阴影算法

╰+哭是因爲堅強的太久メ 2022-09-30 07:07 167阅读 0赞

![Image 1][] ![2408546_13043458116o3Z.jpg][]

图中有两条光线，一条沿视角方向，另一条偏右

采用阴影体算法，实现了多光源，多物体的阴影。

这个算法没有采用顶点shader，所以速度比较慢，学习算法用，在实际工程中用效率就太低了。。。

算法还有个缺陷，只对凸多面体有效，如果物体是凹多面体阴影就会错误。因为我不知道如何确定凹多面体的轮廓，从而构造一个封闭的阴影体。非常希望有牛人能指点一二！

主要代码如下

计算轮廓

/\* 计算轮廓（silhouette edge），采用的算法，伪代码如下 for 三角形tri in 所有三角形 如果tri是前向面 for 边e1 in 三角形tri的三条边 如果e1在轮廓列表中，从轮廓列表中移除e1 否则把e1加入轮廓列表中 默认物体是凸多面体。。。 对凹多面体，如何用算出来的轮廓构造阴影体，至今还不知道，希望有前辈能提点提点。 \*/ void ShadowVolume::Build(const Model &model, const Vector3 &lightPos) \{ m\_lightPos = Vector3(lightPos); m\_silhouetteEdges.clear(); m\_silhouettePoints.clear(); for(int j = 0; j < model.m\_indexNum; j += 3) \{ Vector3 v1 = model.m\_vbuf\[model.m\_ibuf\[j\]\]; Vector3 v2 = model.m\_vbuf\[model.m\_ibuf\[j + 1\]\]; Vector3 v3 = model.m\_vbuf\[model.m\_ibuf\[j + 2\]\]; Vector3 n = (v2 - v1).crossProduct(v3 - v1); n.normalise(); if(n.dotProduct(lightPos - v1) > EPSILON) \{ // 注意此处要减去v1，才是正确的可见面判别。否则在离物体很近的时候会出现误差 AddEdge(SilEdge(v1, v2)); AddEdge(SilEdge(v2, v3)); AddEdge(SilEdge(v3, v1)); \} \} list<SilEdge>::iterator it = m\_silhouetteEdges.begin(); m\_silhouettePoints.push\_back((\*it).edges\[0\]); m\_silhouettePoints.push\_back((\*it).edges\[1\]); m\_silhouetteEdges.erase(it); bool erased = true; // 把轮廓边缘的点提取出来，构造出多边形的点集合。下面的方法很糟糕，可是我想不到更好的办法了。。 while(m\_silhouetteEdges.size() != 0 && erased) \{ erased = false; for(it = m\_silhouetteEdges.begin();it != m\_silhouetteEdges.end(); it++) \{ SilEdge &edge = \*it; if(edge.edges\[0\] == m\_silhouettePoints\[m\_silhouettePoints.size() - 1\]) \{ if(edge.edges\[1\] != m\_silhouettePoints\[0\]) \{ m\_silhouettePoints.push\_back(edge.edges\[1\]); \} m\_silhouetteEdges.erase(it); erased = true; break; \} if(edge.edges\[1\] == m\_silhouettePoints\[m\_silhouettePoints.size() - 1\]) \{ if(edge.edges\[0\] != m\_silhouettePoints\[0\]) \{ m\_silhouettePoints.push\_back(edge.edges\[0\]); \} m\_silhouetteEdges.erase(it); erased = true; break; \} if(edge.edges\[1\] == m\_silhouettePoints\[0\] || edge.edges\[0\] == m\_silhouettePoints\[0\]) \{ m\_silhouetteEdges.erase(it); erased = true; break; \} \} \} \}

绘制阴影体

void DrawVolume() \{ m\_infinitePoints.clear(); for(size\_t i = 0; i < m\_silhouettePoints.size(); i++) \{ Vector3 a = m\_silhouettePoints\[i\] - m\_lightPos; Vector3 infinite = a; infinite.normalise(); infinite \*= 100; infinite += a; m\_infinitePoints.push\_back(infinite); \} glBegin(GL\_POLYGON); for(int i = m\_infinitePoints.size() - 1; i >= 0; i--) \{ glVertex3f(m\_infinitePoints\[i\].x, m\_infinitePoints\[i\].y, m\_infinitePoints\[i\].z); \} glEnd(); glBegin(GL\_POLYGON); for(size\_t i = 0; i < m\_silhouettePoints.size(); i++) \{ glVertex3f(m\_silhouettePoints\[i\].x, m\_silhouettePoints\[i\].y, m\_silhouettePoints\[i\].z); \} glEnd(); glBegin(GL\_QUADS); int n = m\_silhouettePoints.size(); for(int i = 0; i < n; i++) \{ glVertex3f(m\_silhouettePoints\[(i+1) % n\].x, m\_silhouettePoints\[(i+1) % n\].y, m\_silhouettePoints\[(i+1) % n\].z); glVertex3f(m\_silhouettePoints\[i\].x, m\_silhouettePoints\[i\].y, m\_silhouettePoints\[i\].z); glVertex3f(m\_infinitePoints\[i\].x, m\_infinitePoints\[i\].y, m\_infinitePoints\[i\].z); glVertex3f(m\_infinitePoints\[(i+1) % n\].x, m\_infinitePoints\[(i+1) % n\].y, m\_infinitePoints\[(i+1) % n\].z); \} glEnd(); \}

用蒙版缓存绘制阴影

// 用蒙版缓存(stencil buffer)的方法绘制阴影。 void Render() \{ // 绘制没有光线时的场景 for(size\_t k = 0; k < mLight.size(); k++) \{ glDisable(GL\_LIGHT0 + k); \} glBlendFunc(GL\_ONE, GL\_ONE); drawWall(); for(size\_t i = 0; i < mObjects.size(); i++) \{ mObjects\[i\]->Render(); \} //处理每条光线 for(size\_t k = 0; k < mLight.size(); k++) \{ glClear(GL\_STENCIL\_BUFFER\_BIT); // 计算每一个物体阴影体 for(size\_t i = 0; i < mObjects.size(); i++) \{ mObjects\[i\]->BuildShadow(mLight\[k\]->position); \} glDepthFunc(GL\_LESS); // 绘制阴影体，采用卡马克的depth fail算法 glColorMask(GL\_FALSE, GL\_FALSE, GL\_FALSE, GL\_FALSE); glDepthMask(GL\_FALSE); glEnable(GL\_CULL\_FACE); glEnable(GL\_STENCIL\_TEST); glEnable(GL\_POLYGON\_OFFSET\_FILL); // 对重叠表面的处理 glPolygonOffset(0.0f, 100.0f); glCullFace(GL\_FRONT); glStencilFunc(GL\_ALWAYS, 0x0, 0xff); glStencilOp(GL\_KEEP, GL\_INCR, GL\_KEEP); for(size\_t i = 0; i < mObjects.size(); i++) \{ mObjects\[i\]->RenderShadow(); \} glCullFace(GL\_BACK); glStencilFunc(GL\_ALWAYS, 0x0, 0xff); glStencilOp(GL\_KEEP, GL\_DECR, GL\_KEEP); for(size\_t i = 0; i < mObjects.size(); i++) \{ mObjects\[i\]->RenderShadow(); \} glDepthFunc(GL\_LEQUAL); glDisable(GL\_CULL\_FACE); glDisable(GL\_POLYGON\_OFFSET\_FILL); // 绘制整个场景，如果蒙版缓存的值为1，说明有阴影，不绘制 glColorMask(GL\_TRUE, GL\_TRUE, GL\_TRUE, GL\_TRUE); glDepthMask(GL\_TRUE); glStencilFunc(GL\_EQUAL, 0x0, 0xff); glStencilOp(GL\_KEEP, GL\_KEEP, GL\_KEEP); // 开启当前光线的光照效果 glEnable(GL\_BLEND); glBlendFunc(GL\_ONE, GL\_ONE); glEnable(GL\_LIGHT0 + k); for(size\_t i = 0; i < mObjects.size(); i++) \{ mObjects\[i\]->Render(); \} drawWall(); glDisable(GL\_LIGHT0 + k); glDisable(GL\_BLEND); \} \}

[代码打包下载][Link 1]

[Image 1]: 
[2408546_13043458116o3Z.jpg]: /images/20220708/bfb487510d5847eca1b9091d0e92054f.png
[Link 1]: http://download.csdn.net/source/3245306