Fixed ex 20,21

devshgraphicsprogramming · devshgraphicsprogramming · commit 6abb8fb7968a · 2021-02-03T17:57:51.000+01:00
diff --git a/examples_tests/20.Megatexture/main.cpp b/examples_tests/20.Megatexture/main.cpp
@@ -10,9 +10,7 @@
 //! I advise to check out this file, its a basic input handler
 #include "../common/QToQuitEventReceiver.h"
 #include <nbl/video/IGPUVirtualTexture.h>
-#include <nbl/asset/CMTLPipelineMetadata.h>
 #include "nbl/ext/FullScreenTriangle/FullScreenTriangle.h"
-#include <nbl/asset/filters/CMipMapGenerationImageFilter.h>
 
 //#include "nbl/ext/ScreenShot/ScreenShot.h"
 using namespace nbl;
@@ -177,12 +175,12 @@ struct SPushConstants
 static_assert(sizeof(SPushConstants)<=asset::ICPUMeshBuffer::MAX_PUSH_CONSTANT_BYTESIZE, "doesnt fit in push constants");
 
 constexpr uint32_t texturesOfInterest[TEX_OF_INTEREST_CNT]{
-    asset::CMTLPipelineMetadata::EMP_AMBIENT,
-    asset::CMTLPipelineMetadata::EMP_DIFFUSE,
-    asset::CMTLPipelineMetadata::EMP_SPECULAR,
-    asset::CMTLPipelineMetadata::EMP_SHININESS,
-    asset::CMTLPipelineMetadata::EMP_OPACITY,
-    asset::CMTLPipelineMetadata::EMP_BUMP
+    asset::CMTLMetadata::CRenderpassIndependentPipeline::EMP_AMBIENT,
+    asset::CMTLMetadata::CRenderpassIndependentPipeline::EMP_DIFFUSE,
+    asset::CMTLMetadata::CRenderpassIndependentPipeline::EMP_SPECULAR,
+    asset::CMTLMetadata::CRenderpassIndependentPipeline::EMP_SHININESS,
+    asset::CMTLMetadata::CRenderpassIndependentPipeline::EMP_OPACITY,
+    asset::CMTLMetadata::CRenderpassIndependentPipeline::EMP_BUMP
 };
 
 core::smart_refctd_ptr<asset::ICPUSpecializedShader> createModifiedFragShader(const asset::ICPUSpecializedShader* _fs, const asset::ICPUVirtualTexture* _vt)
@@ -290,19 +288,23 @@ int main()
 
     asset::IAssetLoader::SAssetLoadParams lp;
     auto meshes_bundle = am->getAsset("sponza.obj", lp);
-    assert(!meshes_bundle.isEmpty());
+    assert(!meshes_bundle.getContents().empty());
+
+    const auto meta = meshes_bundle.getMetadata()->selfCast<const asset::COBJMetadata>();
+
     auto mesh = meshes_bundle.getContents().begin()[0];
     auto mesh_raw = static_cast<asset::ICPUMesh*>(mesh.get());
 
+    // all pipelines will have the same metadata
+    const asset::CMTLMetadata::CRenderpassIndependentPipeline* pipelineMetadata = nullptr;
     core::vector<commit_t> vt_commits;
     //modifying push constants and default fragment shader for VT
-    for (uint32_t i = 0u; i < mesh_raw->getMeshBufferCount(); ++i)
+    for (auto mb : mesh_raw->getMeshBuffers())
     {
         SPushConstants pushConsts;
         memset(pushConsts.map_data, 0xff, TEX_OF_INTEREST_CNT*sizeof(pushConsts.map_data[0]));
         pushConsts.extra = 0u;
 
-        auto* mb = mesh_raw->getMeshBuffer(i);
         auto* ds = mb->getAttachedDescriptorSet();
         if (!ds)
             continue;
@@ -333,24 +335,24 @@ int main()
             pushConsts.map_data[k] = reinterpret_cast<uint64_t*>(&texData)[0];
         }
 
-        auto* pipeline = mb->getPipeline();
-        auto* metadata = static_cast<asset::CMTLPipelineMetadata*>( pipeline->getMetadata() );
+        pipelineMetadata = static_cast<const asset::CMTLMetadata::CRenderpassIndependentPipeline*>(meta->getAssetSpecificMetadata(mb->getPipeline()));
 
         //copy texture presence flags
-        pushConsts.extra = metadata->getMaterialParams().extra;
-        pushConsts.ambient = metadata->getMaterialParams().ambient;
-        pushConsts.diffuse = metadata->getMaterialParams().diffuse;
-        pushConsts.emissive = metadata->getMaterialParams().emissive;
-        pushConsts.specular = metadata->getMaterialParams().specular;
-        pushConsts.IoR = metadata->getMaterialParams().IoR;
-        pushConsts.opacity = metadata->getMaterialParams().opacity;
-        pushConsts.shininess = metadata->getMaterialParams().shininess;
+        pushConsts.extra = pipelineMetadata->m_materialParams.extra;
+        pushConsts.ambient = pipelineMetadata->m_materialParams.ambient;
+        pushConsts.diffuse = pipelineMetadata->m_materialParams.diffuse;
+        pushConsts.emissive = pipelineMetadata->m_materialParams.emissive;
+        pushConsts.specular = pipelineMetadata->m_materialParams.specular;
+        pushConsts.IoR = pipelineMetadata->m_materialParams.IoR;
+        pushConsts.opacity = pipelineMetadata->m_materialParams.opacity;
+        pushConsts.shininess = pipelineMetadata->m_materialParams.shininess;
         memcpy(mb->getPushConstantsDataPtr(), &pushConsts, sizeof(pushConsts));
 
         //we dont want this DS to be converted into GPU DS, so set to nullptr
         //dont worry about deletion of textures (invalidation of pointers), they're grabbed in VTtexDataMap
         mb->setAttachedDescriptorSet(nullptr);
     }
+    assert(pipelineMetadata);
 
     core::smart_refctd_ptr<asset::ICPUDescriptorSetLayout> ds0layout;
     {
@@ -383,15 +385,15 @@ int main()
         pipelineLayout = core::make_smart_refctd_ptr<asset::ICPUPipelineLayout>(&pcrng, &pcrng + 1, core::smart_refctd_ptr(ds0layout), nullptr, core::smart_refctd_ptr(ds2layout), nullptr);
     }
 
-    for (uint32_t i = 0u; i < mesh_raw->getMeshBufferCount(); ++i)
+    for (auto mb : mesh_raw->getMeshBuffers())
     {
-        auto* mb = mesh_raw->getMeshBuffer(i);
-
         auto* pipeline = mb->getPipeline();
+
         auto newPipeline = core::smart_refctd_ptr_static_cast<asset::ICPURenderpassIndependentPipeline>(pipeline->clone(0u));//shallow copy
         //leave original ds1 layout since it's for UBO with matrices
         if (!pipelineLayout->getDescriptorSetLayout(1u))
             pipelineLayout->setDescriptorSetLayout(1u, core::smart_refctd_ptr<asset::ICPUDescriptorSetLayout>(pipeline->getLayout()->getDescriptorSetLayout(1u)));
+
         newPipeline->setLayout(core::smart_refctd_ptr(pipelineLayout));
         {
             auto* fs = pipeline->getShaderAtIndex(asset::ICPURenderpassIndependentPipeline::ESSI_FRAGMENT_SHADER_IX);
@@ -405,8 +407,6 @@ int main()
             }
             newPipeline->setShaderAtIndex(asset::ICPURenderpassIndependentPipeline::ESSI_FRAGMENT_SHADER_IX, newfs.get());
         }
-        auto* metadata = static_cast<asset::CMTLPipelineMetadata*>(pipeline->getMetadata());
-        am->setAssetMetadata(newPipeline.get(), core::smart_refctd_ptr<asset::IAssetMetadata>(metadata));
 
         //set new pipeline (with overriden FS and layout)
         mb->setPipeline(std::move(newPipeline));
@@ -435,7 +435,7 @@ int main()
     //we can safely assume that all meshbuffers within mesh loaded from OBJ has same DS1 layout (used for camera-specific data)
     //so we can create just one DS
     
-    asset::ICPUDescriptorSetLayout* ds1layout = mesh_raw->getMeshBuffer(0u)->getPipeline()->getLayout()->getDescriptorSetLayout(1u);
+    asset::ICPUDescriptorSetLayout* ds1layout = mesh_raw->getMeshBuffers().begin()[0]->getPipeline()->getLayout()->getDescriptorSetLayout(1u);
     uint32_t ds1UboBinding = 0u;
     for (const auto& bnd : ds1layout->getBindings())
         if (bnd.type==asset::EDT_UNIFORM_BUFFER)
@@ -446,9 +446,8 @@ int main()
 
     size_t neededDS1UBOsz = 0ull;
     {
-        auto pipelineMetadata = static_cast<const asset::IPipelineMetadata*>(mesh_raw->getMeshBuffer(0u)->getPipeline()->getMetadata());
-        for (const auto& shdrIn : pipelineMetadata->getCommonRequiredInputs())
-            if (shdrIn.descriptorSection.type==asset::IPipelineMetadata::ShaderInput::ET_UNIFORM_BUFFER && shdrIn.descriptorSection.uniformBufferObject.set==1u && shdrIn.descriptorSection.uniformBufferObject.binding==ds1UboBinding)
+        for (const auto& shdrIn : pipelineMetadata->getRequiredShaderInputs())
+            if (shdrIn.descriptorSection.type==asset::IRenderpassIndependentPipelineMetadata::ShaderInput::ET_UNIFORM_BUFFER && shdrIn.descriptorSection.uniformBufferObject.set==1u && shdrIn.descriptorSection.uniformBufferObject.binding==ds1UboBinding)
                 neededDS1UBOsz = std::max<size_t>(neededDS1UBOsz, shdrIn.descriptorSection.uniformBufferObject.relByteoffset+shdrIn.descriptorSection.uniformBufferObject.bytesize);
     }
 
@@ -519,26 +518,25 @@ int main()
 		camera->render();
 
         core::vector<uint8_t> uboData(gpuubo->getSize());
-        auto pipelineMetadata = static_cast<const asset::IPipelineMetadata*>(mesh_raw->getMeshBuffer(0u)->getPipeline()->getMetadata());
-        for (const auto& shdrIn : pipelineMetadata->getCommonRequiredInputs())
+        for (const auto& shdrIn : pipelineMetadata->getRequiredShaderInputs())
         {
-            if (shdrIn.descriptorSection.type==asset::IPipelineMetadata::ShaderInput::ET_UNIFORM_BUFFER && shdrIn.descriptorSection.uniformBufferObject.set==1u && shdrIn.descriptorSection.uniformBufferObject.binding==ds1UboBinding)
+            if (shdrIn.descriptorSection.type==asset::IRenderpassIndependentPipelineMetadata::ShaderInput::ET_UNIFORM_BUFFER && shdrIn.descriptorSection.uniformBufferObject.set==1u && shdrIn.descriptorSection.uniformBufferObject.binding==ds1UboBinding)
             {
                 switch (shdrIn.type)
                 {
-                case asset::IPipelineMetadata::ECSI_WORLD_VIEW_PROJ:
+                case asset::IRenderpassIndependentPipelineMetadata::ECSI_WORLD_VIEW_PROJ:
                 {
                     core::matrix4SIMD mvp = camera->getConcatenatedMatrix();
                     memcpy(uboData.data()+shdrIn.descriptorSection.uniformBufferObject.relByteoffset, mvp.pointer(), shdrIn.descriptorSection.uniformBufferObject.bytesize);
                 }
                 break;
-                case asset::IPipelineMetadata::ECSI_WORLD_VIEW:
+                case asset::IRenderpassIndependentPipelineMetadata::ECSI_WORLD_VIEW:
                 {
                     core::matrix3x4SIMD MV = camera->getViewMatrix();
                     memcpy(uboData.data() + shdrIn.descriptorSection.uniformBufferObject.relByteoffset, MV.pointer(), shdrIn.descriptorSection.uniformBufferObject.bytesize);
                 }
                 break;
-                case asset::IPipelineMetadata::ECSI_WORLD_VIEW_INVERSE_TRANSPOSE:
+                case asset::IRenderpassIndependentPipelineMetadata::ECSI_WORLD_VIEW_INVERSE_TRANSPOSE:
                 {
                     core::matrix3x4SIMD MV = camera->getViewMatrix();
                     memcpy(uboData.data()+shdrIn.descriptorSection.uniformBufferObject.relByteoffset, MV.pointer(), shdrIn.descriptorSection.uniformBufferObject.bytesize);
@@ -549,9 +547,8 @@ int main()
         }       
         driver->updateBufferRangeViaStagingBuffer(gpuubo.get(), 0ull, gpuubo->getSize(), uboData.data());
 
-        for (uint32_t i = 0u; i < gpumesh->getMeshBufferCount(); ++i)
+        for (auto gpumb : gpumesh->getMeshBuffers())
         {
-            video::IGPUMeshBuffer* gpumb = gpumesh->getMeshBuffer(i);
             const video::IGPURenderpassIndependentPipeline* pipeline = gpumb->getPipeline();
 
             driver->bindGraphicsPipeline(pipeline);
diff --git a/examples_tests/21.DynamicTextureIndexing/main.cpp b/examples_tests/21.DynamicTextureIndexing/main.cpp
@@ -7,8 +7,7 @@
 
 //! I advise to check out this file, its a basic input handler
 #include "../common/QToQuitEventReceiver.h"
-#include <nbl/asset/CCPUMeshPacker.h>
-//#include "nbl/ext/ScreenShot/ScreenShot.h"
+#include "nbl/asset/utils/CCPUMeshPacker.h"
 
 using namespace nbl;
 using namespace core;
@@ -63,13 +62,12 @@ int main()
     auto mesh_raw = static_cast<asset::ICPUMesh*>(mesh.get());
 
     //saving cache to file
-    //qnc->saveCacheToFile(asset::CQuantNormalCache::E_CACHE_TYPE::ECT_2_10_10_10, fs, "../../tmp/normalCache101010.sse");
-    
-    const uint32_t mbCount = mesh_raw->getMeshBufferCount();
+    qnc->saveCacheToFile(asset::CQuantNormalCache::E_CACHE_TYPE::ECT_2_10_10_10, fs, "../../tmp/normalCache101010.sse");
 
-    core::vector<ICPUMeshBuffer*> meshBuffers(mbCount);
-    for (uint32_t i = 0u; i < mbCount; i++)
-        meshBuffers[i] = mesh_raw->getMeshBuffer(i);
+    const auto meta = meshes_bundle.getMetadata()->selfCast<const COBJMetadata>();
+    
+    const auto meshbuffers = mesh_raw->getMeshBuffers();
+    core::vector<ICPUMeshBuffer*> meshBuffers(meshbuffers.begin(),meshbuffers.end());
 
         //divide mesh buffers into sets, where sum of textures used by meshes in one set is <= 16
     struct MBRangeTexturesPair
@@ -87,7 +85,7 @@ int main()
 
         for (auto it = meshBuffers.begin(); it < meshBuffers.end(); it++)
         {
-            auto ds3 = dynamic_cast<CMTLPipelineMetadata*>((*it)->getPipeline()->getMetadata())->getDescriptorSet();
+            auto ds3 = (*it)->getAttachedDescriptorSet();
             ICPUImageView* tex = dynamic_cast<ICPUImageView*>(ds3->getDescriptors(0u).begin()->desc.get());
             ICPUImageView* texBump = dynamic_cast<ICPUImageView*>(ds3->getDescriptors(6u).begin()->desc.get());
 
@@ -140,7 +138,7 @@ int main()
     core::vector<core::smart_refctd_ptr<IGPUBuffer>> gpuIndirectDrawBuffer(mbRangesTex.size());
     for(size_t i = 0u; i < mbRangesTex.size(); i++)
     {
-        asset::SVertexInputParams& inputParams = mesh_raw->getMeshBuffer(0u)->getPipeline()->getVertexInputParams();
+        asset::SVertexInputParams& inputParams = mesh_raw->getMeshBuffers().begin()[0]->getPipeline()->getVertexInputParams();
         asset::MeshPackerBase::AllocationParams allocationParams;
 
         auto& mbRange = mbRangesTex[i].mbRanges;
@@ -180,7 +178,7 @@ int main()
         uint32_t mbIdx = 0u;
         for (auto it = mbRange.begin(); it != mbRange.end(); it++)
         {
-            auto ds3 = dynamic_cast<CMTLPipelineMetadata*>((*it)->getPipeline()->getMetadata())->getDescriptorSet();
+            auto ds3 = (*it)->getAttachedDescriptorSet();
             ICPUImageView* tex = dynamic_cast<ICPUImageView*>(ds3->getDescriptors(0u).begin()->desc.get());
             ICPUImageView* texBump = dynamic_cast<ICPUImageView*>(ds3->getDescriptors(6u).begin()->desc.get());
 
diff --git a/include/nbl/asset/utils/CCPUMeshPacker.h b/include/nbl/asset/utils/CCPUMeshPacker.h
@@ -150,7 +150,7 @@ MeshPackerBase::ReservedAllocationMeshBuffers CCPUMeshPacker<MDIStructType>::all
 	size_t vtxCnt = 0u;
 	for (auto it = begin; it != end; it++)
 	{
-		ICPUMeshBuffer* mb = it->get();
+		ICPUMeshBuffer* mb = *it;
 		idxCnt += mb->getIndexCount();
 		vtxCnt += IMeshManipulator::upperBoundVertexID(mb);
 	}

Original file line number	Diff line number	Diff line change
`@@ -150,7 +150,7 @@ MeshPackerBase::ReservedAllocationMeshBuffers CCPUMeshPacker<MDIStructType>::all`
`150`	`150`	`size_t vtxCnt = 0u;`
`151`	`151`	`for (auto it = begin; it != end; it++)`
`152`	`152`	`{`
`153`		`- ICPUMeshBuffer* mb = it->get();`
	`153`	`+ ICPUMeshBuffer* mb = *it;`
`154`	`154`	`idxCnt += mb->getIndexCount();`
`155`	`155`	`vtxCnt += IMeshManipulator::upperBoundVertexID(mb);`
`156`	`156`	`}`