Misbehaving patch normal

[R Reynolds]

This is the first time I've ever seen this (v19.0o). Attached is a model of what will ultimately be the hinge on the back door of a 1950's vintage truck trailer. The attached image shows that it has one strange patch that doesn't respond well to "refind normals" which has the effect of pointing its normal opposite to all its neighbours and messing up the smoothness of the surface. The only way to correct it is to select just the patch and flip it's normal. It's not much of a bug but can Steffen fix it?

Fruehauf_hinge.mdl

[robcat2075]

I dunno, i typically always have to fix some patches manually.

But for me it's smoother with the patch pointing in.

[serg2]

You need to check the durations of the curves framing this patch. Sometimes there are spiral structures that do not allow a patch to be closed.

[serg2]

I looked at the file. My assumption turned out to be untenable.

 

[yoda64]

Try "Renumber CP's"  , than "Refind Normals" .

I think the problem is coming from some insert operations, needs more observation.

[robcat2075]

After ReNumber CPs and Refind Normals, the normals were all pointing outward but the smoothness of the shading got worse...

 

This is also interesting. I wondered if detaching and re-attaching spline intersections might change things, but after the Renumber and Refind, Detach became unavailable for this intersection...

[yoda64]

Try with the attached model

 

Fruehauf_hinge_renumber.mdl

[robcat2075]

54 minutes ago, yoda64 said:

Try with the attached model

That model seems to have all the normals outward already, so Renumber and Refind Normals doesn't change anything?

It also still has the problem I noted above with the patch that shades well only if its normal is pointing in.

 

[yoda64]

This is the original model, after Renumber and Refind Normals.

For the shading problem .

Have a look at the bias values for CP#219 and CP#223 , they should have identical values to achive the same shading.