On Oct 9, 2013, at 10:52 AM, Matthew Hunt <m...@pobox.com> wrote: >> >>> It seems to me that if you had a rear illuminated sensor, with no space >>> between the pixels, and it had no bayer filter, then aliasing/moire would >>> not happen, because the light value would be averaged over the whole sample. >>> >>> It's the discontinuous aspect of what is effectively three overlayed photos >>> that is causing the aliasing. >> >> Aliasing occurs regardless of how close together the cells are because the >> cells are discrete. This kind of aliasing is a function of discrete >> mathematics. > > I think you're dismissing Larry's point too easily. When we first > learn about aliasing, it's usually in the context of the Nyquist > sampling theorem, which (in the time domain) assumes instantaneous > sampling. In a pixel array, "instantaneous sampling" would correspond > infinitely small (point-sized) pixels (separated by a certain pitch). > > Enlarging the sensitive area of the pixels (relative to their pitch) > does provide some anti-aliasing effect. ...
Some amount of antialiasing effect is not the same as eliminating aliasing, making it "not happen". Aliasing will continue to exist regardless of how close together the bins are because they are handled as discrete elements in an 2D array. Larry's statement about what causes aliasing is incorrect. Aliasing occurs at the discrete discriminant: simplistically, does this object belong in bin a or bin b? Aliasing in discrete image collection apparatus occurs when the sampling frequency combined with the nature of the phenomenon creates a multi-valued solution to that question, the object is replicated into several bins (usually at different levels of intensity). Lots of things can help to attenuate aliasing, some to the point at which it is no longer visible or significant, but that doesn't mean aliasing does not happen. It's one of the issues when using discrete mathematics to represent continuous, Real-valued phenomena. G -- PDML Pentax-Discuss Mail List PDML@pdml.net http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.
