Currently, a trivial task to render out specific attributes for a set of meshes, placed in the world, always requires a fully custom solution.
The closest thing is base color only scene capture, but unfortunately it is too slow to be used for anything and disallows blending. If you’d need to capture something with blending, it pushes you into using scene color capture, ramping up totally unnecessary costs to a new level.
I propose and request having a lightweight scene capture, that would allow rendering only unlit emissive materials(including negative emissive) and/or depth only, and perhaps stencil, with nothing, but frustum culling and optionally sorting on the rendering side, bypassing all the bulkiness of scene capture, such as depth pre-pass and occlusion culling.
There is a good line of environmental effects that can be achieved using it and it would have great synergy with Niagara and RDG.
Unlike special cases, this is universally needed.
Another thing that scene capture desperately needs is ability to sync temporal jitter between active camera and scene capture.
Introducing and supporting class as described would cut down time investments and lower the bar of accessibility to a whole line of rendering effects for good deal of your licencees.
Kindly consider it.