Tweet
So generally speaking reprojection area field of 1.0 is the most ideal scenario due to photo consistency filter usage, AKA at value of 1.0 we can receive the (mathematically) all / pixel information that our image contains and (theoretically) it will be converted on our mesh by triangle projection.
Now it rise a few questions in my mind:
- if the above is true, can we say, that high megapixel cameras are able to help us in the highest geometrical micro detail reconstruction? Should we aim for a higher megapixel sensor VS large pixel size? (I do not consider the capturing distance factor for now, so lets say a lowpass filter less-no AA filter full frame sensor with 24 vs 42 megapixel)
- of course we are talking about drastically raised polygon count at value of 1.0 so blockers can play a role, as visualizing the reconstruction - vRAM, Mesh reconstruction - RAM
- do the photo consistency filter has any limits at projection area value of 1.0? Or theoretically it do not have any? Would we see great benefits from a 42 megapixel sensor vs a 24 one at value 1.0?
Did anyone made such a test?
Now it rise a few questions in my mind:
- if the above is true, can we say, that high megapixel cameras are able to help us in the highest geometrical micro detail reconstruction? Should we aim for a higher megapixel sensor VS large pixel size? (I do not consider the capturing distance factor for now, so lets say a lowpass filter less-no AA filter full frame sensor with 24 vs 42 megapixel)
- of course we are talking about drastically raised polygon count at value of 1.0 so blockers can play a role, as visualizing the reconstruction - vRAM, Mesh reconstruction - RAM
- do the photo consistency filter has any limits at projection area value of 1.0? Or theoretically it do not have any? Would we see great benefits from a 42 megapixel sensor vs a 24 one at value 1.0?
Did anyone made such a test?