We’ve got a setup where a scene capture component is rendering a small set of primitives (particles mostly, but it could be anything) into a render target, which is then used as an input for other materials to drive some special effects. However, because the scene capture component is using the same FScene as the rest of the world, when it goes to render its view it still pays the cost (time, memory) of processing visibility and everything for all of the many thousands of primitives that are active in the main scene, even though it’s set up to only draw the small set of primitives we’ve specified.
Before I go down the path of either having all of these effect primitives being captured separately as part of the main rendering path, or adding some flag to the rendering pipeline to modify its behavior when rendering this view (like making it always assume that everything is invisible, without going through the normal occlusion/frustum culling/etc. paths), I wanted to see if there was some way I could separate them out from the main scene into a new scene, and still keep the current system using a scene capture component on this small, custom scene that only contains the things I want drawn into the render target. Ideally, I’d like to keep the changes required as limited as possible, and it feels like being able to have the SceneCaptureComponent keep track of its own little scene would let me do what I want without having to add special-case behavior to the main rendering pipeline.
Is this a reasonable thing to do? What kind of problems am I likely to encounter if I do attempt to do something like this?