How do i better quallity of surface: pwr rail sweep

[JohnyCageII]

Hi,

i am working on new plate of car. My question sounds: How can i make a constant distance between each "line" of rail sweep surface ? Mentioned problem is described on the attached picture. Thanx for your time.

[JohnyCageII]

described picture

[Clawhammer]

My experience with surface modeling is quite limited but sometimes "Use given rails" would give you a more constant distance. (It's under "rail tangency") Probably there is another way that I can't recall. Something about increasing or evenly spread these knots, perhaps by rail synchronisation, but I doubt it would get much better then that!

This is the main reason why I want to learn surface modeling, I cant boolean my way out in my next project!

[JohnyCageII]

Firstly thanx for your answer. You will be convert to "box" modelig style or what do you mean "This is the main reason why I want to learn surface modeling".

[Clawhammer]

Well I do model 95% with Powernurbs but I use a lot of solid operations, this has mainly to do with the nature of my work (putting 300 windows in a ship etc) but I would like to slide more to the design and smart surface stuff. :)

[PiXeL_MoNKeY]

Johny,
Can you post the file that it comes from? Be a little easier to investigate, and possibly be able to provide a solution.

-Eric

[nPower_Michael]

That surface looks like it has a pretty high quality already. If you want, you can go into the Surface Analysis rollout and turn on a Gaussian curvature map and see how smoothly your curve varies (ideally you will see a mixture of red, green, and blue without any really sharp variation, if you see all green it usually means somewhere there's a small (spatially) but very steep variation in curvature, which is usually a bad sign).

The lines you are seeing are the surface knots. They essentially represent the places where given rows or columns of control points stop influencing the surface. In other words, if you were to go into your shape curve and start dragging a control point up or down, the resulting sweep would only change shape in a certain range which is bounded by two of the (non-adjacent) knot "lines". This is the "local influence" property of NURBS which makes it so intuitive to work with (if you move a control point, it only affects the curve in places "near" the control point).

Now, the reason you are seeing that gap is a little bit tricky: By default, curves are created with uniformly spaced internal knots in parameter space. However, the knots you are seeing are evaluated and represented in 3d space. Further, the knots at the ends of an open curve have a multiplicity built in (ie multiple knots in the exact same place in parameter space)... which 'unbalances' the functions that determine how strongly each control point affects the curve at each point in parameter space. One result of this is that for the default knot set up, and fairly nicely arranged control points, most of the internal knots evaluate to be vaguely close to a control point.. but the second and second-to-last control points do not have a knot that lives 'close' to them... (for the sake of full disclosure, I do not wish to imply that each control point 'owns' a knot. Its just that for uniform knots and very nicely spaced control points, the internal knots end up 'close' to control points due to the way the basis functions balance out).

A watered down version of this would be to say that (at least for degree 3 curves/surfaces) the second/second-to-last (or first/last INTERNAL) control points have the most influence on the parameter space in between the end point knots and the first/last internal knots... while all the in between control points have reasonably heavy influence on parameter space regions that are close to a internal knot. (Note that this only works out if the knot spacings in parameter space ARE uniform. If the parameter space knots get moved around, all bets are off on this topic).

In summary, the surface you have is actually quite typical of a nice high-quality NURBS surface, and you shouldn't worry too much about the end 'gaps' in knot spacing as they arise naturally.