You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
thanks for your implementation of this library. I'm currently evaluating libraries for boolean operations and counted your martinez implementation as a top contender. During my evaluation I've noticed the intersection computation completely fails for polygons with very short segment lengths: in my demo I'm approximating bezier-curves through polygons. The intersection algorithm seems to work (much better) with longer line segments - that is what I think is the reason.
Here are two videos of the demo. I'm computing intersection between a circle and some ornament multipolygon I found on the internet. I'm moving the circle interactively with the mouse - the resulting intersection is shown in red color.
I wondered if I was interpreting the result data wrong or if the input data was malformed - but this would contradict the algorithm performing well on longer segments. Are there any parameters in the algorithm that are tweakable (accuracy or "epsilon") to support shorter segments? Is this a known issue? I've also tried the polybool library which is about 10 times slower but has pretty accurate results also for short segment lengths. Would you suggest me trying your Greiner-Horman implementation?
The text was updated successfully, but these errors were encountered:
martin19
changed the title
Intersections fails with short segment lengths
Intersection fails with short segment lengths
Jun 2, 2020
Hi,
thanks for your implementation of this library. I'm currently evaluating libraries for boolean operations and counted your martinez implementation as a top contender. During my evaluation I've noticed the intersection computation completely fails for polygons with very short segment lengths: in my demo I'm approximating bezier-curves through polygons. The intersection algorithm seems to work (much better) with longer line segments - that is what I think is the reason.
Here are two videos of the demo. I'm computing intersection between a circle and some ornament multipolygon I found on the internet. I'm moving the circle interactively with the mouse - the resulting intersection is shown in red color.
https://imgur.com/fJx7Bqd (10 division per bezier segment)
https://imgur.com/TUyDcKP (50 division per bezier segment)
I wondered if I was interpreting the result data wrong or if the input data was malformed - but this would contradict the algorithm performing well on longer segments. Are there any parameters in the algorithm that are tweakable (accuracy or "epsilon") to support shorter segments? Is this a known issue? I've also tried the polybool library which is about 10 times slower but has pretty accurate results also for short segment lengths. Would you suggest me trying your Greiner-Horman implementation?
The text was updated successfully, but these errors were encountered: