Surface Edit Points

Might’ve been asked for a few times in the past, but could it be feasible to request display and editing surface edit / Greville points even just as a test command?

Given it’s ‘sort of’ possible in the SubD workflow.

Hello - these are technically possible for surfaces but from what I understand from the bigger brains, the math is such that it is very hard, or impossible, to keep things under control in a useful way in the UI.

-Pascal

@pascal Has edit/Greville points for surface modification been seriously reviewed in the last few years? Advancements in computer hardware may have changed the landscape. The primary challenge using edit/Greville points to modify surfaces compared to for surfaces would be the much larger matrices to be solved for surfaces. In the past such matrices may not have fit into memory which would drastically slow the solution but with 16GB common it should be less of an issue. Also solution of such matrices is generally amenable to parallel processing using multiple cores/threads.

I struggle to understand what the UI issue would be other than possibly sluggish response. A user would want to turn on the edit/Greville points, then move one or more and have the surface change shape appropriately.

Hi David - what I recall is that moving the edit points on a surface gets ‘squirrely’ very quickly - the surface is apt to shoot off into space - not that I have tried it, since it is not exposed, but I think that was the hesitation - it just would not be useful. But I will ask a developer to clarify…

-Pascal

Would he squirrely nature be comparable to, for example, interpolating a curve on a surface, whereby by after a couple of points are drawn the interpolation will become a bit wild?

My guess is the origin of “moving the edit points on a surface gets ‘squirrely’ very quickly” is due to the common practice of using higher degrees with surfaces with a given number of control points rather than some fundamental difference between surfaces and curves.

Oscillations with edit points become pronounced as the degree increases for both surfaces and curves. That is fundamental to how spline curves work, and is one of the major reasons working with control points is usually preferred.

Exactly the same based on my understanding of the underlying math.

Hi All - the deloper I chatted with last week thinks it could work OK- at least for relatively minor edits. I’ll get it on the pile, Meantime, HBar may be of some help.

-Pascal

1 Like

Yeah, this isn’t something I would anticipate using on a surface anymore than, say, 10 x 10 degree 3 in u and V.

Worst case oscillations are for a single span degree 7 curve or surface. Here is what happens when moving one edit point on “8 point” curves of degree 1, 3, 5 and 7.

In contrast if the number of spans is equal the oscillations are similar.

And for single span curves.

With edit points the lower the degree and the larger the number of points is usually better with less noticeable oscillations. See above.

Yeah that’s reasonable. In this case, I am most concerned with edit points on the surface edges, and in some cases on trimmed surfaces too, in the trimmed away region.

here a topic discussing this slightly