3 ms·
Delightful post. > in theory, once per half-pixel, according to Nyquist I'd have thought this should be once per two pixels instead. Nyquist says there's no a
by tgb 6y ago
Delightful post.
> in theory, once per half-pixel, according to Nyquist
I'd have thought this should be once per two pixels instead. Nyquist says there's no aliasing between functions with wavelength L if the sampling at intervals of L/2. So sampling at once pixel should imply a 2-pixel wavelength minimum without aliasing. Assuming the author is right, what am I messing up?
- andreareina 6y agoIf L is 1 pixel, then L/2 would be 0.5 pixels.
- tgb 6y agoBut L/2 should be the sampling interval, which is fixed at 1 pixel. For example, a signal with a wavelength of 1 pixel (or 1/2 pixel) would be identical to a constant signal.