Built-in contour component produces incomplete and missing contour curves from grid surface

I am trying to get the contour curves out of a grid surface from a GeoTiff file and most of them are either missing or incomplete. As a workarround I converted the surface to a low-res mesh but obviously the results did not have enough quality to be useful.

Replies to This Discussion

Permalink Reply by David Rutten on July 29, 2017 at 9:58am

Can you post the mesh? There's very little anyone can do just by looking at a screenshot of the erroneous result.

Permalink Reply by UsernameAndPassword on July 29, 2017 at 10:27am

You can find a 3DM file containing the surface attached to this message.

Attachments:

Surface.zip, 51.6 MB

Permalink Reply by David Rutten on July 29, 2017 at 4:02pm

RhinoWip crashes trying to contour that surface.

What seems to work better is if you use a mesh instead:

Permalink Reply by UsernameAndPassword on July 29, 2017 at 4:19pm

Yes, that is what I already tried but if I zoom in the results are not as good as the few contours I manage to obtain directly from the surface. Also, you cannot directly obtain NURBS curves from a mesh and faking it through the polyline vertexes often results in contour overlapping which is obviously wrong for such representation. Still they look better than the ones I get from QGIS.

I understand that this might have more to do with RhinoSDK than Grasshopper. Anyway, trying with very small portions of the surface also results in missing segments so this component may not be reliable for complex surfaces.

Permalink Reply by David Rutten on July 30, 2017 at 8:56am

Yeah there's something wonky in the Rhino core about this. It somehow doesn't like to perform these intersections. I'll put it on the RhinoWip bug list, but I have no idea when the relevant developer will have time to try and fix this.

I will say that since you're sampling a heightfield image, the information between the sample point is not genuine new information. The fact that a nurbs surface gives you nurbs contours doesn't mean that those nurbs curves are somehow more accurate than the polylines you'd get from a mesh contouring. Your original data is discrete, and the best you can hope for is a smooth interpolation of that data.

Permalink Reply by David Rutten on July 30, 2017 at 9:07am

Incidentally I'm getting somewhat better results if I vertically scale the (nurbs) landscape. This way the horizontal section planes aren't quite so nearly tangent with the surface in so many places. But it's still not correct, some sections are missing entirely.

Permalink Reply by UsernameAndPassword on July 30, 2017 at 10:40am

By your findings it also seems to be a precission issue.

I know that information is not actually stored inside the GeoTIFF as it goes beyond the file's resolution but it is that proper interpolation what makes the later manipulation of the surface easier for my purposes.

RSS