alanr0

Agreed. One could equally say that dynamic range has increased as a side effect of improved low light performance, but this too is a misleading over-simplification. In crude terms, engineering ...

Diffraction should be symmetric (at least for well-rounded aperture blades). Sensor PSF will have 2-fold rotational symmetry, 4-fold with the usual square (rather than rectangular) pixels. We lose ...

Even order is necessary, but not sufficient for rotational invariance. Astigmatism has even order: Z(2,2)  ~ r^2 cos(2 theta), Z(2,-2) ~ r^2 sin(2 theta) Secondary astigmatism and quatrefoil have ...

Frans, As Jack says , the PSF is real, not complex. One interpretation of PSF is the system response to a delta function input. In our case lens PSF is the image plane intensity distribution from ...

Loss of contrast at high spatial frequencies due to diffraction can be improved by deconvolution, but only up to a well-defined limit. A standard way to deconvolve an image is to take its 2D ...

Looks reasonable. It doesn't specify whether the response is quantum efficiency (electrons/photon), or radiometric (volts/Joule), but it could be QE. Here are two more examples showing QE vs ...

What aspects of the graphs do you find inconsistent with typical photographic sensor responses? As far as I know the silicon is doped, but with rather low levels of acceptor or donor atoms (e.g. ...

As Jack says, the dividing line can become rather fuzzy. Recall that "post processing" also applies to editing out-of-camera JPEG images. In this case, the "post" is "after the camera has produced ...

We - or at least most of us - do not wholeheartedly endorse Wikipedia, but it is often pretty reliable, especially for more popular topics in science and engineering. It is also highly convenient ...

Can you clarify what you mean by a "primary source" in a scientific or engineering context? For example, what about ISO12232:2006E "Photography — Digital still cameras Determination of exposure ...

Some plausible, but mildly conflicting answers culled from the web: Thomas Ulrich on Quora refers to M A B Deakin's historical articles [ 2 ] [ 3 ] in which Laplace himself attributes the technique ...

I am not convinced by your "must". Do you mean "lightness", used in Hunter Lab space, or "lightness" used in CIE L*a*b* space?  In either case L or L* is dominated by the luminance (Y) channel ...

Is this true? For a simple monochrome conversion in which you add C+M+Y, there will indeed be twice the signal for the same exposure, though you may need a shorter exposure with an ETR strategy. Fo ...

JPEG uses Discrete Cosine Transform . DCT is a DFT of discretely sampled real quantity, which is assumed to be symmetric about the origin. DFT of a real symmetric function is real, and contains ...

A colleague recently raised this topic in relation to truncation of Gaussian beams by the aperture of a lens. I ran simulations showing the dependence on apodisation strength (expressed in terms ...

Actually it is spelt out in the dpreview forums rules here : So the OP may be making a mountain out of an ant hill, but posting in a format-specific forum, with another post to a broader audience ...

I mean, c'mon?! Is this your biggest gripe about the site? The name of the forum is " Pentax SLR Talk ". That is still accurate and fully inclusive. Both APS-C and FF cameras are discussed here. We ...

No need for CPU clock and ADC clocks to be identical. JPEG conversion takes time. Writing to flash is slow.  In enthusiast and professional cameras, data is copied to buffer memory so that writing ...

I guess in sampling theory you generally don't keep the deltas and their infinities hanging around for long.  The next conceptual step integrates over each delta in turn, and produces an array of ...

A Kronecker delta has no size.  It takes two integer arguments, and in the present context only applies to discretely sampled data.  That is not the real world.