You should always have your images scanned as "colour slides", even colour negative film, in order to make sure the entire information captured on the film is sampled and does not get lost in any process of "auto-correction" which typically happens when the scanner's software or scanning software tries to remove the mask.
YOU WILL NEED TO HAVE YOUR IMAGES SCANNED @48BIT COLOURDEPTH in order to have enough colour information to remove the mask.
I. What is the mask for?
Much has be written about it, so I will just give a short summary.
Film was developped, obviously, for purely analoge environments originally, which provided fairly little options for post-production colour correction. Problem was and is that there is very limited amount of chemicals which can be used, and those available have a certain undesired side effect: the do not produce entirely clean colour in each layer (Red- Green- and Blue-sensitive layer) as the dyes are impure. Result: Magenta-Layer (green-sensitive) contains not only magenta, but also yellow. Cyan-layer (red-sensitive) contains not only cyan, but also magenta and yellow. Consequently, in the print red would have a blue-greenish (or basically greyish) tint, and green a bluish tint.
That issues is being solved by adding a certain amount of magenta and yellow to the unexposed part of the affected layers to compensate for that effect. E.g.:
Cyan layer unexposed without masking: 0% Cyan, 0% Magenta, 0% Yellow
Cyan layer fully exposed, without masking: 100%Cyan, some 30% Magenta, some 30% Yellow (the latter two unwantedly, obviously).
Cyan layer unexposed with masking: 0% Cyan, some 30% Magenta, some 30% Yellow
Cyan layer fully exposed with masking: 100%Cyan, some 30% Magenta, some 30% Yellow (the latter two unwantedly coming from exposure, because the masking appears only on unexposed film (therefore it is called "masking")).
Obviously, through masking it has become possible to simply subtract the some 30% of magenta and yellow from the entire image while printing (e.g. increasing the amount of blue and green light reaching the print), which will result in:
Cyan layer unexposed with masking: 0% Cyan, some 30% - some 30% = 0% Magenta , some 30% - some 30% = 0% Yellow
Cyan layer fully exposed with masking: 100%Cyan, some 30% - some 30% = 0% Magenta, some 30% - some 30% = 0% Yellow.
The result is pure colours. In the digital age this process would not necessarily be needed anymore, but likewise in the digital age film is not being technically developped a lot anymore (which is a pity as it is so capable compared to e.g. a Canon 5d II and similar), although there are a few unmasekd colour negative films available.
II. How NOT to remove the orange mask
There are a couple of descriptions on the web which by several steps try to remove the mask itself (be seeking to calculate the orginal mask out of the colour layers and then to only remove the mask as such). This, in my view, does not make sense at all. Not only is it overly complicated, it is also counter-productive. What you would get is, at best, the original impure dyes. It is WAY easier and more useful, to simply simulate the oroginal process of reducing the amout of yellow and magenta in the ENTIRE IMAGE, to get the pure colour.
III. How I remove the mask
I am not sure that this is technically accurate and I may change my opinion on this, but the results so far have been quite convincing in my view.
I am using Photoshop, in which I have automated the following process in order to a) make availabe through pressing a function key simply and let it run automatically and b) to have identical results for each image (correction can be done manually afterwards, if needed).
Ideally, you will create (once) a CMY profile, meaning a profile NOT using the black channel of CMYK, but mapping all colour on the the CMY channels. The result we be somewhat similar to what was on film originally ("somewhat" due to colour shifts produced by the scan).
The automated function should do the following:
1. Convert to your custom CMY profile.
2. In the channel mixer reduce
a) constant of Magenta by 33%
b) constant of Yellow by 66%
3. Flatten image.
4. Convert back to RGB.
The result should provide you with fairly decent colour providing lots of space for subtle correction and showing a highly detailed colour distinction. In theory, you would not have needed to convert into CMY at all, but for reasons unclear to me results tend to become better this way; it appears that the conversion of colour space affects the result (?).
... the echo of a distant time ...