XCEngine/engine/third_party/physx/snippets/snippetdeformablevolumeskinning/SnippetDeformableVolumeSkinning.cpp

// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions
// are met:
//  * Redistributions of source code must retain the above copyright
//    notice, this list of conditions and the following disclaimer.
//  * Redistributions in binary form must reproduce the above copyright
//    notice, this list of conditions and the following disclaimer in the
//    documentation and/or other materials provided with the distribution.
//  * Neither the name of NVIDIA CORPORATION nor the names of its
//    contributors may be used to endorse or promote products derived
//    from this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS ''AS IS'' AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
// PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT OWNER OR
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY
// OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Copyright (c) 2008-2025 NVIDIA Corporation. All rights reserved.
// Copyright (c) 2004-2008 AGEIA Technologies, Inc. All rights reserved.
// Copyright (c) 2001-2004 NovodeX AG. All rights reserved.  

// ****************************************************************************
// This snippet demonstrates how to setup deformable volumes.
// ****************************************************************************

#include <ctype.h>
#include "PxPhysicsAPI.h"
#include "../snippetcommon/SnippetPrint.h"
#include "../snippetcommon/SnippetPVD.h"
#include "../snippetutils/SnippetUtils.h"
#include "../snippetdeformablevolumeskinning/SnippetDeformableVolumeSkinning.h"
#include "../snippetdeformablevolume/MeshGenerator.h"
#include "extensions/PxTetMakerExt.h"
#include "extensions/PxDeformableVolumeExt.h"

#include "PxDeformableSkinning.h"
#include "gpu/PxPhysicsGpu.h"
#include "extensions/PxCudaHelpersExt.h"
#include "extensions/PxDeformableSkinningExt.h"
#include "extensions/PxRemeshingExt.h"

using namespace physx;
using namespace physx::Ext;
using namespace meshgenerator;

static PxDefaultAllocator		gAllocator;
static PxDefaultErrorCallback	gErrorCallback;
static PxFoundation* gFoundation = NULL;
static PxPhysics* gPhysics = NULL;
static PxCudaContextManager* gCudaContextManager = NULL;
static PxDefaultCpuDispatcher* gDispatcher = NULL;
static PxScene* gScene = NULL;
static PxMaterial* gMaterial = NULL;
static PxPvd* gPvd = NULL;
PxArray<DeformableVolume> gDeformableVolumes;
PxArray<SkinnedMesh> gSkinnedMeshes;
BasePostSolveCallback* gSkinning;

template<typename T>
class HostAndDeviceBuffer
{
public:
	PxCudaContextManager* mContextManager;
	T* mDeviceData;
	T* mHostData; //Pinned host memory
	PxU32 mNumElements;

	HostAndDeviceBuffer() :
		mContextManager(NULL), mDeviceData(NULL), mHostData(NULL), mNumElements(0)
	{}

	HostAndDeviceBuffer(PxCudaContextManager* contextManager, PxU32 numElements) :
		mContextManager(contextManager), mDeviceData(NULL), mHostData(NULL), mNumElements(0)
	{
		allocate(numElements);
	}

	void initialize(PxCudaContextManager* contextManager, PxU32 numElements)
	{
		mContextManager = contextManager;
		allocate(numElements);
	}

	void initialize(PxCudaContextManager* contextManager, const T* dataSource, PxU32 numElements)
	{
		mContextManager = contextManager;
		allocate(numElements);
		PxMemCopy(mHostData, dataSource, numElements * sizeof(T));
	}

	void allocate(PxU32 numElements)
	{
		release();
		mDeviceData = PxCudaHelpersExt::allocDeviceBuffer<T>(*mContextManager, numElements);
		mHostData = PxCudaHelpersExt::allocPinnedHostBuffer<T>(*mContextManager, numElements);
		mNumElements = numElements;
	}

	void copyDeviceToHost(PxU32 numElementsToCopy = 0xFFFFFFFF)
	{
		PxCudaHelpersExt::copyDToH(*mContextManager, mHostData, mDeviceData, PxMin(numElementsToCopy, mNumElements));
	}

	void copyHostToDevice(PxU32 numElementsToCopy = 0xFFFFFFFF)
	{
		PxCudaHelpersExt::copyHToD<T>(*mContextManager, mDeviceData, mHostData, PxMin(numElementsToCopy, mNumElements));
	}

	void copyDeviceToHostAsync(CUstream stream, PxU32 numElementsToCopy = 0xFFFFFFFF)
	{
		PxCudaHelpersExt::copyDToHAsync(*mContextManager, mHostData, mDeviceData, PxMin(numElementsToCopy, mNumElements), stream);
	}

	void release()
	{
		PxCudaHelpersExt::freeDeviceBuffer(*mContextManager, mDeviceData);
		PxCudaHelpersExt::freePinnedHostBuffer(*mContextManager, mHostData);
	}
};

struct VolumeSkinningHelper
{
	PxDeformableVolume* mDeformableVolume;
	HostAndDeviceBuffer<PxU32> mDeformableVolumeTets;
	HostAndDeviceBuffer<PxTetrahedronMeshEmbeddingInfo> mSkinningInfo;
	HostAndDeviceBuffer<PxVec3> mSkinnedVertices;
	PxU32 mNumSkinnedVertices;

	VolumeSkinningHelper() : mDeformableVolume(NULL), mNumSkinnedVertices(0)
	{ }

	VolumeSkinningHelper(PxCudaContextManager* contextManager, PxDeformableVolume* deformableVolume, PxVec3* skinnedPointsRestPosition, PxU32 nbSkinnedPoints)
		: mDeformableVolume(deformableVolume)
	{
		PxTetrahedronMesh& simulationMesh = *deformableVolume->getSimulationMesh();

		PxU32 nbTetrahedra = simulationMesh.getNbTetrahedrons();

		bool uses16bit = simulationMesh.getTetrahedronMeshFlags() & PxTetrahedronMeshFlag::e16_BIT_INDICES;
		mDeformableVolumeTets.initialize(contextManager, 4 * nbTetrahedra);

		if (uses16bit)
		{
			const PxU16* tetIndices = reinterpret_cast<const PxU16*>(simulationMesh.getTetrahedrons());
			for (PxU32 i = 0; i < mDeformableVolumeTets.mNumElements; ++i)
				mDeformableVolumeTets.mHostData[i] = tetIndices[i];
		}
		else
		{
			const PxU32* tetIndices = reinterpret_cast<const PxU32*>(simulationMesh.getTetrahedrons());
			for (PxU32 i = 0; i < mDeformableVolumeTets.mNumElements; ++i)
				mDeformableVolumeTets.mHostData[i] = tetIndices[i];
		}

		mSkinnedVertices.initialize(contextManager, skinnedPointsRestPosition, nbSkinnedPoints);
		mNumSkinnedVertices = nbSkinnedPoints;
		mSkinningInfo.initialize(contextManager, nbSkinnedPoints);

		PxDeformableSkinningExt::initializeInterpolatedVertices(
			mSkinningInfo.mHostData, simulationMesh.getVertices(), mDeformableVolumeTets.mHostData,
			nbTetrahedra, skinnedPointsRestPosition, nbSkinnedPoints);

		mDeformableVolumeTets.copyHostToDevice();
		mSkinnedVertices.copyHostToDevice();
		mSkinningInfo.copyHostToDevice();
	}

	void packageGpuData(PxTetmeshSkinningGpuData& target)
	{
		target.guideVerticesD.data = reinterpret_cast<PxVec3*>(mDeformableVolume->getSimPositionInvMassBufferD());
		target.guideVerticesD.stride = sizeof(PxVec4);
		target.guideTetrahedraD = mDeformableVolumeTets.mDeviceData;
		target.skinningInfoPerVertexD = mSkinningInfo.mDeviceData;
		target.skinnedVerticesD.count = mNumSkinnedVertices;
		target.skinnedVerticesD.stride = sizeof(PxVec3);
		target.skinnedVerticesD.data = mSkinnedVertices.mHostData; //This works because it's pinned memory - no device to host transfer will be required
	}

	void release()
	{
		mDeformableVolumeTets.release();
		mSkinnedVertices.release();
		mSkinningInfo.release();
	}
};


struct PostSolveCallback : BasePostSolveCallback, PxUserAllocated
{
	CUstream mSkinningStream;
	PxCudaContextManager* mContextManager;
	PxDeformableSkinning* skinning;
	PxArray<VolumeSkinningHelper> skinningHelpers;
	HostAndDeviceBuffer<PxTetmeshSkinningGpuData> packagedSkinningData; 


	PostSolveCallback(PxCudaContextManager* contextManager, PxU32 maxNumDeformableVolumes) :
		mContextManager(contextManager)
	{
		const PxU32 CU_STREAM_NON_BLOCKING = 0x1;
		mContextManager->getCudaContext()->streamCreate(&mSkinningStream, CU_STREAM_NON_BLOCKING);

		skinning = PxGetPhysicsGpu()->createDeformableSkinning(contextManager);

		packagedSkinningData.initialize(contextManager, maxNumDeformableVolumes);
		skinningHelpers.resize(maxNumDeformableVolumes);
	}

	void setDeformableVolume(PxU32 index, PxDeformableVolume* deformableVolume, PxVec3* skinnedPointsRestPosition, PxU32 nbSkinnedPoints)
	{
		skinningHelpers[index] = VolumeSkinningHelper(mContextManager, deformableVolume, skinnedPointsRestPosition, nbSkinnedPoints);
	}

	virtual void onPostSolve(CUevent startEvent)
	{
		mContextManager->getCudaContext()->streamWaitEvent(mSkinningStream, startEvent);

		for (PxU32 i = 0; i < skinningHelpers.size(); ++i) 		
			skinningHelpers[i].packageGpuData(packagedSkinningData.mHostData[i]);
		
		packagedSkinningData.copyHostToDevice(skinningHelpers.size());

		skinning->evaluateVerticesEmbeddedIntoVolume(packagedSkinningData.mDeviceData, skinningHelpers.size(), mSkinningStream);

		//mSkinnedVertices.copyDeviceToHostAsync(mSkinningStream);
	}

	virtual void synchronize()
	{
		mContextManager->getCudaContext()->streamSynchronize(mSkinningStream);
	}

	virtual PxVec3* getSkinnedVertices(PxU32 deformableVolumeIndex)
	{
		return skinningHelpers[deformableVolumeIndex].mSkinnedVertices.mHostData;
	}

	~PostSolveCallback()
	{
		mContextManager->getCudaContext()->streamDestroy(mSkinningStream);
		for (PxU32 i = 0; i < skinningHelpers.size(); ++i)
			skinningHelpers[i].release();
		PX_DELETE(skinning);
	}
};

PostSolveCallback* postSolveCallback;

void addDeformableVolume(PxDeformableVolume* deformableVolume, const PxTransform& transform, const PxReal density, const PxReal scale)
{
	PxVec4* simPositionInvMassPinned;
	PxVec4* simVelocityPinned;
	PxVec4* collPositionInvMassPinned;
	PxVec4* restPositionPinned;

	PxDeformableVolumeExt::allocateAndInitializeHostMirror(*deformableVolume, gCudaContextManager, simPositionInvMassPinned, simVelocityPinned, collPositionInvMassPinned, restPositionPinned);

	const PxReal maxInvMassRatio = 50.f;

	PxDeformableVolumeExt::transform(*deformableVolume, transform, scale, simPositionInvMassPinned, simVelocityPinned, collPositionInvMassPinned, restPositionPinned);
	PxDeformableVolumeExt::updateMass(*deformableVolume, density, maxInvMassRatio, simPositionInvMassPinned);
	PxDeformableVolumeExt::copyToDevice(*deformableVolume, PxDeformableVolumeDataFlag::eALL, simPositionInvMassPinned, simVelocityPinned, collPositionInvMassPinned, restPositionPinned);

	DeformableVolume volume(deformableVolume, gCudaContextManager);

	gDeformableVolumes.pushBack(volume);

	PX_EXT_PINNED_MEMORY_FREE(*gCudaContextManager, simPositionInvMassPinned);
	PX_EXT_PINNED_MEMORY_FREE(*gCudaContextManager, simVelocityPinned);
	PX_EXT_PINNED_MEMORY_FREE(*gCudaContextManager, collPositionInvMassPinned);
	PX_EXT_PINNED_MEMORY_FREE(*gCudaContextManager, restPositionPinned);
}

static PxDeformableVolume* createDeformableVolume(const PxCookingParams& params, const PxArray<PxVec3>& triVerts, const PxArray<PxU32>& triIndices, bool useCollisionMeshForSimulation = false)
{
	PxDeformableVolumeMesh* deformableVolumeMesh;

	PxU32 numVoxelsAlongLongestAABBAxis = 8;

	PxSimpleTriangleMesh surfaceMesh;
	surfaceMesh.points.count = triVerts.size();
	surfaceMesh.points.data = triVerts.begin();
	surfaceMesh.triangles.count = triIndices.size() / 3;
	surfaceMesh.triangles.data = triIndices.begin();

	if (useCollisionMeshForSimulation)
	{
		deformableVolumeMesh = PxDeformableVolumeExt::createDeformableVolumeMeshNoVoxels(params, surfaceMesh, gPhysics->getPhysicsInsertionCallback());
	}
	else
	{
		deformableVolumeMesh = PxDeformableVolumeExt::createDeformableVolumeMesh(params, surfaceMesh, numVoxelsAlongLongestAABBAxis, gPhysics->getPhysicsInsertionCallback());
	}

	//Alternatively one can cook a deformable volume mesh in a single step
	//tetMesh = cooking.createDeformableVolumeMesh(simulationMeshDesc, collisionMeshDesc, deformableVolumeDesc, physics.getPhysicsInsertionCallback());
	PX_ASSERT(deformableVolumeMesh);

	if (!gCudaContextManager)
		return NULL;
	PxDeformableVolume* deformableVolume = gPhysics->createDeformableVolume(*gCudaContextManager);
	if (deformableVolume)
	{
		PxShapeFlags shapeFlags = PxShapeFlag::eVISUALIZATION | PxShapeFlag::eSCENE_QUERY_SHAPE | PxShapeFlag::eSIMULATION_SHAPE;

		PxDeformableVolumeMaterial* materialPtr = PxGetPhysics().createDeformableVolumeMaterial(2.e+5f, 0.3f, 0.1f);
		PxTetrahedronMeshGeometry geometry(deformableVolumeMesh->getCollisionMesh());
		PxShape* shape = gPhysics->createShape(geometry, &materialPtr, 1, true, shapeFlags);
		if (shape)
		{
			deformableVolume->attachShape(*shape);
			shape->setSimulationFilterData(PxFilterData(0, 0, 2, 0));
		}
		deformableVolume->attachSimulationMesh(*deformableVolumeMesh->getSimulationMesh(), *deformableVolumeMesh->getDeformableVolumeAuxData());

		gScene->addActor(*deformableVolume);

		addDeformableVolume(deformableVolume, PxTransform(PxVec3(0.f, 0.f, 0.f), PxQuat(PxIdentity)), 100.f, 1.0f);
		deformableVolume->setDeformableBodyFlag(PxDeformableBodyFlag::eDISABLE_SELF_COLLISION, true);
		deformableVolume->setSolverIterationCounts(30);


		PxArray<PxU32> subdividedTriangles = triIndices;
		PxArray<PxVec3> subdividedVertices = triVerts;
		PxRemeshingExt::limitMaxEdgeLength(subdividedTriangles, subdividedVertices, 0.0001f, 2);

		SkinnedMesh mesh;
		for (PxU32 i = 0; i < subdividedTriangles.size(); ++i)
			mesh.mTriangles.pushBack(subdividedTriangles[i]);
		for (PxU32 i = 0; i < subdividedVertices.size(); ++i)
			mesh.mVertices.pushBack(subdividedVertices[i]);

		gSkinnedMeshes.pushBack(mesh);
	}
	return deformableVolume;
}

static void createDeformableVolumes(const PxCookingParams& params)
{
	if (gCudaContextManager == NULL)
	{
		printf("The Deformable Volume feature is currently only supported on GPU\n");
		return;
	}

	PxArray<PxVec3> triVerts;
	PxArray<PxU32> triIndices;

	PxReal maxEdgeLength = 1;

	createCube(triVerts, triIndices, PxVec3(0.0, 9, 0), 2.5);
	PxRemeshingExt::limitMaxEdgeLength(triIndices, triVerts, maxEdgeLength);
	createDeformableVolume(params, triVerts, triIndices);

	createSphere(triVerts, triIndices, PxVec3(0, 4.5, 0), 2.5, maxEdgeLength);
	createDeformableVolume(params, triVerts, triIndices);

	createConeY(triVerts, triIndices, PxVec3(0.1, 11.5, 0), 2.0f, 3.5);
	PxRemeshingExt::limitMaxEdgeLength(triIndices, triVerts, maxEdgeLength);
	createDeformableVolume(params, triVerts, triIndices);

	postSolveCallback = PX_NEW(PostSolveCallback)(gCudaContextManager, PxU32(gSkinnedMeshes.size()));
	gSkinning = postSolveCallback;

	for (PxU32 i = 0; i < gSkinnedMeshes.size(); ++i) 
	{
		SkinnedMesh& skinnedMesh = gSkinnedMeshes[i];
		postSolveCallback->setDeformableVolume(i, gDeformableVolumes[i].mDeformableVolume, &skinnedMesh.mVertices[0], PxU32(skinnedMesh.mVertices.size()));
	}

	gScene->setDeformableVolumeGpuPostSolveCallback(postSolveCallback);
}

void initPhysics(bool /*interactive*/)
{
	gFoundation = PxCreateFoundation(PX_PHYSICS_VERSION, gAllocator, gErrorCallback);
	gPvd = PxCreatePvd(*gFoundation);
	PxPvdTransport* transport = PxDefaultPvdSocketTransportCreate(PVD_HOST, 5425, 10);
	gPvd->connect(*transport, PxPvdInstrumentationFlag::eALL);

	// initialize cuda
	PxCudaContextManagerDesc cudaContextManagerDesc;
	gCudaContextManager = PxCreateCudaContextManager(*gFoundation, cudaContextManagerDesc, PxGetProfilerCallback());
	if (gCudaContextManager && !gCudaContextManager->contextIsValid())
	{
		PX_RELEASE(gCudaContextManager);
		printf("Failed to initialize cuda context.\n");
	}

	PxTolerancesScale scale;
	gPhysics = PxCreatePhysics(PX_PHYSICS_VERSION, *gFoundation, scale, true, gPvd);
	PxInitExtensions(*gPhysics, gPvd);

	PxCookingParams params(scale);
	params.meshWeldTolerance = 0.001f;
	params.meshPreprocessParams = PxMeshPreprocessingFlags(PxMeshPreprocessingFlag::eWELD_VERTICES);
	params.buildTriangleAdjacencies = false;
	params.buildGPUData = true;

	PxSceneDesc sceneDesc(gPhysics->getTolerancesScale());
	sceneDesc.gravity = PxVec3(0.0f, -9.81f, 0.0f);

	if (!sceneDesc.cudaContextManager)
		sceneDesc.cudaContextManager = gCudaContextManager;

	sceneDesc.flags |= PxSceneFlag::eENABLE_GPU_DYNAMICS;
	sceneDesc.flags |= PxSceneFlag::eENABLE_PCM;

	PxU32 numCores = SnippetUtils::getNbPhysicalCores();
	gDispatcher = PxDefaultCpuDispatcherCreate(numCores == 0 ? 0 : numCores - 1);
	sceneDesc.cpuDispatcher = gDispatcher;
	sceneDesc.filterShader = PxDefaultSimulationFilterShader;

	sceneDesc.broadPhaseType = PxBroadPhaseType::eGPU;
	sceneDesc.gpuMaxNumPartitions = 8;

	sceneDesc.solverType = PxSolverType::eTGS;

	gScene = gPhysics->createScene(sceneDesc);
	PxPvdSceneClient* pvdClient = gScene->getScenePvdClient();
	if (pvdClient)
	{
		pvdClient->setScenePvdFlag(PxPvdSceneFlag::eTRANSMIT_CONSTRAINTS, true);
		pvdClient->setScenePvdFlag(PxPvdSceneFlag::eTRANSMIT_CONTACTS, true);
		pvdClient->setScenePvdFlag(PxPvdSceneFlag::eTRANSMIT_SCENEQUERIES, true);
	}

	gMaterial = gPhysics->createMaterial(0.5f, 0.5f, 0.f);

	PxRigidStatic* groundPlane = PxCreatePlane(*gPhysics, PxPlane(0, 1, 0, 0), *gMaterial);
	gScene->addActor(*groundPlane);

	createDeformableVolumes(params);
}

void stepPhysics(bool /*interactive*/)
{
	const PxReal dt = 1.0f / 60.f;

	gScene->simulate(dt);
	gScene->fetchResults(true);

	for (PxU32 i = 0; i < gDeformableVolumes.size(); i++)
	{
		DeformableVolume* dv = &gDeformableVolumes[i];
		dv->copyDeformedVerticesFromGPU();
	}
}

void cleanupPhysics(bool /*interactive*/)
{
	for (PxU32 i = 0; i < gDeformableVolumes.size(); i++)
		gDeformableVolumes[i].release();
	gDeformableVolumes.reset();

	gSkinnedMeshes.reset();

	PX_RELEASE(gScene);
	PX_RELEASE(gDispatcher);
	PX_RELEASE(gPhysics);
	if (gPvd)
	{
		PxPvdTransport* transport = gPvd->getTransport();
		PX_RELEASE(gPvd);
		PX_RELEASE(transport);
	}
	PxCloseExtensions();

	PX_RELEASE(gCudaContextManager);
	PX_RELEASE(gFoundation);

	printf("SnippetDeformableVolumeSkinning done.\n");
}

int snippetMain(int, const char* const*)
{
#ifdef RENDER_SNIPPET
	extern void renderLoop();
	renderLoop();
#else
	static const PxU32 frameCount = 100;
	initPhysics(false);
	for (PxU32 i = 0; i < frameCount; i++)
		stepPhysics(false);
	cleanupPhysics(false);
#endif

	return 0;
}