+11hi!
I would like to use HullAccumulator to create polygones around group of lines spatially apart, but I don't have any attribute to permit to use "group by" option.
Do you have any idea to do that? I don't find a tranformer that permit to give an ID to a group of objects that intersect each other.
Thanks!
+44Send your lines to a counter to assign a unique id, if they don't have one already, then a spatial relator as both requestor and supplier to find the intersections and build a list of the id's each feature intersects with, then a listcombiner (custom transformer from FME hub) to create group ids for all features that relate to each other. This should give you your group by option
If you want to group by connected-ness, then you can use the TopologyBuilder followed by the NetworkTopologyCalculator and use the resulting attribute _network_id as Group By in the HullAccumulator.
+44If you want to group by connected-ness, then you can use the TopologyBuilder followed by the NetworkTopologyCalculator and use the resulting attribute _network_id as Group By in the HullAccumulator.
Because the networktopologycalculator splits the lines, this can lead to slightly different hulls depending on the input geometries and hull type being used.
Because the networktopologycalculator splits the lines, this can lead to slightly different hulls depending on the input geometries and hull type being used.
Interesting, I wasn't aware of that. I must admit I don't quite understand why that would be however, do you have an idea?
+44Interesting, I wasn't aware of that. I must admit I don't quite understand why that would be however, do you have an idea?
No, I don't really understand why. But if you generate a convex hull with a 0 alpha value, the result varies if the lines are split at intersections. Tbh it may be that the output after the topology builder is more correct? But not splitting more closely resembles the output if you had manually assigned a group attribute (I think that's how i first spotted the difference)
No, I don't really understand why. But if you generate a convex hull with a 0 alpha value, the result varies if the lines are split at intersections. Tbh it may be that the output after the topology builder is more correct? But not splitting more closely resembles the output if you had manually assigned a group attribute (I think that's how i first spotted the difference)
Thanks, I learned something new.
