It was brought to my attention that the Distance colouring method has always looked different in my GMP fork compared to the last upstream binary version 2.11.1.
I managed to track this down to the very early history of my fork, when I imported the contents of the 2.11.1 source code zip.
Turns out the source code doesn't quite correspond exactly to the released binary version - it has an extra square root in the Distance function (and possibly other changes).
So my plan of action to resolve this mess and make old parameter files look correct in my fork:
* make ColorMethod 5 correspond to Distance (Linear) for compatibility with 2.11.1 upstream binary version (it was (Distance (Square Root) in my fork, until now, apart from one version)
* make ColorMethod 7 correspond to Distance (Logarithm) for compatibility with my fork version (anyway a new feature in my fork compared to upstream binary)
* add a ColorMethod 8 corresponding to Distance (Square Root) for compatibility with my fork version
Therefore
some settings made with my fork will have to be changed manually, ie change ColorMethod 5 Distance to ColorMethod 8 Distance (Square Root) (or ColorMethod 7 Distance (Logarithm) for that one GMP fork version with log transfer in the de function)
Any comments? This is a bit of a mess all around...
EDIT Here are some test images:
2.9.4 upstream
2.9.11 upstream
linear
logarithm
square root
So it isn't just linear scaling that is the issue, there's some other missing ingredient, and older versions differ between themselves too (but not as much).
What to do?