anim_defines.h:
#ifndef ANIM_DEFINES_H_INCLUDED#define ANIM_DEFINES_H_INCLUDED#include <iostream>#include <d3d9.h>#include <d3dx9.h>#include <vector>struct AnimFrame{ D3DXMATRIX localTransMat; D3DXMATRIX boneOffset; std::vector<unsigned int> meshContainer; struct AnimFrame* child;};extern AnimFrame upperRightArmFrame;extern AnimFrame lowerRightArmFrame;#endif // ANIM_DEFINES_H_INCLUDED/Sets up the struct and declares the anim frame globals with extern to prevent linker errors. I'm only animating two limbs for now.
The present animation recursive function:
void animate_arm_mesh(AnimFrame* frame, D3DXMATRIX* parent) { unsigned int i = 0; D3DXMATRIX local = frame->localTransMat; D3DXMATRIX combined; D3DXMATRIX temp; D3DXMATRIX offsetInverse; D3DXMATRIX offset = frame->boneOffset; float doff = D3DXMatrixDeterminant(&frame->boneOffset); D3DXMatrixInverse(&offsetInverse, &doff, &frame->boneOffset); /////////////////////////////////////////////////////////////////////////// D3DXMatrixMultiply(&temp, &offsetInverse, &local); D3DXMatrixMultiply(&temp, &temp, &offset); D3DXMatrixMultiply(&combined, &temp, parent); /////////////////////////////////////////////////////////////////////////// for(i = 0; i < frame->meshContainer.size(); i++) { currentPos.x = vertices[frame->meshContainer].X; currentPos.y = vertices[frame->meshContainer].Y; currentPos.z = vertices[frame->meshContainer].Z; currentNormals.x = vertices[frame->meshContainer].NORMAL.x; currentNormals.y = vertices[frame->meshContainer].NORMAL.y; currentNormals.z = vertices[frame->meshContainer].NORMAL.z; D3DXVec3TransformCoord(&newPos, ¤tPos, &combined); D3DXVec3TransformNormal(&newNormals, ¤tNormals, &combined); newVertices[frame->meshContainer].X = newPos.x; newVertices[frame->meshContainer].Y = newPos.y; newVertices[frame->meshContainer].Z = newPos.z; newVertices[frame->meshContainer].NORMAL.x = newNormals.x; newVertices[frame->meshContainer].NORMAL.y = newNormals.y; newVertices[frame->meshContainer].NORMAL.z = newNormals.z; } /////////////////////////////////////////////////////////////////////////// AnimFrame* child = frame->child; if(frame->child != NULL) { animate_arm_mesh(child, &combined); } /////////////////////////////////////////////////////////////////////////// VOID* pVoid; v_buffer->Lock(0, 0, (void**)&pVoid, 0); memcpy(pVoid, &newVertices[0], newVertices.size() * sizeof(CUSTOMVERTEX)); v_buffer->Unlock(); return; }This loops through each limb starting at the designated parent. Performs arbitrary axis transformations (rotations only in this case) and updates the mesh container all in one. I know copying verts is bad but I've yet to implement a HLSL version of this that's for the future.
The code that calls it all:
soldier.load_soldier_hierachy(); soldier.set_poses(); D3DXMATRIX identity; D3DXMatrixIdentity(&identity); soldier.animate_arm_mesh(&upperRightArmFrame, &identity);And the set poses function:
void set_poses(void) { D3DXMATRIX matRotate1; D3DXMATRIX matRotate2; D3DXVECTOR3 axis; axis.x = 0.0f; axis.y = 0.0f; axis.z = 1.0f; D3DXMatrixRotationAxis(&matRotate1, &axis, 0.4); upperRightArmFrame.localTransMat = matRotate1; axis.x = 1.0f; axis.y = -1.0f; axis.z = 0.0f; D3DXMatrixRotationAxis(&matRotate1, &axis, -1.571); axis.x = 0.0f; axis.y = 0.0f; axis.z =1.0f; D3DXMatrixRotationAxis(&matRotate2, &axis, 1.571); D3DXMatrixMultiply(&matRotate2, &matRotate1, &matRotate2); lowerRightArmFrame.localTransMat = matRotate2; return; }Now I don't expect *anyone* to gleam a damn thing from my source code as it isn't well explained enough in this topic but I reckon you can get a *feel* for what I'm doing.
Well I have two main questions:
1)Is what I've done above carped in some way or is it ok (for a rough start)? It produces proper results - but that's not the point I'm worried about efficiency and structuring.
2)How do I go about arranging all this professionally? Ok so I can animate. Big deal. What I want to know is how is it best to set-up your complete animation layout. Will I need a seperate class for controlling animations? I was thinking of one header file (or cpp file) with all the required poses pre-calced. An animation controller class could then acquire the current pose and slerp it to the next one. Is this a good idea?
Pretty vague topic really sorry. I just want some input on where to go from here. I can provide full source and resources to anyone interested through pm - use at your own risk! ;o) Thanks, *any* input appreciated.
