Kayvon mentioned in lecture that this is a result of aliasing, but I'm still having trouble understanding what he means by that. Would this image be much more gray if the sampling rate was much higher? and why would aliasing cause the image to repeat a pattern?
akshay
If you zoom in and out on this image, some of the patterns will appear to change or move around--these are the ones caused by aliasing. They appear because the original, infinite-resolution signal rapidly alternates between white and black; however, since we're sampling the signal to display it on a screen, we occasionally sample only when the signal is white and we occasionally sample only when the image is black. In reality, the corresponding pixel should be grey since it's 50% black and 50% white, but the sampling algorithm only sees a single color and assumes that the whole pixel should be that color. Since the frequency of the signal varies across the image, we see weird artifacts as some pixels turn white and some turn black. If the sampling rate were higher, the algorithm would be more likely to see both black and white samples and correctly figure out that the pixel should be grey.
Kayvon mentioned in lecture that this is a result of aliasing, but I'm still having trouble understanding what he means by that. Would this image be much more gray if the sampling rate was much higher? and why would aliasing cause the image to repeat a pattern?
If you zoom in and out on this image, some of the patterns will appear to change or move around--these are the ones caused by aliasing. They appear because the original, infinite-resolution signal rapidly alternates between white and black; however, since we're sampling the signal to display it on a screen, we occasionally sample only when the signal is white and we occasionally sample only when the image is black. In reality, the corresponding pixel should be grey since it's 50% black and 50% white, but the sampling algorithm only sees a single color and assumes that the whole pixel should be that color. Since the frequency of the signal varies across the image, we see weird artifacts as some pixels turn white and some turn black. If the sampling rate were higher, the algorithm would be more likely to see both black and white samples and correctly figure out that the pixel should be grey.