Skip to main content
This page is for AI assistant context only. To use this package, visit community.ntop.com.
Community submission — not verified by nTop. This package was submitted by an outside community member and has not been reviewed or verified by nTop. Treat its accuracy, safety, and compatibility claims with appropriate caution.

​
Summary

Distribute points and frames non-uniformly along a spline, packing more into high-curvature regions and fewer into flat sections.

​
Package Details

Generates a set of points (and matching reference frames) distributed non-uniformly along a spline, packing more points into high-curvature regions and fewer into flat/straight regions.

​
Why not just space points evenly?

Evenly-spaced points either waste density on straight sections or under-resolve sharp bends. This block instead builds a curvature-weighted arc-length reparameterization, so point density tracks local geometric complexity.

​
How it works

  1. Sample curvature densely along the curve (step size = Increment).
  2. Convert curvature into a weight list (higher curvature → higher weight, with a floor so flat regions still get some density).
  3. Build a running total of the weights — a discrete integral.
  4. Normalize that into a monotonic 0→1 curve — effectively a CDF (cumulative distribution function) over arc length, weighted by curvature.
  5. Pick Target Points evenly-spaced levels in that 0→1 range.
  6. For each level, invert the CDF to find the actual arc-length distance it corresponds to.
  7. Evaluate the curve at each resulting distance to get the final point/frame outputs.

​
Installation

Clone this repo or copy the package folder into your nTop workspace:
The package lives at packages/ajayprasad97/points-or-frames-based-on-curvature/.

​
Usage

  1. Open the provided .ntop file in nTop 5.47+.
  2. Input a spline.
  3. Set Target Points to the total number of points/frames to generate.

​
Inputs & outputs

​
Tuning

  • Increment: controls sampling resolution of the underlying curvature scan. Smaller = more accurate but slower. Should scale with curve length rather than staying a fixed absolute value.
  • The internal weight ramp (currently 0.2 → 1.0) controls how aggressively curvature pulls in extra points vs. how much minimum density flat sections retain. Raise the floor for more even coverage; lower it for more aggressive clustering at bends.

​
Known limitations

  • Very short or near-straight curves may need a smaller Increment to resolve curvature accurately.
  • Cumulative sum is computed with an O(n²) custom block, fine at the ~400-500 sample counts used by default; revisit if curvature sampling resolution grows much finer.

​
License

MIT.

​
Source

https://community.ntop.com/packages/ajayprasad97/points-or-frames-based-on-curvature