include <random>
 
 namespace SHADE
 {
+  void SHSSAO::Init(Handle<SHVkLogicalDevice> logicalDevice, Handle<SHVkCommandBuffer> cmdBuffer) noexcept
+  {
+    // create memory on CPU side
+    ssaoData = std::make_unique<SamplesAndKernel>();
+
+    // Initialize a distribution to get values from 0 to 1
+    std::uniform_real_distribution<float> distrib{0.0f, 1.0f};
+
+    // generator for random number
+    std::default_random_engine generator;
+
+    // generate samples
+    auto& samples = ssaoData->samples;
+    for (uint32_t i = 0; i < NUM_SAMPLES; ++i)
+    {
+      samples[i] = SHVec4
+      {
+        distrib(generator) * 2.0f - 1.0f, // -1.0f - 1.0f
+        distrib(generator) * 2.0f - 1.0f, // -1.0f - 1.0f
+        distrib(generator),               //  0.0f - 1.0f so that sample space is a hemisphere
+        0.0f
+      };
+
+      // This makes sure that most points are closer to fragment's position
+      float scale = 1.0f / static_cast<float>(NUM_SAMPLES);
+      scale = std::lerp(0.1f, 1.0f, scale * scale);
+      samples[i] *= scale;
+    }
+
+    // generate rotation vector
+    auto& rotationVectors = ssaoData->rotatationVectors;
+    for (uint32_t i = 0; i < NUM_ROTATION_VECTORS; ++i)
+    {
+      samples[i] = SHVec4
+      {
+        distrib(generator) * 2.0f - 1.0f, // -1.0f - 1.0f
+        distrib(generator) * 2.0f - 1.0f, // -1.0f - 1.0f
+        0.0f,                             // we want to rotate about the z axis in tangent space
+        0.0f
+      };
+    }
+
+    // Get aligned size for buffer
+    uint32_t alignedSize = logicalDevice->PadSSBOSize(sizeof (SamplesAndKernel));
+
+    // Create buffer
+    ssaoDataBuffer = logicalDevice->CreateBuffer(alignedSize, ssaoData.get(), alignedSize, vk::BufferUsageFlagBits::eStorageBuffer | vk::BufferUsageFlagBits::eTransferDst, VMA_MEMORY_USAGE_AUTO, {});
+
+    ssaoDataBuffer->TransferToDeviceResource(cmdBuffer);
+  }
 
 }

SHADE_Engine/src/Graphics/MiddleEnd/PostProcessing/SHSSAO.h

@@ -2,14 +2,36 @@
 
 #include "Resource/SHHandle.h"
 #include "Graphics/SHVulkanIncludes.h"
+#include "Math/Vector/SHVec4.h"
 
 namespace SHADE
 {
+  class SHVkBuffer;
+  class SHVkLogicalDevice;
+  class SHVkCommandBuffer;
 
   class SHSSAO
   {
   private:
+    static constexpr uint32_t NUM_SAMPLES = 64;
+    static constexpr uint32_t NUM_ROTATION_VECTORS = 16;
 
+    struct SamplesAndKernel
+    {
+      //! distances from a pixel we want to sample
+      std::array<SHVec4, NUM_SAMPLES>         samples;
 
+      std::array<SHVec4, NUM_ROTATION_VECTORS> rotatationVectors;
+    };
+
+  private:
+    //! Samples and kernel on CPU side
+    std::unique_ptr<SamplesAndKernel> ssaoData;
+
+    //! For passing SSAO samples and kernel to GPU
+    Handle<SHVkBuffer> ssaoDataBuffer;
+
+  public:
+    void Init (Handle<SHVkLogicalDevice> logicalDevice, Handle<SHVkCommandBuffer> cmdBuffer) noexcept;
   };
 }

SHADE_Engine/src/Graphics/Queues/SHVkQueue.cpp

@@ -129,4 +129,9 @@ namespace SHADE
     return vkQueue;
   }
 
+  void SHVkQueue::WaitIdle(void) const noexcept
+  {
+    vkQueue.waitIdle();
+  }
+
 }

SHADE_Engine/src/Graphics/Queues/SHVkQueue.h

@@ -50,6 +50,8 @@ namespace SHADE
     void       SubmitCommandBuffer (std::initializer_list<Handle<SHVkCommandBuffer>> cmdBuffers, std::initializer_list<Handle<SHVkSemaphore>> signalSems = {}, std::initializer_list<Handle<SHVkSemaphore>> waitSems = {}, vk::PipelineStageFlags waitDstStageMask = {}, Handle<SHVkFence> const& fence = {}) noexcept;
     vk::Result Present             (Handle<SHVkSwapchain> const& swapchain, std::initializer_list<Handle<SHVkSemaphore>> waitSems, uint32_t frameIndex) noexcept;
     vk::Queue GetVkQueue() noexcept;
+
+    void WaitIdle (void) const noexcept;
   };
 }
 

SHADE_Engine/src/Graphics/RenderGraph/SHRenderGraphNodeCompute.cpp

@@ -164,4 +164,9 @@ namespace SHADE
     }
   }
 
+  void SHRenderGraphNodeCompute::ModifyWriteDescBufferComputeResource(uint32_t set, uint32_t binding, std::span<Handle<SHVkBuffer>> const& buffers, uint32_t offset, uint32_t range) noexcept
+  {
+    computeResource->descSet->ModifyWriteDescBuffer(set, binding, buffers, offset, range);
+  }
+
 }

SHADE_Engine/src/Graphics/RenderGraph/SHRenderGraphNodeCompute.h

@@ -18,6 +18,7 @@ namespace SHADE
   class SHRenderGraphResource;
   class SHVkShaderModule;
   class SHVkCommandBuffer;
+  class SHVkBuffer;
 
 
   class SHRenderGraphNodeCompute
@@ -37,8 +38,6 @@ namespace SHADE
     static constexpr uint32_t workGroupSizeX = 16;
     static constexpr uint32_t workGroupSizeY = 16;
 
-    using ExternalDescSetBindCommands = std::function<void(Handle<SHVkCommandBuffer>, Handle<SHVkPipelineLayout>)>;
-
     //! To run the dispatch command
     Handle<SHVkPipeline>                       computePipeline;
 
@@ -48,6 +47,7 @@ namespace SHADE
     //! Descriptor set group to hold the images for reading (STORAGE_IMAGE)
     std::array<Handle<SHVkDescriptorSetGroup>, SHGraphicsConstants::NUM_FRAME_BUFFERS> graphResourceDescSets;
 
+    //! Compute resources
     Handle<ComputeResource> computeResource;
 
     //! vector of resources needed by the subpass compute
@@ -73,6 +73,8 @@ namespace SHADE
 
     void SetDynamicOffsets (std::span<uint32_t> perFrameSizes) noexcept;
 
+    void ModifyWriteDescBufferComputeResource (uint32_t set, uint32_t binding, std::span<Handle<SHVkBuffer>> const& buffers, uint32_t offset, uint32_t range) noexcept;
+
     friend class SHRenderGraph;
     friend class SHRenderGraphNode;
   };