Thanks for dropping by, Lupo.
We're basically applying the mod N arithmetic to the deepest well. This converts it from one that produces 360 degrees shift at the design frequency to one that produces zero degrees of shift. The situation is the same for multiples of the design frequency.
It is only one cell affected in the N7 and higher, although the rarely used N5 2D inverse has 9 cells changed.
Using the substitution brings the results into line with the products that are commercially available - although really, there are only variations on the N7 out there.
The resulting inverse panels are the same depth as the non-inverse panels, meaning that for any order, the choice between the two comes down to how they look.
There doesn't seem to be any barker style modulation used with 2D panels, so any QRDude support for non-optimised 2D inverses would be purely academic.
In the interest of keeping the calculator as simple as possible, I'm going to make a judgement call:
For 1D inverses, the subsitution will not be used. (There is plenty of choices for normal panels using the phase wheel. Inverses would normally be used for modulation, so should have the correct maths)
For 2D inverses, the substitution will allways be applied. (No barker style modulation possible - panels only to be used in arrays of the one design)
I was thinking of using a disabled menu option as a reminder:
Paul, that French site is a good find. They also have some nice lighweight plastic N7's...