How much information do smaller pixels buy us? 0 ▲ the last word 29 minutes ago · 8 min read1555 words · Tech · hide In the previous post, I discussed how prior knowledge of a subject can allow us to measure geometric features much smaller than the diffraction limit or the pixel spacing. The example was a pair of parallel black lines of known width, separated by a narrow gap whose width we wanted to estimate. By fitting a mathematical model of the target and the imaging system to the measured pixel values, we could estimate the gap even when it was too narrow to be visually resolved. That raises another question. Suppose we keep the optical image projected onto the sensor unchanged, but increase the sensor’s sampling frequency by making the pixels smaller. How much does that improve the accuracy of our gap measurement, and at what point does further sampling cease to be useful? Let’s assume that the lens is diffraction-limited, that we know its point spread function, and that photon shot noise is the only significant source of noise. We’ll also hold the exposure and the total number of detected… Commenting on Bubbles has been removed. Why? If you want to talk about this post, write to the author, use the comments on their blog if it has them, or reply with a "RE:" post on your own blog. Suggestions? feedback@bubbles.town