3dsmax表情动画的导出

傷城~ 2022-08-06 08:27 379阅读 0赞

**作者****任晓宇**

**1** **基本概念**

1，Morpher Modifier

在3dsmax中，使用Morpher Modifier可以改变mesh, patch, NURBS model的形状，同时支持Material morphing，通常用来实现复杂的表情动画。

为Mesh添加Morpher Modifier：选择一个mesh－＞Modify面板－＞Modifier List－＞Morpher。

2，Channel

每个channel对应一个不同的Mesh，代表一种表情。

通过指定在不同关健帧时每个channel的权重，来实现不同表情的组合过渡。

**2** **准备工作**

3dsmax6 sdk并没有暴露访问morpher modifier的接口，但是我们可以在3dsmax6/maxsdk/samples/modifiers/morpher/wm3.h中找到它的定义和ClassID：

\#define MR3\_CLASS\_ID           Class\_ID(0x17bb6854, 0xa5cba2a3)

…

class MorphR3 : public Modifier, TimeChangeCallback

…

\};

将此文件include到你的工程中就可以访问morpher modifier接口了。注意，把下面两行代码注释才可以联接通过：

…

//static GetMorphMod theModPickmode;

…

//static GetMorphNode thePickMode;

**3** **实现代码**

1，得到morpher modifier

Modifier \*wxFindMorpherModifier(INode \*pINode)

\#if MAX\_RELEASE >= 4000

// get the object reference of the node

Object \*pObject;

pObject = pINode->GetObjectRef();

if(pObject == 0) return 0;

// loop through all derived objects

while(pObject->SuperClassID() == GEN\_DERIVOB\_CLASS\_ID)

IDerivedObject \*pDerivedObject;

pDerivedObject = static\_cast<IDerivedObject \*>(pObject);

// loop through all modifiers

int stackId;

for(stackId = 0; stackId < pDerivedObject->NumModifiers(); stackId++)

// get the modifier

Modifier \*pModifier;

pModifier = pDerivedObject->GetModifier(stackId);

// check if we found the morpher modifier

if(pModifier->ClassID() == MR3\_CLASS\_ID) return pModifier;

// continue with next derived object

pObject = pDerivedObject->GetObjRef();

\#endif

return 0;

2，得到每个channel相对于原始Mesh的变化顶点和偏移量：

MorphR3\*pMorpherModifier = static\_cast<MorphR3\*>( wxFindMorpherModifier(pNode));

if(pMorpherModifier != NULL )

intiNumPoses = (int)pMorpherModifier->chanBank.size();

for(int i = 0; i<iNumPoses; ++i )

morphChannel&currChannel =pMorpherModifier->chanBank\[i\];

if( !currChannel.mActive || !currChannel.mActiveOverride ||!currChannel.cblock || !currChannel.mNumPoints )

continue;

wxPose pose;

//表情的名称

memcpy( pose.chName, currChannel.mName, sizeof(char)\*64);

wxPose::wxOffsetVertex offsetVertex;

//遍历所有顶点

for( int iVertexID = 0; iVertexID<currChannel.mNumPoints;++iVertexID )

//判断是否有偏移

if(currChannel.mDeltas\[iVertexID\] == Point3(0,0,0) )

continue;

//保存顶点ID和偏移量

offsetVertex.index =iVertexID;

offsetVertex.pos =currChannel.mPoints\[iVertexID\] - m\_aVertexArray\[iVertexID\].vPos;

pose.m\_VertexOffsetMap.push\_back(offsetVertex);

3，得到channel的关健帧信息

// 权重的变化范围

float fRange = currChannel.mSpinmax - currChannel.mSpinmin;

Control \*Controller = currChannel.cblock->GetController(0);

IKeyControl \*pIKeyControl = GetKeyControlInterface(Controller);

// 关健帧数目

int iNumKeys = pIKeyControl->GetNumKeys();

for( int j = 0; j < iNumKeys; ++j )

IBezFloatKey FKey;

pIKeyControl->GetKey( j, &FKey);

// 权重

fInfluence = FKey.val / fRange;

// 时间

fTime = FKey.time;

**4** **TODO**

1，morphChannel类的数据成员mDeltas，mWeights的具体含义。发现mDeltas里面近似保存了每个顶点的偏移量乘以0.01，有何作用？mWeights按名字理解是顶点权重，不理解。

// Actualmorphable points

std::vector<Point3>              mPoints;

std::vector<Point3>              mDeltas;

std::vector<double>              mWeights;

2，求权重范围的方法。目前的方法是求SpinControl的取值范围：

float fRange = currChannel.mSpinmax - currChannel.mSpinmin;

在实验过程中发现这个范围并不准确，当SpinControl取值范围为｛0,100｝时，会得到大于100和小于0的数。

3，支持导出material morpher数据。支持material morpher可以做到材质的混合与过渡，实现更逼真的效果。比如在笑的同时脸色变红润，眼角出现皱纹等等。

4，目前我将表情动画的数据序列化到模型格式里，但是3dsmax6不能将morpher modifier编辑的结果存成单独的文件（类似于骨骼动画里的.bip文件），如果一个模型对应多个morph动画文件就不容易序列化了。不知道高版本的max有没有此功能。

5, 各个MorphTarget法向量的计算

for each face....

FaceEx \*pFaceEx = pGameMesh->GetFace(faceId);

// 得到位置  
 int vertexId = pFaceEx->vert\[faceVertexId\];  
 Point3 vertex;  
 if( pGameMesh->GetVertex(vertexId, vertex) )  
  pVertexData->m\_position = vertex;

// 得到法向量  
 int normId = pFaceEx->norm\[faceVertexId\];  
 Point3 normal;  
 if( pGameMesh->GetNormal(normId, normal) )  
  pVertexData->m\_normal = normal;

// 得到纹理坐标  
 // mapchannel代表一层纹理坐标, 如果mapchannel个数为1,则直接使用GetTexVertex()函数  
 // 否则就遍历所有的TextureMap,计算本TextureMap的纹理坐标  
 if( pGameMesh->GetActiveMapChannelNum().Count() == 1 )  
 \{  
  int texCoordId = pFaceEx->texCoord\[faceVertexId\];  
  Point2 texCoord;  
  if( pGameMesh->GetTexVertex(texCoordId, texCoord) )  
   pVertexData->AddTextureCoordinate( texCoord.x, 1.f - texCoord.y );  
 \}  
 else  
 \{  
  IGameMaterial \*pGameMaterial = pGameMesh->GetMaterialFromFace( pFaceEx );  
  int iNumTexture = pGameMaterial->GetNumberOfTextureMaps();  
  for( int i = 0; i < iNumTexture; ++i )  
  \{  
   IGameTextureMap \*pTexture = pGameMaterial->GetIGameTextureMap(i);  
   TCHAR \*tcName = pTexture->GetBitmapFileName();

int iChannel = pTexture->GetMaxTexmap()->GetMapChannel();  
   DWORD vertexId\[3\];  
   Point3 uvVert;  
   if( pGameMesh->GetMapFaceIndex(iChannel, faceId, vertexId) )  
   \{  
    if( pGameMesh->GetMapVertex(iChannel, vertexId\[faceVertexId\], uvVert) )  
    \{  
     IGameUVGen \*pGameUVGen = pTexture->GetIGameUVGen();  
     GMatrix gmUV = pGameUVGen->GetUVTransform();  
     uvVert = uvVert \* gmUV;

pVertexData->AddTextureCoordinate( uvVert.x, 1.f - uvVert.y );  
    \}  
   \}  
  \}  
 \}

// 得到skin信息  
 if( pGameMesh->IsObjectSkinned() )  
 \{  
  IGameSkin \*pGameSkin = NULL;  
  pGameSkin = pGameMesh->GetIGameSkin();

int iBoneID;  
  float fWeight;  
  int type = pGameSkin->GetVertexType(vertexId);  
  if(type==IGameSkin::IGAME\_RIGID)  
  \{  
   iBoneID = pGameSkin->GetBoneID(vertexId,0);  
   IGameNode \*pBoneNode = g\_Exporter.GetIGameScene()->GetIGameNode(ULONG(iBoneID));  
   TCHAR \*chBoneName = pBoneNode->GetName();

fWeight = 1.f;  
   pVertexData->AddInfluence(iBoneID, fWeight);  
  \}  
  else //blended  
  \{  
   for(int b=0;b<pGameSkin->GetNumberOfBones(vertexId);++b)  
   \{  
    iBoneID = pGameSkin->GetBoneID(vertexId,b);  
    IGameNode \*pBoneNode = g\_Exporter.GetIGameScene()->GetIGameNode(ULONG(iBoneID));  
    TCHAR \*chBoneName = pBoneNode->GetName();

fWeight = pGameSkin->GetWeight(vertexId,b);  
    pVertexData->AddInfluence(iBoneID, fWeight);  
   \}  
  \}  
 \}