Merging coplanar faces causes artefacts to appear in surfaces

When I use MergeCoplanarFace on the attached model I get artefacts in the first face after I merge an adjacent face. I have the same problem if I use MergeAllCoplanarFaces:

The object was created as one half shape and that was mirrored and unioned to get the item shown.

Merge coplanar.3dm (2.8 MB)

(Rhino 7.2.21005.13001, 05/01/2021)



Hi Jeremy - I see that, thanks - the vertex tolerance goes way off on the merged faces.
RH-62250 MergeAllCoplanarfaces



I’m seeing a similar bug.

The command seems to facet some surfaces that are curved. Left Frame before, and Right Frame after the MergeCoplanarFaces command.

MergeCoPlanarFacesBug.3dm (1.1 MB)

Hi Andrew - I do not know if this has anything to do with what you are shooting for in this model but it looks to me like you can vastly simplify your object- granted what you have misbehaves in MergeAllCoplanarFaces, and I’ll see if we can sort that out but cleaner modeling would not hurt, and does the right thing in that command.

FrameSimpler.3dm (213.6 KB)


Thanks Pascal,
I know there are cleaner ways to model that shape. I can loft the mitred profile as you have set up, or I can extrude the original profile, then mitre the corners. But all of those methods take a number of steps. Granted, if it creates cleaner geo, so it might be worth the time.

I find it’s super quick to model mouldings just by using Sweep1 around the perimeter shape (in this case a rectangle). The trade-offs are:

  • the seam line that’s left behind. (which I usually Merge for planar surfaces and edit out for curved surfaces)
  • The quality of the geo from a Sweep1 is never as good as an extrude.

Maybe I should just model it your way…
Or, maybe Sweep1 could be tuned up to generate better geo? :wink:

Hi Andrew - Sweep 1 should get it right, but I was thinking of this curve, mostly -

I’d fix that up - needs a fraction of those control points.


I see.
In fact, that is partially my point. If I do a Sweep1, then Section the result, I get curves with too many control points like that. I must have used that resulting curve to generate another Sweep.

Is there a way that the resulting surfaces have the identical number of control points as the original profile curve? Last I checked it came down to file tolerances which generated those extra points.

Hi Andrew - I took the edge curves here:

which was presumably the input to the sweep. Anyway yes, if you trimmed or took section there is a good chance the curve was made more complex. That was all I wanted to point out - the curve that it appears you used was messier than ideal and I suppose that contributed to the messing up of the surface on merging, even though it should not do that.


1 Like

I would convert any non-linear curve to Degree 5, because that will produce the smoothest transition in the case of S-shaped profiles. Also, degree 5 curves are modify-friendly and their curvature graph always stays smooth no matter how you move its control points. It’s no surprise that they are used in the automotive design as the primary geometry to build degree 5 surfaces.

1 Like