I'm currently working on a software which displays some data using linear and logarithmic scales. Since this data may contain a lot of noise, we want to give the user the possibility to set a noise threshold to highlight the outlier values. Because of the nature of the software, these outliers are on both sides of the 0-line.

Just removing the data without changes to the scales is easy, since we just need to adjust the series data of the curves. But it would be better to collapse the threshold area completely since this may take a lot of screen space. As example, the y axis should have the following labels: 10000 1000 +/-100 -1000 -10000 where +/- 100 acts as 0-line and each value with y <= 100 is drawn on this line.

The only solution I came up with, is to introduce a lot of transformation logic to transform "raw value" into "value - threshold" and determining the correct label and position for those transformed values. I think this approach is cumbersome, but currently the only one I can think of. Does Qwt support such a scenario or are there any facilities which may ease the implementation?

The best solution would obviously a single spot to set the transformation - but until now, I could find any. I hope someone can provide some hints about this.