Could someone explain to me why it is common practice (or maybe not) to use Linear Fit in cases of narrowband imaging in order to align the Luminances of Ha/OIII/SII before RGB channel combination?
I guess it's used just as an "easy" way to equalize the three narrowband images. Using LinearFit is in this case as arbitrary as any other tool such as HistogramTransformation, curves, etc., but more automatic.
Shouldnt the relative strength of each channel be presented as is and isnt that a better representation of the actual conditions?
Not necessarily. It depends on what you want to communicate with the image. For example, in galaxy images mixing Ha data with broadband RGB data the Ha image has to be amplified if one wants to represent HII regions; otherwise Ha and red are usually indistinguishable. In this case, by boosting Ha we are increasing the information represented in the final image. This is plausible as long as we apply uniform criteria to the
whole image, based on physical properties of the objects,
and as long as we inform the spectator that we have applied these techniques. Here is
an example with M51.
