Question

Is the placement of vertices for circular buffers around a point predictable ?

  • 26 January 2023
  • 4 replies
  • 10 views

Hello, when using the Bufferer transformer on points, a circular polygon is generated. The number of vertices constructing the circle is a function of parameters like the corner endcap type + interpolation angle. I'm good with this.

 

What I would like to know is if there is some logic to how the vertices of the buffer are generated. That is, how is the first point on the circle determined (since all others would follow from there based on the number of vertices / interpolation angle).

 

To provide context, we are capturing points that fall within the buffers and we have some edge cases that appear would be captured if only the circle were rotated or we increased the number of vertices. We want to see if there is a way other than increasing the number of vertices since this tends to slow down downstream transformers.

 

In the picture below, you will see that depending on the rotation of the points, the green dot may or may not be captured. In the case shown, it's captured by the grey filled circle (only portion shown) while it's missed by the red outlined circle.

 

Circles 

Thank you


4 replies

Userlevel 1
Badge +21

I'm not sure how this would compare performance wise with your data, but could you not use a NeighborFinder instead to find the distance from your point to a point that is the centre of the buffer? This would pick up these edge cases

 

imageAs you can see, a point fails the test of falling within the point that has been buffered to 5m but is found with a maximum distance of 5 with the NeighborFinder

I don't know either but this is a great suggestion ! I will definitely explore this option which may make a lot more sense. You know, these basic processes seem simple. We do them all the time. However, we take for granted that we really sometimes need to know what's happening in order to anticipate potential issues.

 

Thanks for the reply ! @ebygomm

Userlevel 5
Badge +25

I'm wondering if using a 2DEllipseReplacer would give you a true arc instead of a stroked one like the Bufferer does, circumventing the issue alltogether (although I think @ebygomm​'s suggestion is probably the way to go)

Thanks also for the suggestion @redgeographics. Yes, while the neighbourfinder is probably the best way to go, for some, they will still need to see the buffers so a true arc may help.

Reply