Concept of Multi-passes

How we render the cube, in different render passes.

As a complex deferred rendering system, unreal engine not render mesh in one draw call. Instead, it contains many passes.

If you are not familiar with this concept, please check this image from https://unrealcommunity.wiki/unreal-schematics-2d6859

For example, our cube need to be rendered in both early Z pass for scene depth, and in base pass for GBuffer.

In other words, please imagine a table like this:

• Why do we need to divide the Mesh into two Batches: Mesh Batch 0 and Mesh Batch 1?

Because they have different materials, the shaders used for rendering such as GBufferPass may also be different.

• Why do we need to generate separate Draw Calls for each Pass for Mesh Batch 0?

Because for the same Mesh, we have completely different drawing methods in different Passes:

1. For the Early Z Pass, we only need to output depth.
2. For the GBuffer Pass, we need to output GBuffer data. Therefore, the bound Render Target must be switched.
3. For the Shadow Pass, our projection matrix will be completely different from 1 and 2.

Mesh Draw Command and Mesh Pass Processor

This means that even if we only have a single FMeshBatch, we need to generate more than one draw call. We must save the information related to each draw call somewhere. This is what we call a FMeshDrawCommand. And we need someone to take the responsibility to generate FMeshDrawCommand for each different pass, we call that one FMeshPassProcessor .

If you want to know more, here is the official document.