When creating a surface that closes at a point, I have problem making the
surface around the point to be clean. (as in the right object below, using networksrf)
Using curvaturegraph, I can see that where the tip intersects the lines are not clean, but wasn’t sure how to clean it. (wanted to make the end three control points on both curves to be in a line to have continouse curvature, but wasn’t sure how to do like a set point function in a linear way)
As a way forward, I just trimmed out middle part and used blendsrf so make the surface labelled trial 2.
Is there a good way to make the initial curves clean?
I was thinking about changing cplane so the curves are on a plane to start with, but after that still stuck in setting points to a line…
hi Toshi, does not look so bad, i would go with Trail one. the only thing is that the curves dont actually touch each other in front as pointed out in the picture, i wonder how it succeeds in network surface like that, some peculiarity there. if you get those to touch each other maybe it buffs out whatever you are unhappy with.
Don’t have a simple answer for you, unfortunately, but wanted to use your frustration to illustrate this type of scenario - capping a closed loft - as an area that could use some love. Such remains a process nuisance.
Results always depend on the closed shape, and desired termination tip shape. Some NURBS studs may provide a patchwork solution, but you’ll spend forever on it, and exploration/editing it will be pain, and still may never get clean enough results. Easing and speeding the efficiency of this type of fundamental termination would be a big value add.
Future ‘modeling update’ improvements will be welcome…
Until then, alternatives are likely your best bet if a clean tooling ready surface is the goal.
A. TSpline design the tip, and TSpline ‘match’ the end of the TSpline to the end of the nurbs loft.
B. There are new patch workflows too, such as the following:
The input curves for NetworkSrf do not have to intersect, but must be arranged in two sets. There are some limitations on the arrangement of each set of curves.
For those interested in how the surface algorithms work here is my guess on how NetworkSrf works: it first rebuilds the input curves with certain characteristics, then creates a surface similar to a Coons bicubically blended patch. The input curves are rebuilt as degree 3 curves with each set of rebuilt curves having the same number of control points. The number of control points is determined by the tolerances in NetworkSrf. Then the surface is determined as the the sum of a lofted surface through each set of curves minus the edge curve surface from the edge curves.
cool, not a single curve touching the other… like patch… but even though NetworkSrf can cause non joinable surfaces at adjacent surfaces depending on the tolerance it still is better than patch which i never got to join into anything adjacent.
anyway something to play with and to comprehend now, thanks david.