A tile that provides an influence value of
\(i\) has the influence effect
\(f\) on another tile at relative position
\(\Delta x, \Delta y\):
\(f(\Delta x,\Delta y,i)=\frac{i}{\Delta x^2+\Delta y^2+1}\)
with 1.0 being the max and 0.0 being the min value for influence effect.
So if you want a single tile to have a specific effect
\(e\) at a distance
\(d=\sqrt{\Delta x^2+\Delta y^2}\) you can calculate the required influence value
\(i\) by
\(e=\frac{i}{\Delta x^2+\Delta y^2+1}\\
\Leftrightarrow i=e\cdot(\Delta x^2+\Delta y^2 + 1)\\
\Leftrightarrow i=e\cdot(d^2+1)\\
\Leftrightarrow i =e\cdot d^2+e\)
So your rough calculation is quite good, at least for a single tiles. To get an influence effect of
\(0.5\) at a distance of 50 tiles you would need an influence value of
\(i=0.5\cdot2,500+0.5\approx1,250\).
But all of your tiles (36 for a 6x6 sized building) have an influence effect, so in reality a value around
\(\frac{2,500}{36}\approx70\) seems to be reasonable (neglecting the different distances of the tiles of the building).
The precision for influences is limited to reduce memory consumption. Therefore huge numbers (like 5000

) may lead to an overflow which in return may lead to strange influence effects.