The pressure-dependent brush angle jitter is inconsistent and could be more effective. Here is the problem and suggestions.

When control is set to "pen pressure", the behavior is so completely differently that we might have changed apps.

This "pressure-dependent" behavior is inconsistent with intuition and with the no-control situation.

BTW, I have found angle jitter to be useful in modeling texture of cloth.

**Here's how it works now:**When control is off, for angle jitter setting of x%, for each separate stroke,**e****ach angle seems to be a sample from a fixed probability distribution**whose width depends on x. The angle is measured relative to an invisible line (horizontal line if tip angle is 0) across the drawing surface, and independent of cursor direction.When control is set to "pen pressure", the behavior is so completely differently that we might have changed apps.

**The angle now seems to take a random walk around a circle,**with the "*step size*" given by x% of something. This is very different, as now all possible angles appear, and there is unwelcome correlation for small jitter settings.This "pressure-dependent" behavior is inconsistent with intuition and with the no-control situation.

**The fix:**With pen pressure, the angle still should be drawn from the same fixed probability distribution, but where the*width*of that distribution is proportional to pressure times the jitter setting. With low pressure and/or small % settings, there are only small angles, and with larger pressure and/or large settings, there are larger angles, but the angle value is always drawn from the same scaled probability distribution (at least as much as possible). It would also be particularly useful if the angle could always be measured from a perpendicular to the pen direction.BTW, I have found angle jitter to be useful in modeling texture of cloth.

Be the first to post a reply!