I’ve noticed a bug – when you’re using a parameter for the Fill() method giving an array on graphs you want to scan, the FloodFill is still called for ALL graphs and that impacts a performance.
Seems that FloodFill() needs the same “NavGraph[] graphsToScan = null” parameter.
Hi
This will have been implicitly fixed in 4.1.22 I believe (though I haven’t verified it actually).
The optimization you are considering was not previously possible, because there could have been off-mesh links connecting different graphs, and thus the flood fill would have to run for all graphs in some cases.