CryEngine 渲染流程

1. 渲染涉及的线程： MainThread、RenderThread、CullingThread和JobThread。 
   MainThread： 用于控制渲染流程 
   RenderThread: 用于渲染底层图形元素 
   CullingThread：用于计算裁剪 
   JobThread： 用于协同以上线程的通信
2. 线程协同的序列图

1. Cull Job,注册Culling任务：

2. DECLARE_JOB("CheckOcclusion", TOcclusionCheckJob, NAsyncCull::CCullThread::CheckOcclusion);

3. DECLARE_JOB("PrepareOcclusion", TOcclusionPrepareJob, NAsyncCull::CCullThread::PrepareOcclusion);

4. DECLARE_JOB("PrepareOcclusion_ReprojectZBuffer", TOcclusionPrepareReprojectJob, NAsyncCull::CCullThread::PrepareOcclusion_ReprojectZBuffer);

5. DECLARE_JOB("PrepareOcclusion_ReprojectZBufferLine", TOcclusionPrepareReprojectLineJob, NAsyncCull::CCullThread::PrepareOcclusion_ReprojectZBufferLine);

6. DECLARE_JOB("PrepareOcclusion_ReprojectZBufferLineAfterMerge", TOcclusionPrepareReprojectLineJob2, NAsyncCull::CCullThread::PrepareOcclusion_ReprojectZBufferLineAfterMerge);

7. DECLARE_JOB("PrepareOcclusion_RasterizeZBuffer", TOcclusionPrepareRasterizeJob, NAsyncCull::CCullThread::PrepareOcclusion_RasterizeZBuffer);

1. 裁剪处理： 
   MainThread中调用： 
   C3DEngine::PrepareOcclusion 
   | 
   CObjManager::PrepareCullbufferAsync 
   | 
   CCullThread::PrepareCullbufferAsync –> TOcclusionPrepareJob job; // 从主线程创建任务，任务则将从JobThread线程执行，此后每个Job结束后，又创建新的Job向下推进： 
   | ， 
   CCullThread::PrepareOcclusion –> TOcclusionPrepareReprojectJob job; 
   | 
   CCullThread::PrepareOcclusion_ReprojectZBuffer –> TOcclusionPrepareRasterizeJob job; 
   | 
   CCullThread::PrepareOcclusion_ReprojectZBufferLine –> RASTERIZER.ReprojectHWDepthBuffer;TOcclusionPrepareReprojectLineJob2 job; 
   | 
   CCullThread::PrepareOcclusion_ReprojectZBufferLineAfterMerge –> RASTERIZER.ReprojectHWDepthBufferAfterMerge TOcclusionPrepareRasterizeJob job; 
   | 
   CCullThread::PrepareOcclusion_RasterizeZBuffer –>TOcclusionCheckJob job; 
   | 
   CCullThread::CheckOcclusion

以上流程，大体思路是，用当前相机生成一个裁剪视锥和深度缓存，通过TestAABB，将模型的深度信息写入深度缓存（类似于实现zbuffer深度绘制算法）。 
除此，CryEngine还有其OCM处理流程（OCM Occlusion Mesh查询遮挡模型)，通过加载预先生成的OCM模型，可以快速裁剪。