I’ve added 17 more cameras to my test suite. The results:
Same trends apply. Older CCD cameras are worse for SMI, except for the NEX-5N, which has lousy SMI. Not a lot of difference wrt the CFA’s effect on noise.
Photography meets digital computer technology. Photography wins -- most of the time.
I’ve added 17 more cameras to my test suite. The results:
Same trends apply. Older CCD cameras are worse for SMI, except for the NEX-5N, which has lousy SMI. Not a lot of difference wrt the CFA’s effect on noise.
Hi, JimK.
After reading your six articles about this theme, I have a question. I am just an amateur in photography thus bear with me for my knowledge level.
In reality, the radiance energy is filtered according to the CFA‘s spectra, thus this might affect the SNR of the sensor if the illuminance and exposure time are the same on the sensor plane. If some CFA spectra designs filter out larger parts in wavelength, then SNR is lower as less energy remains. Am I correct?
And in your conclusion, as you stated in the 1st article of this simulation methodology:
1, Normalization for each color channel;
and 2, ETTR for a color plane.
You deliberately exclude the attenuation of the filters.
I think that is why the conclusion you made does not fully match my first thought about the relationship between SNR(relative to visual noise) and CFA spectra design.
The filters are normalized so their spectral peaks are at 100% transmission. I have ways in the program to deal with additional losses, but hardly any data on those losses.