Scarfski/assimp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

General

Browse Source 
- changing Assimp's coordinate system from RH z-up to RH y-up
 - fixing coordinate system for LWO, 3DS, ASE, MD5, MDL, B3D, IRR, IRRMESH 
 - converttolh moved to three separate steps -> flipuv, flipwinding, makelh

LWO 
 - fixing texture coordinate generation -> mapping axis is correct now 
 - fixing z-fighting bug

ASE 
 - fixing crash due to invalid normal setup 
 - fixing parenting bug 
 - code cleanup

IRR 
 - code cleanup
 - fixing placement of externally loaded meshes 

MDL 
 - fixing texture coordinate space

PLY 
 - cleanup 
 - two-sided maat property is now set 

git-svn-id: https://assimp.svn.sourceforge.net/svnroot/assimp/trunk@366 67173fc5-114c-0410-ac8e-9d2fd5bffc1f
This commit is contained in:
aramis_acg
2009-03-15 00:40:30 +00:00
parent 68d7e43056
commit c2d8881549
48 changed files with 1518 additions and 1802 deletions
Download Patch File Download Diff File Expand all files Collapse all files

380
code/ConvertToLHProcess.cpp
View File

@@ -39,136 +39,220 @@ OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
---------------------------------------------------------------------------
*/
/** @file Implementation of the post processing step to convert all imported data
* to a left-handed coordinate system.
/** @file MakeLeftHandedProcess.cpp
* @brief Implementation of the post processing step to convert all
* imported data to a left-handed coordinate system.
*
* Face order & UV flip are also implemented here, for the sake of a
* better location.
*/
#include "AssimpPCH.h"
#include "ConvertToLHProcess.h"
using namespace Assimp;
// The transformation matrix to convert from DirectX coordinates to OpenGL coordinates.
const aiMatrix3x3 Assimp::ConvertToLHProcess::sToOGLTransform(
1.0f, 0.0f, 0.0f, 0.0f, 0.0f, 1.0f, 0.0f, 1.0f, 0.0f
);
// The transformation matrix to convert from OpenGL coordinates to DirectX coordinates.
const aiMatrix3x3 Assimp::ConvertToLHProcess::sToDXTransform(
1.0f, 0.0f, 0.0f, 0.0f, 0.0f, 1.0f, 0.0f, 1.0f, 0.0f
);
#ifndef ASSIMP_BUILD_NO_MAKELEFTHANDED_PROCESS
// ------------------------------------------------------------------------------------------------
// Constructor to be privately used by Importer
ConvertToLHProcess::ConvertToLHProcess()
{
bTransformVertices = false;
}
MakeLeftHandedProcess::MakeLeftHandedProcess()
{}
// ------------------------------------------------------------------------------------------------
// Destructor, private as well
ConvertToLHProcess::~ConvertToLHProcess()
{
// nothing to do here
}
MakeLeftHandedProcess::~MakeLeftHandedProcess()
{}
// ------------------------------------------------------------------------------------------------
// Returns whether the processing step is present in the given flag field.
bool ConvertToLHProcess::IsActive( unsigned int pFlags) const
bool MakeLeftHandedProcess::IsActive( unsigned int pFlags) const
{
if (pFlags & aiProcess_ConvertToLeftHanded)
{
bTransformVertices = (0 != (pFlags & aiProcess_PreTransformVertices) ? true : false);
return true;
}
return false;
return 0 != (pFlags & aiProcess_MakeLeftHanded);
}
// ------------------------------------------------------------------------------------------------
// Executes the post processing step on the given imported data.
void ConvertToLHProcess::Execute( aiScene* pScene)
void MakeLeftHandedProcess::Execute( aiScene* pScene)
{
// Check for an existent root node to proceed
if (NULL == pScene->mRootNode)
{
DefaultLogger::get()->error("ConvertToLHProcess fails, there is no root node");
return;
}
ai_assert(pScene->mRootNode != NULL);
DefaultLogger::get()->debug("MakeLeftHandedProcess begin");
DefaultLogger::get()->debug("ConvertToLHProcess begin");
// recursively convert all the nodes
ProcessNode( pScene->mRootNode, aiMatrix4x4());
// transform vertex by vertex or change the root transform?
// We can't do the coordinate system transformation earlier
// in the pipeline - most steps assume that we're in OGL
// space. So we need to transform all vertices a second time
// here.
if (bTransformVertices)
{
aiMatrix4x4 mTransform;
ConvertToDX(mTransform);
for (unsigned int i = 0; i < pScene->mNumMeshes;++i)
{
aiMesh* pcMesh = pScene->mMeshes[i];
// transform all vertices
for (unsigned int n = 0; n < pcMesh->mNumVertices;++n)
pcMesh->mVertices[n] = mTransform * pcMesh->mVertices[n];
// transform all normals
if (pcMesh->HasNormals())
{
for (unsigned int n = 0; n < pcMesh->mNumVertices;++n)
pcMesh->mNormals[n] = mTransform * pcMesh->mNormals[n];
}
// transform all tangents and all bitangents
if (pcMesh->HasTangentsAndBitangents())
{
for (unsigned int n = 0; n < pcMesh->mNumVertices;++n)
{
pcMesh->mTangents[n] = mTransform * pcMesh->mTangents[n];
pcMesh->mBitangents[n] = mTransform * pcMesh->mBitangents[n];
}
}
}
}
else
{
// transform the root node of the scene, the other nodes will follow then
ConvertToDX( pScene->mRootNode->mTransformation);
}
// transform all meshes accordingly
for( unsigned int a = 0; a < pScene->mNumMeshes; a++)
// process the meshes accordingly
for( unsigned int a = 0; a < pScene->mNumMeshes; ++a)
ProcessMesh( pScene->mMeshes[a]);
// process all materials - we need to adjust UV transformations
for( unsigned int a = 0; a < pScene->mNumMaterials; a++)
// process the materials accordingly
for( unsigned int a = 0; a < pScene->mNumMaterials; ++a)
ProcessMaterial( pScene->mMaterials[a]);
// transform all animation channels affecting the root node as well
// transform all animation channels as well
for( unsigned int a = 0; a < pScene->mNumAnimations; a++)
{
aiAnimation* anim = pScene->mAnimations[a];
for( unsigned int b = 0; b < anim->mNumChannels; b++)
{
aiNodeAnim* nodeAnim = anim->mChannels[b];
if( strcmp( nodeAnim->mNodeName.data, pScene->mRootNode->mName.data) == 0)
ProcessAnimation( nodeAnim);
ProcessAnimation( nodeAnim);
}
}
DefaultLogger::get()->debug("ConvertToLHProcess finished");
// flipping a single vector component means inverting face order ...
FlipWindingOrderProcess flipper;
flipper.Execute(pScene);
DefaultLogger::get()->debug("MakeLeftHandedProcess finished");
}
// ------------------------------------------------------------------------------------------------
// Recursively converts a node, all of its children and all of its meshes
void MakeLeftHandedProcess::ProcessNode( aiNode* pNode, const aiMatrix4x4& pParentGlobalRotation)
{
// mirror all base vectors at the local Z axis
pNode->mTransformation.c1 = -pNode->mTransformation.c1;
pNode->mTransformation.c2 = -pNode->mTransformation.c2;
pNode->mTransformation.c3 = -pNode->mTransformation.c3;
pNode->mTransformation.c4 = -pNode->mTransformation.c4;
// now invert the Z axis again to keep the matrix determinant positive.
// The local meshes will be inverted accordingly so that the result should look just fine again.
pNode->mTransformation.a3 = -pNode->mTransformation.a3;
pNode->mTransformation.b3 = -pNode->mTransformation.b3;
pNode->mTransformation.c3 = -pNode->mTransformation.c3;
pNode->mTransformation.d3 = -pNode->mTransformation.d3; // useless, but anyways...
// continue for all children
for( size_t a = 0; a < pNode->mNumChildren; ++a)
ProcessNode( pNode->mChildren[a], pParentGlobalRotation * pNode->mTransformation);
}
// ------------------------------------------------------------------------------------------------
// Converts a single mesh to left handed coordinates.
void MakeLeftHandedProcess::ProcessMesh( aiMesh* pMesh)
{
// mirror positions, normals and stuff along the Z axis
for( size_t a = 0; a < pMesh->mNumVertices; ++a)
{
pMesh->mVertices[a].z *= -1.0f;
if( pMesh->HasNormals())
pMesh->mNormals[a].z *= -1.0f;
if( pMesh->HasTangentsAndBitangents())
{
pMesh->mTangents[a].z *= -1.0f;
pMesh->mBitangents[a].z *= -1.0f;
}
}
// mirror offset matrices of all bones
for( size_t a = 0; a < pMesh->mNumBones; ++a)
{
aiBone* bone = pMesh->mBones[a];
bone->mOffsetMatrix.a3 = -bone->mOffsetMatrix.a3;
bone->mOffsetMatrix.b3 = -bone->mOffsetMatrix.b3;
bone->mOffsetMatrix.d3 = -bone->mOffsetMatrix.d3;
bone->mOffsetMatrix.c1 = -bone->mOffsetMatrix.c1;
bone->mOffsetMatrix.c2 = -bone->mOffsetMatrix.c2;
bone->mOffsetMatrix.c4 = -bone->mOffsetMatrix.c4;
}
// mirror bitangents as well as they're derived from the texture coords
if( pMesh->HasTangentsAndBitangents())
{
for( unsigned int a = 0; a < pMesh->mNumVertices; a++)
pMesh->mBitangents[a] *= -1.0f;
}
}
// ------------------------------------------------------------------------------------------------
// Converts a single material to left handed coordinates.
void ConvertToLHProcess::ProcessMaterial (aiMaterial* mat)
void MakeLeftHandedProcess::ProcessMaterial( aiMaterial* _mat)
{
for (unsigned int a = 0; a < mat->mNumProperties;++a)
{
MaterialHelper* mat = (MaterialHelper*)_mat;
for (unsigned int a = 0; a < mat->mNumProperties;++a) {
aiMaterialProperty* prop = mat->mProperties[a];
if (!::strcmp( prop->mKey.data, "$tex.uvtrafo"))
{
ai_assert( prop->mDataLength >= sizeof(aiUVTransform));
// Mapping axis for UV mappings?
if (!::strcmp( prop->mKey.data, "$tex.mapaxis")) {
ai_assert( prop->mDataLength >= sizeof(aiVector3D)); /* something is wrong with the validation if we end up here */
aiVector3D* pff = (aiVector3D*)prop->mData;
pff->z *= -1.f;
}
}
}
// ------------------------------------------------------------------------------------------------
// Converts the given animation to LH coordinates.
void MakeLeftHandedProcess::ProcessAnimation( aiNodeAnim* pAnim)
{
// position keys
for( unsigned int a = 0; a < pAnim->mNumPositionKeys; a++)
pAnim->mPositionKeys[a].mValue.z *= -1.0f;
// rotation keys
for( unsigned int a = 0; a < pAnim->mNumRotationKeys; a++)
{
/* That's the safe version, but the float errors add up. So we try the short version instead
aiMatrix3x3 rotmat = pAnim->mRotationKeys[a].mValue.GetMatrix();
rotmat.a3 = -rotmat.a3; rotmat.b3 = -rotmat.b3;
rotmat.c1 = -rotmat.c1; rotmat.c2 = -rotmat.c2;
aiQuaternion rotquat( rotmat);
pAnim->mRotationKeys[a].mValue = rotquat;
*/
pAnim->mRotationKeys[a].mValue.x *= -1.0f;
pAnim->mRotationKeys[a].mValue.y *= -1.0f;
}
}
#endif // !! ASSIMP_BUILD_NO_MAKELEFTHANDED_PROCESS
#ifndef ASSIMP_BUILD_NO_FLIPUVS_PROCESS
// # FlipUVsProcess
// ------------------------------------------------------------------------------------------------
// Constructor to be privately used by Importer
FlipUVsProcess::FlipUVsProcess()
{}
// ------------------------------------------------------------------------------------------------
// Destructor, private as well
FlipUVsProcess::~FlipUVsProcess()
{}
// ------------------------------------------------------------------------------------------------
// Returns whether the processing step is present in the given flag field.
bool FlipUVsProcess::IsActive( unsigned int pFlags) const
{
return 0 != (pFlags & aiProcess_FlipUVs);
}
// ------------------------------------------------------------------------------------------------
// Executes the post processing step on the given imported data.
void FlipUVsProcess::Execute( aiScene* pScene)
{
DefaultLogger::get()->debug("FlipUVsProcess begin");
for (unsigned int i = 0; i < pScene->mNumMeshes;++i)
ProcessMesh(pScene->mMeshes[i]);
for (unsigned int i = 0; i < pScene->mNumMaterials;++i)
ProcessMaterial(pScene->mMaterials[i]);
DefaultLogger::get()->debug("FlipUVsProcess finished");
}
// ------------------------------------------------------------------------------------------------
// Converts a single material
void FlipUVsProcess::ProcessMaterial (aiMaterial* _mat)
{
MaterialHelper* mat = (MaterialHelper*)_mat;
for (unsigned int a = 0; a < mat->mNumProperties;++a) {
aiMaterialProperty* prop = mat->mProperties[a];
// UV transformation key?
if (!::strcmp( prop->mKey.data, "$tex.uvtrafo")) {
ai_assert( prop->mDataLength >= sizeof(aiUVTransform)); /* something is wrong with the validation if we end up here */
aiUVTransform* uv = (aiUVTransform*)prop->mData;
// just flip it, that's everything
@@ -179,8 +263,53 @@ void ConvertToLHProcess::ProcessMaterial (aiMaterial* mat)
}
// ------------------------------------------------------------------------------------------------
// Converts a single mesh to left handed coordinates.
void ConvertToLHProcess::ProcessMesh( aiMesh* pMesh)
// Converts a single mesh
void FlipUVsProcess::ProcessMesh( aiMesh* pMesh)
{
// mirror texture y coordinate
for( unsigned int a = 0; a < AI_MAX_NUMBER_OF_TEXTURECOORDS; a++) {
if( !pMesh->HasTextureCoords( a))
break;
for( unsigned int b = 0; b < pMesh->mNumVertices; b++)
pMesh->mTextureCoords[a][b].y = 1.0f - pMesh->mTextureCoords[a][b].y;
}
}
#endif // !ASSIMP_BUILD_NO_FLIPUVS_PROCESS
#ifndef ASSIMP_BUILD_NO_FLIPWINDING_PROCESS
// # FlipWindingOrderProcess
// ------------------------------------------------------------------------------------------------
// Constructor to be privately used by Importer
FlipWindingOrderProcess::FlipWindingOrderProcess()
{}
// ------------------------------------------------------------------------------------------------
// Destructor, private as well
FlipWindingOrderProcess::~FlipWindingOrderProcess()
{}
// ------------------------------------------------------------------------------------------------
// Returns whether the processing step is present in the given flag field.
bool FlipWindingOrderProcess::IsActive( unsigned int pFlags) const
{
return 0 != (pFlags & aiProcess_FlipWindingOrder);
}
// ------------------------------------------------------------------------------------------------
// Executes the post processing step on the given imported data.
void FlipWindingOrderProcess::Execute( aiScene* pScene)
{
DefaultLogger::get()->debug("FlipWindingOrderProcess begin");
for (unsigned int i = 0; i < pScene->mNumMeshes;++i)
ProcessMesh(pScene->mMeshes[i]);
DefaultLogger::get()->debug("FlipWindingOrderProcess finished");
}
// ------------------------------------------------------------------------------------------------
// Converts a single mesh
void FlipWindingOrderProcess::ProcessMesh( aiMesh* pMesh)
{
// invert the order of all faces in this mesh
for( unsigned int a = 0; a < pMesh->mNumFaces; a++)
@@ -189,77 +318,6 @@ void ConvertToLHProcess::ProcessMesh( aiMesh* pMesh)
for( unsigned int b = 0; b < face.mNumIndices / 2; b++)
std::swap( face.mIndices[b], face.mIndices[ face.mNumIndices - 1 - b]);
}
// mirror texture y coordinate
for( unsigned int a = 0; a < AI_MAX_NUMBER_OF_TEXTURECOORDS; a++)
{
if( pMesh->HasTextureCoords( a))
{
for( unsigned int b = 0; b < pMesh->mNumVertices; b++)
pMesh->mTextureCoords[a][b].y = 1.0f - pMesh->mTextureCoords[a][b].y;
}
}
// mirror bitangents as well as they're derived from the texture coords
if( pMesh->HasTangentsAndBitangents())
{
for( unsigned int a = 0; a < pMesh->mNumVertices; a++)
pMesh->mBitangents[a] = -pMesh->mBitangents[a];
}
}
// ------------------------------------------------------------------------------------------------
// Converts the given animation to LH coordinates.
void ConvertToLHProcess::ProcessAnimation( aiNodeAnim* pAnim)
{
// position keys
for( unsigned int a = 0; a < pAnim->mNumPositionKeys; a++)
ConvertToDX( pAnim->mPositionKeys[a].mValue);
return;
// rotation keys
for( unsigned int a = 0; a < pAnim->mNumRotationKeys; a++)
{
aiMatrix3x3 rotmat = pAnim->mRotationKeys[a].mValue.GetMatrix();
ConvertToDX( rotmat);
pAnim->mRotationKeys[a].mValue = aiQuaternion( rotmat);
}
}
// ------------------------------------------------------------------------------------------------
// Static helper function to convert a vector/matrix from DX to OGL coords
void ConvertToLHProcess::ConvertToOGL( aiVector3D& poVector)
{
poVector = sToOGLTransform * poVector;
}
// ------------------------------------------------------------------------------------------------
void ConvertToLHProcess::ConvertToOGL( aiMatrix3x3& poMatrix)
{
poMatrix = sToOGLTransform * poMatrix;
}
// ------------------------------------------------------------------------------------------------
void ConvertToLHProcess::ConvertToOGL( aiMatrix4x4& poMatrix)
{
poMatrix = aiMatrix4x4( sToOGLTransform) * poMatrix;
}
// ------------------------------------------------------------------------------------------------
// Static helper function to convert a vector/matrix from OGL back to DX coords
void ConvertToLHProcess::ConvertToDX( aiVector3D& poVector)
{
poVector = sToDXTransform * poVector;
}
// ------------------------------------------------------------------------------------------------
void ConvertToLHProcess::ConvertToDX( aiMatrix3x3& poMatrix)
{
poMatrix = sToDXTransform * poMatrix;
}
// ------------------------------------------------------------------------------------------------
void ConvertToLHProcess::ConvertToDX( aiMatrix4x4& poMatrix)
{
poMatrix = aiMatrix4x4(sToDXTransform) * poMatrix;
}
#endif // !! ASSIMP_BUILD_NO_FLIPWINDING_PROCESS