Closed Aeva closed 8 years ago
Light size is hard coded at the moment, but now 4 ray casting samples are generated instead of just the one.
Highlights are still using the light's origin view matrix, which may or may not be ok as is.
Some further experimentation is needed - the shadows don't seem to get softer with depth, but idk maybe its fine?
I don't think I'm going to continue with this approach, as the quality of the shadows looks worse the greater the light size is. It would need more samples to look good, and it does seem to be not so cheap in other ways even though it only needs one texture lookup.
It was a fun experiment, but I don't think it is what I want.
I had an idea just now that may or may not work, but I think is worth exploring. It works like so:
The accumulations passes would use the light's actual view matrix, not the random ones since they would not be accessible.The offsets could be a standard spacing with just a random starting angle, provided via a uniform var, allowing the exact matrices to be recreated easily.extra credit: MRT could be used to add more samples, though that seems excessive