Custom Curve Bias function
Illustrations showing how Bias affects cursor magnitude & direction with differing x/y sensitivities. See the User Guide for more information on how to use Bias in Custom Curve.
Angle & magnitude
Cursor path for a circular mouse motion
Controls
Bias curve
Readout
The relationship between horizontal and vertical sensitivity
Because of how a mouse is used on a desk, moving it sideways and moving it forward and back are not really the same gesture. Left and right, the wrist and arm have plenty of room to work with; forward and back is the more constrained motion, drawing on different muscles and offering a different degree of control. It is reasonable, then, to want some aspect of sensitivity to depend on which way the mouse is being moved.
Standard per-axis sensitivity (Bias 0%)
Standard per-axis sensitivity scales horizontal and vertical movement separately. This keeps the two settings independent: horizontal distance depends only on the x-component sensitivity, and vertical distance depends only on the y-component sensitivity. However, this also changes the direction of off-axis movement. For example, if you move at 45° with two different sensitivities, the cursor travels at a different angle. The directional error is zero along the axes and largest between them, so it changes as the cursor moves through an arc. This is one of the reasons why gamers often say they prefer “1:1” sens when trying out differing x/y sensitivities — it is not the change in sensitivity, but the angular deviation.
It is not only the angle that can seem arbitrary with different x/y sensitivities and off-axis inputs; the resulting magnitude can seem so too. The shape of the cursor path is also affected, bearing less resemblance to the shape of the movement made on the pad, as shown in the cursor path illustration.
The Bias method (Bias 100%)
Bias instead applies a single scale factor to the entire movement based on the input angle. The cursor direction therefore matches the direction of the mouse on the pad at every angle. The two sensitivity settings are no longer independent: diagonal output becomes a defined product of both sensitivity values, with the final scaling factor for any angle determined by the Bias curve, applied proportionately in logarithmic space. The practical effect is that the two sensitivities can be set for comfort and control alone — at whatever ratio that calls for, rather than to whatever angular deviation can be tolerated.
The illustrations here use two fixed sensitivities for clarity, but nothing about the method depends on that: Bias holds direction just as well when each axis is driven by its own acceleration curve.
The Bias slider performs a linear blend between the two methods of sensitivity application.