# f-number equivalent between m43 and FF

Started Mar 25, 2014 | Discussions thread
Re: Defining The Signal: part 2

D Cox wrote:

Jack Hogan wrote:

1) The light intensity at specific points in The 2D Image projected on the sensing plane during exposure

For photographic purposes we can express light intensity in a number of variations of radiometric and photometric units (including W/m^2/nm, lux, photons - of a given wavelength or not - per second per meter squared etc.).

So this definition seems pretty reasonably, but it totally removes photon shot noise (traditionally counted as "noise"). Is this what you want to do?

I also believe that we also often refer to The Signal as being further downstream, at the output of the sensor in a spatially and temporally sampled version of The Image, in units of photoelectrons.

If you're interested in the signal-to-noise characteristics downstream of the "sensor" (including A/D conversion, any pre-filtering or amplification, etc.) then sure. But is that what we're talking about here?

D Cox wrote:

Well, such debates on terminology can be never-ending.

There's never one "correct" definition. There are, however, bad definitions depending on what you're interested in discussing.

My opinion is that the signal is the photograph. To be more precise, it is the array of numbers in a digital image that provides information about objects and light.

Besides the vagueness (e.g., the numbers stored at what stage?): If this is the signal, then haven't you precluded any possibility of "noise"? I.e., "What I measured with my instrument is defined as the signal." Does it make sense to define things this way?

I don't think a single measurement (one pixel, one exposure) should be called a signal. There has to be a set of measurements (different pixels in photography, samples at different times in audio).

So, in "The light intensity at specific points" the important things is that it is plural. The information (the picture) comes from comparing those points: the signal is the differences between the numbers from the pixels. (In audio, it is the differences in the numbers from sample to sample.)

You may not agree.

I think words are adding to the confusion here. Please write out what you mean precisely. For example, take the audio case: Suppose the "true" input to some digital audio system is a function F(t) of time t. You get some samples out, S[n], at discrete times n. So what are you calling the signal here? And what do you mean by "the differences in the numbers from sample to sample"?

Complain
Post ()
Keyboard shortcuts: