In my PCG graph, I have two lines of ‘code’ that have the same input of surface points to sample from, but the density, extend (X and Y) and looseness differ:
.
Because the area that the graph is input, it can cause an error in any Surface Sampler node, when density, extends and/or looseness are above certain levels for that area. For instance, if PCGVolume is set to scale X=1200, Y=1900, the number of points in the lower SurfaceSampler exceeds 160.000.000 - too high. Set to 1/2th (600x950), the number is 7.300.000 which seems acceptable.
So I have two questions: What is the maximum number of points that Surface Sampler can handle, and is there some calculation in which I can tweak density, extends and looseness with a given number points, so these parameters can be calculated without having to re-generate the graph?
Generate less points? (ei: feed in a smaller area to sample?)
No where does it say that you can’t re-run PCG several times in different areas… so if you are concerned about “this zone is too small”, well, don’t be. Duplicate the thing and run it again over a different zone.
The nicety of it is that once you are done the stuff/instances are placed and can even be completely unlinked off PCG.
Matter of fact, I’d suggest just outputting mesh name → transform into a CSV, and creating a system of your own that adds the instances into the level using other (more compatible with older systems) methods - like Foliage and custom HISMC inside dedicated BPs.
