Hi!
Is it possible splitting the entire scene into different regions, render each of the portion separately and finally merge them together in UE 4.27?
For example, in this figures, the entire scene has been divided into three parts (inner, middle, outer). If we can process each of the part separately, then merge together-- that is my idea. This will be probably predefined regions. How to do it in unreal engine?
Why though?
Just a curiosity, also lowering rendering cost. Is there a way to do so?
Profilegpu in command console, look at whats killing your framerate
No, you understand this wrong. I am not thinking about current computation consumption, I am thinking of testing a new way (layered) of rendering.
Maybe the Movie Render Queue High Resolution function is what you are looking for. It splits the render into several tiles, renders the tiles one after another, then it combines all tiles to get the final image. It was introduced to create renders in way bigger resolutions than your gpu could normally handle, since with that the gpu has only to deal with a small part of the whole image.
Of course, it comes with some limitations (like several post processes do not work with it, or autoexposure) , not sure, if they have improved this since it´s introduction, or were able to lift some of those limitations for 4.27 or 5.0.
The problem i see with your method is probably the same, that the MRQ has. With only a tile (or a layer in your case) of the whole image, some effects are probably hard to apply correctly, since they would require information from regions of the image, that are in different tile or layer.
Misread it, sorry. It would take a lot of engine modification, and eye tracking to even begin to use it. Even then something like VRS would be a lot better than re-rendering due to the extra control, and re-rendering adds draw calls, which will degrade fps, instead of improving it.
