First of all I’m really impressed by nanite system. So as the topic suggests I’m trying to figure out something. Nanite is solving LOD problem by keeping (around) 2 mil triangles on screen, right? I came to the conclusion that we can bypass the problem of doubling processing power needed for mirrors if we try to pass geometry data instead of rendering image for mirror designated area.
Now the question for this topic. Is it even possible or do i miss something?
This part is if you want to hear my “algorythm” of implementing this in UE.
- Position “camera” (camera explenation in second point) that will look at geometry viewed by mirror observer
- “Camera” will gather geometry data (or mabye even whole objects)
- (This part I’m not sure about) Project objects to the other side of mirror and use mirror frame as filter to render right objects (in mirror render reflected objects and outside render normally)