Originally Posted by TheANIMAL
Oh, well... good then.
But you would have thought that they would have found an adaptive way to make the resolution higher as it gets closer to the camera or perhaps they already have. The global illumination looks a bit muddy is all i'm saying.
You did read the papers which I posted in previous page?
They use currently 3 cascades for propagation volumes which do just that.
They also state that they also use SSGI methods for smaller details.
I do agree that it is not anywhere near perfect, but when you consider the flexibility and needed computation power, it does look good.