A smooth field through scalar or vector Point Values, fitted with Gaussian radial basis functions whose width is the average point spacing divided by the factor. Each value sums the points its Cutoff takes: by default those within 3.72 widths, where a point left out would have added less than 1e-6 of its own peak. Takes at most 2000 points; the kernel is provisional and may change.
The Point Values to fit: numbers or vectors.
Sharpness of the fit: the Gaussian width is the average point spacing divided by this, so a smaller value spreads each sample further and a larger one keeps it local.
Which points each value sums. Auto: those within 3.72 w, where w is the Gaussian width (the average point spacing divided by the Factor); a point left out would have added less than exp(−3.72²), under 1e-6, of the peak of its own Gaussian. Radius: those within the Cutoff radius. Every point: all of them, the exact sum, the slowest with many points.
Each value sums only the points within this distance in mm. A point left out would have added less than exp(−(R/w)²) of the peak of its own Gaussian, where R is this radius and w the Gaussian width (the average point spacing divided by the Factor): from R = 3.72 w on, less than 1e-6 of it.
Values the block gives besides its output. The editor lists them under Properties, and an input can read one by reference.
The field read as a Solid whose surface is where the field is zero.
The box the field is evaluated in.
The field with its sign flipped (−f), which swaps inside and outside.
The field divided by its gradient length (f/|∇f|) so it reads close to a true distance; the zero set does not move.