Originally posted by falconeye Don't cite me wrong.
You are a difficult person to communicate with.
I didn't cite you wrong. You didn't provide any proof that an AA filter is "totally reversible transformation (in the sense that it can be 100% "undone")."
If you don't want to post in the thread, send me your proof by PM.
I don't need any maths - although I do have a major in Pure Maths and I am somewhat familiar with what you are talking about.
All you need to do is start with an image taken with no AA filter (with frequency information extending past Nyquist), apply AA, apply your deconvolution, and show me you will get back the original image, with every pixel restored.
Note: this implies restoration of frequencies above Nyquist (and yes, I am aware by Shannon's theorem that there is no "useful" information above Nyquist) because otherwise it's still an unfair comparison - you are comparing one image with frequencies above Nyquist (implying presence of aliasing artefacts) with another image that is sharpened, but filtered.
Making claims like there's an extra 5% of detail around Nyquist that could potentially be recovered ... Sorry, not really relevant given there are so many other factors affecting image quality between the two cameras that any improvement would be insignificant.