Joined
·
1,149 Posts
I remember when I first joined this forum there is a certain mysticism surrounding the notion of "native CPI" and somehow relating the photodiode pixel-pitch to the supported set of CPIs.
Now, after a period of my advocacy, we all understand that subpixel movements are inherently tied to the process of image correlation itself, and moving by "integer multiples of pixels" are completely trivial, that the CPI settings are a purely arbitrary conversion unit internally used by the sensor chip -- reveaing Morier's analogy of "subdividing pixels" as an incorrectly oversimplified illustration, and that the supported CPI sets are just convenient units obtained from subdividing a subset area (i.e. 50cpi step means a reference area of 16x16 pixels at a density of 800dpi).
What are some other examples of these kinds of evolution in our collective knowledge? Things like frame averaging errors, established standard protocol for quantifying sensor attributes, etc. Can you point to the origin of each historic development in this subforum?
Now, after a period of my advocacy, we all understand that subpixel movements are inherently tied to the process of image correlation itself, and moving by "integer multiples of pixels" are completely trivial, that the CPI settings are a purely arbitrary conversion unit internally used by the sensor chip -- reveaing Morier's analogy of "subdividing pixels" as an incorrectly oversimplified illustration, and that the supported CPI sets are just convenient units obtained from subdividing a subset area (i.e. 50cpi step means a reference area of 16x16 pixels at a density of 800dpi).
What are some other examples of these kinds of evolution in our collective knowledge? Things like frame averaging errors, established standard protocol for quantifying sensor attributes, etc. Can you point to the origin of each historic development in this subforum?