A logarithmic scaling of the negative part of the y-axis would let scores about -999,999 (alomost a million) take the same place as -6,000, a value about -100,000 would have the position of a value about -5,000; -10,000->-4,000 … and so on.
If the score stay negative, the automatic zooming of the y-axis (which is already implemented) would automatically fit it to full scale.
In this case, I meant a logarithmic scaling with a base of 10, of course.
In respect to what infjamc wrote, a function to rescale the score-value for displaying (before zooming) could look like this (written as lua-function):
function yrescale(y)
if y>=-10000 then
y2=y
else
y2=-log(abs(y),1000)*10000
end – if
return y2
end – function
whereas log(abs(y),1000) means the logarithm for the absolute value of y to the base of 10,000.
By using a tau-function, you could also do "soft clipping".
I can work out a formula, which fits to the tau-model, if you want.
change y2=-log(abs(y),1000)*10000
to y2=(4-log(abs($A3);10))*1000-10000,
then all values equal and above -10,000 will have the original value.
In contrast, a value of -100,000 will have the position of -11,000,
a value of -1,000,000 will have the position of -12,000,
and so on.
sorry, I meant y2=(4-log(abs(y),10))*1000-10000.
For now, I heard from some users with fast PCs, that it is impossible to stop rebuild or scripts at "good positions", because their PC ist "too fast", so they HAVE to stop and look for a good positions in the undo bar, so negative values in the bar would help much.
The undo graph will also handle negative values, using Advanced GUI, in the next update.
This went out in the last update. Marking as resolved.
Wow, thanks again!