Is Grasshopper less accurate than Rhino?

[victor_c_cheung]

I m realizing that doing a 4 point sphere with the rhino command prompt gives a much more accurate result than using grasshopper.

Using the same list of Grasshopper generated points:

• Grey sphere that actually goes through all 4 points: Bake points => switch to rhino = > do a 4 point sphere

• Red sphere that doesn’t go through any of the 4 points (much smaller radius): Grasshopper nodes as seen on image

Anyone know why this is happening?

[kts.nguyenhoangloc (Loc)]

Do you follow the same order of points in Grasshopper while picking points in Rhino?

[gankeyu (Keyu Gan)]

As for Sphere 4pt, no. GH should have the same accuracy as Rhino’s, because internally it uses Sphere.FitSphereToPoints provided by Rhinoceros.

[DanielPiker]

Did you check the centre and radius? It might just appear different because of the Grasshopper preview mesh resolution

[victor_c_cheung]

I checked this in different ways, following same order of point selection (though there should only be one sphere able to fit 4 points right?):

All points generated in GH => GH ‘4pt Sphere’ command => radius at 35, sphere doesn’t intersect points

Baked points in rhino => input from rhino as points to GH => GH ‘4pt Sphere’ command => radius at 35, sphere doesn’t intersect points

Baked points in rhino => ‘4pt Sphere’ command from rhino command => radius at 50, sphere intersect points

Checked attached script on different machines to same result… Also have been looking around accuracy / tolerance in GH and Rhino but settings change don t seem to have an impact. would accuracy reduce because the points have not been baked?

attached the script if someone wants to try it.

4pt sphere test.gh (19.6 KB)

[kenzan]

Has anyone come up with a reason? This is definitely happening to me regularly, where the sphere is supposed to be tangent to a cylinder, but is dipping inside the face of the cylinder.

Is this a display problem or actually a geometry problem?

Thanks

[maje90 (Riccardo Majewski)]

Without further information, yes, it likely is just a display problem… which is not really a problem.
Both Rhino and GH have options to increase meshing quality.

Try enable flat shading…

[kenzan]

I realized that my geometry was fundamentally incorrect, and I ended up doing it a different way.

But I will be on the lookout and try flat shading in the future.

Thanks!

[rhino19 (Henry Segerman)]

I just ran into this problem with Grasshopper’s Sphere 4Pt component (in the current beta build of Rhino8 (8.0.23292.13304, 2023-10-19).

This minimal grasshopper script shows the problem well.

The slider on the left goes from -0.1 to 0.1. The corresponding sphere should continuously change with the slider, with the radius getting very big as the slider gets close to z = 0. But it doesn’t: the radius of the sphere never goes above about 4.3.

Baking the four points (with z = -0.011) and then running Rhino’s equivalent command produces the white sphere, which looks much more plausible. In comparison Grasshopper’s output is in red:

[jeremy5 (Jeremy)]

And FWIW the red sphere is not centred on the Z axis where it ought to be, given your first three points.

[maje90 (Riccardo Majewski)]

Definitively a bug.
Rhino.Geometry.Sphere.FitSphereToPoints doesn’t work properly.

sphere fit bug.gh (31.0 KB)

@DanielPiker @wim @Gijs

[Gijs (Gijs de Zwart)]

that looks wrong indeed
RH-77868 Rhino.Geometry.Sphere.FitSphereToPoints not working correctly.

[laurent_delrieu]

There is also this “bug” here

Sphere fit Invalid surface

Hello I think it is a coordinate problem, surely very far from the origin or too big. [image] fit sphere BUG.gh (11.5 KB)

[rhino19 (Henry Segerman)]

Thanks for the replacement code!