CryEnginev渲染流程

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：

   DECLARE_JOB("CheckOcclusion", TOcclusionCheckJob, NAsyncCull::CCullThread::CheckOcclusion);
   DECLARE_JOB("PrepareOcclusion", TOcclusionPrepareJob, NAsyncCull::CCullThread::PrepareOcclusion);
   DECLARE_JOB("PrepareOcclusion_ReprojectZBuffer", TOcclusionPrepareReprojectJob, NAsyncCull::CCullThread::PrepareOcclusion_ReprojectZBuffer);
   DECLARE_JOB("PrepareOcclusion_ReprojectZBufferLine", TOcclusionPrepareReprojectLineJob, NAsyncCull::CCullThread::PrepareOcclusion_ReprojectZBufferLine);
   DECLARE_JOB("PrepareOcclusion_ReprojectZBufferLineAfterMerge", TOcclusionPrepareReprojectLineJob2, NAsyncCull::CCullThread::PrepareOcclusion_ReprojectZBufferLineAfterMerge);
   DECLARE_JOB("PrepareOcclusion_RasterizeZBuffer", TOcclusionPrepareRasterizeJob, NAsyncCull::CCullThread::PrepareOcclusion_RasterizeZBuffer);
   

2. 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。