359 lines
12 KiB
C++
359 lines
12 KiB
C++
/*
|
|
---------------------------------------------------------------------------
|
|
Open Asset Import Library (assimp)
|
|
---------------------------------------------------------------------------
|
|
|
|
Copyright (c) 2006-2021, assimp team
|
|
|
|
All rights reserved.
|
|
|
|
Redistribution and use of this software 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 the assimp team, nor the names of its
|
|
contributors may be used to endorse or promote products
|
|
derived from this software without specific prior
|
|
written permission of the assimp team.
|
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
"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.
|
|
---------------------------------------------------------------------------
|
|
*/
|
|
|
|
/** @file OptimizeGraph.cpp
|
|
* @brief Implementation of the aiProcess_OptimizGraph step
|
|
*/
|
|
|
|
#ifndef ASSIMP_BUILD_NO_OPTIMIZEGRAPH_PROCESS
|
|
|
|
#include "OptimizeGraph.h"
|
|
#include "ProcessHelper.h"
|
|
#include "ConvertToLHProcess.h"
|
|
#include <assimp/Exceptional.h>
|
|
#include <assimp/SceneCombiner.h>
|
|
#include <stdio.h>
|
|
|
|
using namespace Assimp;
|
|
|
|
#define AI_RESERVED_NODE_NAME "$Reserved_And_Evil"
|
|
|
|
/* AI_OG_USE_HASHING enables the use of hashing to speed-up std::set lookups.
|
|
* The unhashed variant should be faster, except for *very* large data sets
|
|
*/
|
|
#ifdef AI_OG_USE_HASHING
|
|
// Use our standard hashing function to compute the hash
|
|
#define AI_OG_GETKEY(str) SuperFastHash(str.data, str.length)
|
|
#else
|
|
// Otherwise hope that std::string will utilize a static buffer
|
|
// for shorter node names. This would avoid endless heap copying.
|
|
#define AI_OG_GETKEY(str) std::string(str.data)
|
|
#endif
|
|
|
|
// ------------------------------------------------------------------------------------------------
|
|
// Constructor to be privately used by Importer
|
|
OptimizeGraphProcess::OptimizeGraphProcess() :
|
|
mScene(),
|
|
nodes_in(),
|
|
nodes_out(),
|
|
count_merged() {
|
|
// empty
|
|
}
|
|
|
|
// ------------------------------------------------------------------------------------------------
|
|
// Destructor, private as well
|
|
OptimizeGraphProcess::~OptimizeGraphProcess() {
|
|
// empty
|
|
}
|
|
|
|
// ------------------------------------------------------------------------------------------------
|
|
// Returns whether the processing step is present in the given flag field.
|
|
bool OptimizeGraphProcess::IsActive(unsigned int pFlags) const {
|
|
return (0 != (pFlags & aiProcess_OptimizeGraph));
|
|
}
|
|
|
|
// ------------------------------------------------------------------------------------------------
|
|
// Setup properties for the post-processing step
|
|
void OptimizeGraphProcess::SetupProperties(const Importer *pImp) {
|
|
// Get value of AI_CONFIG_PP_OG_EXCLUDE_LIST
|
|
std::string tmp = pImp->GetPropertyString(AI_CONFIG_PP_OG_EXCLUDE_LIST, "");
|
|
AddLockedNodeList(tmp);
|
|
}
|
|
|
|
// ------------------------------------------------------------------------------------------------
|
|
// Collect new children
|
|
void OptimizeGraphProcess::CollectNewChildren(aiNode *nd, std::list<aiNode *> &nodes) {
|
|
nodes_in += nd->mNumChildren;
|
|
|
|
// Process children
|
|
std::list<aiNode *> child_nodes;
|
|
for (unsigned int i = 0; i < nd->mNumChildren; ++i) {
|
|
CollectNewChildren(nd->mChildren[i], child_nodes);
|
|
nd->mChildren[i] = nullptr;
|
|
}
|
|
|
|
// Check whether we need this node; if not we can replace it by our own children (warn, danger of incest).
|
|
if (locked.find(AI_OG_GETKEY(nd->mName)) == locked.end()) {
|
|
for (std::list<aiNode *>::iterator it = child_nodes.begin(); it != child_nodes.end();) {
|
|
|
|
if (locked.find(AI_OG_GETKEY((*it)->mName)) == locked.end()) {
|
|
(*it)->mTransformation = nd->mTransformation * (*it)->mTransformation;
|
|
nodes.push_back(*it);
|
|
|
|
it = child_nodes.erase(it);
|
|
continue;
|
|
}
|
|
++it;
|
|
}
|
|
|
|
if (nd->mNumMeshes || !child_nodes.empty()) {
|
|
nodes.push_back(nd);
|
|
} else {
|
|
delete nd; /* bye, node */
|
|
return;
|
|
}
|
|
} else {
|
|
|
|
// Retain our current position in the hierarchy
|
|
nodes.push_back(nd);
|
|
|
|
// Now check for possible optimizations in our list of child nodes. join as many as possible
|
|
aiNode *join_master = nullptr;
|
|
aiMatrix4x4 inv;
|
|
|
|
const LockedSetType::const_iterator end = locked.end();
|
|
|
|
std::list<aiNode *> join;
|
|
for (std::list<aiNode *>::iterator it = child_nodes.begin(); it != child_nodes.end();) {
|
|
aiNode *child = *it;
|
|
if (child->mNumChildren == 0 && locked.find(AI_OG_GETKEY(child->mName)) == end) {
|
|
|
|
// There may be no instanced meshes
|
|
unsigned int n = 0;
|
|
for (; n < child->mNumMeshes; ++n) {
|
|
if (meshes[child->mMeshes[n]] > 1) {
|
|
break;
|
|
}
|
|
}
|
|
if (n == child->mNumMeshes) {
|
|
if (!join_master) {
|
|
join_master = child;
|
|
inv = join_master->mTransformation;
|
|
inv.Inverse();
|
|
} else {
|
|
child->mTransformation = inv * child->mTransformation;
|
|
|
|
join.push_back(child);
|
|
it = child_nodes.erase(it);
|
|
continue;
|
|
}
|
|
}
|
|
}
|
|
++it;
|
|
}
|
|
if (join_master && !join.empty()) {
|
|
join_master->mName.length = ::ai_snprintf(join_master->mName.data, MAXLEN, "$MergedNode_%i", count_merged++);
|
|
|
|
unsigned int out_meshes = 0;
|
|
for (std::list<aiNode *>::const_iterator it = join.cbegin(); it != join.cend(); ++it) {
|
|
out_meshes += (*it)->mNumMeshes;
|
|
}
|
|
|
|
// copy all mesh references in one array
|
|
if (out_meshes) {
|
|
unsigned int *meshIdxs = new unsigned int[out_meshes + join_master->mNumMeshes], *tmp = meshIdxs;
|
|
for (unsigned int n = 0; n < join_master->mNumMeshes; ++n) {
|
|
*tmp++ = join_master->mMeshes[n];
|
|
}
|
|
|
|
for (const aiNode *join_node : join) {
|
|
for (unsigned int n = 0; n < join_node->mNumMeshes; ++n) {
|
|
|
|
*tmp = join_node->mMeshes[n];
|
|
aiMesh *mesh = mScene->mMeshes[*tmp++];
|
|
|
|
// Assume the transformation is affine
|
|
// manually move the mesh into the right coordinate system
|
|
|
|
// Check for odd negative scale (mirror)
|
|
if (join_node->mTransformation.Determinant() < 0) {
|
|
// Reverse the mesh face winding order
|
|
FlipWindingOrderProcess::ProcessMesh(mesh);
|
|
}
|
|
|
|
// Update positions, normals and tangents
|
|
const aiMatrix3x3 IT = aiMatrix3x3(join_node->mTransformation).Inverse().Transpose();
|
|
for (unsigned int a = 0; a < mesh->mNumVertices; ++a) {
|
|
|
|
mesh->mVertices[a] *= join_node->mTransformation;
|
|
|
|
if (mesh->HasNormals())
|
|
mesh->mNormals[a] *= IT;
|
|
|
|
if (mesh->HasTangentsAndBitangents()) {
|
|
mesh->mTangents[a] *= IT;
|
|
mesh->mBitangents[a] *= IT;
|
|
}
|
|
}
|
|
}
|
|
delete join_node; // bye, node
|
|
}
|
|
delete[] join_master->mMeshes;
|
|
join_master->mMeshes = meshIdxs;
|
|
join_master->mNumMeshes += out_meshes;
|
|
}
|
|
}
|
|
}
|
|
// reassign children if something changed
|
|
if (child_nodes.empty() || child_nodes.size() > nd->mNumChildren) {
|
|
|
|
delete[] nd->mChildren;
|
|
|
|
if (!child_nodes.empty()) {
|
|
nd->mChildren = new aiNode *[child_nodes.size()];
|
|
} else
|
|
nd->mChildren = nullptr;
|
|
}
|
|
|
|
nd->mNumChildren = static_cast<unsigned int>(child_nodes.size());
|
|
|
|
if (nd->mChildren) {
|
|
aiNode **tmp = nd->mChildren;
|
|
for (std::list<aiNode *>::iterator it = child_nodes.begin(); it != child_nodes.end(); ++it) {
|
|
aiNode *node = *tmp++ = *it;
|
|
node->mParent = nd;
|
|
}
|
|
}
|
|
|
|
nodes_out += static_cast<unsigned int>(child_nodes.size());
|
|
}
|
|
|
|
// ------------------------------------------------------------------------------------------------
|
|
// Execute the post-processing step on the given scene
|
|
void OptimizeGraphProcess::Execute(aiScene *pScene) {
|
|
ASSIMP_LOG_DEBUG("OptimizeGraphProcess begin");
|
|
nodes_in = nodes_out = count_merged = 0;
|
|
mScene = pScene;
|
|
|
|
meshes.resize(pScene->mNumMeshes, 0);
|
|
FindInstancedMeshes(pScene->mRootNode);
|
|
|
|
// build a blacklist of identifiers. If the name of a node matches one of these, we won't touch it
|
|
locked.clear();
|
|
for (std::list<std::string>::const_iterator it = locked_nodes.begin(); it != locked_nodes.end(); ++it) {
|
|
#ifdef AI_OG_USE_HASHING
|
|
locked.insert(SuperFastHash((*it).c_str()));
|
|
#else
|
|
locked.insert(*it);
|
|
#endif
|
|
}
|
|
|
|
for (unsigned int i = 0; i < pScene->mNumAnimations; ++i) {
|
|
for (unsigned int a = 0; a < pScene->mAnimations[i]->mNumChannels; ++a) {
|
|
aiNodeAnim *anim = pScene->mAnimations[i]->mChannels[a];
|
|
locked.insert(AI_OG_GETKEY(anim->mNodeName));
|
|
}
|
|
}
|
|
|
|
for (unsigned int i = 0; i < pScene->mNumMeshes; ++i) {
|
|
for (unsigned int a = 0; a < pScene->mMeshes[i]->mNumBones; ++a) {
|
|
|
|
aiBone *bone = pScene->mMeshes[i]->mBones[a];
|
|
locked.insert(AI_OG_GETKEY(bone->mName));
|
|
|
|
// HACK: Meshes referencing bones may not be transformed; we need to look them.
|
|
// The easiest way to do this is to increase their reference counters ...
|
|
meshes[i] += 2;
|
|
}
|
|
}
|
|
|
|
for (unsigned int i = 0; i < pScene->mNumCameras; ++i) {
|
|
aiCamera *cam = pScene->mCameras[i];
|
|
locked.insert(AI_OG_GETKEY(cam->mName));
|
|
}
|
|
|
|
for (unsigned int i = 0; i < pScene->mNumLights; ++i) {
|
|
aiLight *lgh = pScene->mLights[i];
|
|
locked.insert(AI_OG_GETKEY(lgh->mName));
|
|
}
|
|
|
|
// Insert a dummy master node and make it read-only
|
|
aiNode *dummy_root = new aiNode(AI_RESERVED_NODE_NAME);
|
|
locked.insert(AI_OG_GETKEY(dummy_root->mName));
|
|
|
|
const aiString prev = pScene->mRootNode->mName;
|
|
pScene->mRootNode->mParent = dummy_root;
|
|
|
|
dummy_root->mChildren = new aiNode *[dummy_root->mNumChildren = 1];
|
|
dummy_root->mChildren[0] = pScene->mRootNode;
|
|
|
|
// Do our recursive processing of scenegraph nodes. For each node collect
|
|
// a fully new list of children and allow their children to place themselves
|
|
// on the same hierarchy layer as their parents.
|
|
std::list<aiNode *> nodes;
|
|
CollectNewChildren(dummy_root, nodes);
|
|
|
|
ai_assert(nodes.size() == 1);
|
|
|
|
if (dummy_root->mNumChildren == 0) {
|
|
pScene->mRootNode = nullptr;
|
|
throw DeadlyImportError("After optimizing the scene graph, no data remains");
|
|
}
|
|
|
|
if (dummy_root->mNumChildren > 1) {
|
|
pScene->mRootNode = dummy_root;
|
|
|
|
// Keep the dummy node but assign the name of the old root node to it
|
|
pScene->mRootNode->mName = prev;
|
|
} else {
|
|
|
|
// Remove the dummy root node again.
|
|
pScene->mRootNode = dummy_root->mChildren[0];
|
|
|
|
dummy_root->mChildren[0] = nullptr;
|
|
delete dummy_root;
|
|
}
|
|
|
|
pScene->mRootNode->mParent = nullptr;
|
|
if (!DefaultLogger::isNullLogger()) {
|
|
if (nodes_in != nodes_out) {
|
|
ASSIMP_LOG_INFO("OptimizeGraphProcess finished; Input nodes: ", nodes_in, ", Output nodes: ", nodes_out);
|
|
} else {
|
|
ASSIMP_LOG_DEBUG("OptimizeGraphProcess finished");
|
|
}
|
|
}
|
|
meshes.clear();
|
|
locked.clear();
|
|
}
|
|
|
|
// ------------------------------------------------------------------------------------------------
|
|
// Build a LUT of all instanced meshes
|
|
void OptimizeGraphProcess::FindInstancedMeshes(aiNode *pNode) {
|
|
for (unsigned int i = 0; i < pNode->mNumMeshes; ++i) {
|
|
++meshes[pNode->mMeshes[i]];
|
|
}
|
|
|
|
for (unsigned int i = 0; i < pNode->mNumChildren; ++i)
|
|
FindInstancedMeshes(pNode->mChildren[i]);
|
|
}
|
|
|
|
#endif // !! ASSIMP_BUILD_NO_OPTIMIZEGRAPH_PROCESS
|