Is there a tool/program/way (!) for this ?!

[Nyro]

Yes, there is a way: transforming the edges on that loop so the resulting shape looks good.

I find myself stating the same fact over and over these days: you have to understand how Smoothing works.

In a nutshell: it uses the average between neighbouring vertices to calculate (recursively) a vertex' new position. If the algorithm subdivides (which the smooth mesh preview does; you can tell it to display the subdivisions in the attribute editor), then the newly created vertex will be placed between two original vertices. The new vertex' position is, as stated above, a result of the average between the two original vertices.
This is a very simplistic explanation, but it does capture the most important aspect of smoothing: The farther apart two original vertices are, the more space there is to average a new vertex, or if you want to visualize it (not technically correct, but makes it more understandable): if you want to draw a smooth curve between two points, that becomes easier if the points are farther apart. If the points are close together, then the curve between them will increasingly look like a straight line: i.e. a corner

This is what's happening to the torus below: When creating a new torus from scratch, the edges are calculated so that, when smoothed, the resulting mesh looks round (smooth). If you then simply add another edgeloop (i.e. vertices) somewhere between the original edges, then you have to position the new edge loop according to the average. In your image, you decreased the space between the original edges (see example above), and so gave the smoothing algorithm less space to average. This results in a more edgy look. If you move the new edgeloop outwards (so that the unsmoothed, original mesh resembles a ring more closely) then the problem can be solved.

[pauljs75]

If you have slide and vector rotation operations in your modeler, I know one easy way. Slide that loop until it's flush against a neighboring one. Then without deselecting it, rotate it on Y about the center of the torus until it's back between the loops.

If your modeler doesn't do it that way, then I dunno. :shrug:

[Nyro]

If you have slide and vector rotation operations in your modeler, I know one easy way. Slide that loop until it's flush against a neighboring one. Then without deselecting it, rotate it on Y about the center of the torus until it's back between the loops.

If your modeler doesn't do it that way, then I dunno. :shrug:

Looks like he's using Maya. In Maya, there's the Slide Edge tool (in the Edit Mesh menu) and finally there's an option in the tool settings for the Rotation tool to set the rotation pivot to the object pivot, which should be at the center of the torus. So your solution should work.

[SadjadMp]

Hi friends, really sorry about returning late, my internet connection date was expired.
I’m sure my explaining was not perfect ! really thanks friend for your asks, especially mr A. D. for explaining perfect.
Maybe I know how smooth works ( I use Maya ), in this case ( torus ) when I add an extra loop there “ neighbor edge loops AND the edge loops behind those “ effects on the new added edge loop together. ( I hope say what I want to say ! ) but I want a tool or way to solve this. And moving and positioning the new edge loop manually or rotating is not so tender ! I have doubt on T-Splines ! but I’m not sure ! and I know T-Splines are not a Poly tool ! I want just a way or tool to have more control on the aria of those two edge loops ! ( but no BURBS ! )

[Nyro]

I fear there is no way to automatically do that. If there is, I'm sorry but I don't know about it.

Certain things in modeling have to be done manually, or else we wouldn't need Modelers, we could let the computer do everything ;)

[SadjadMp]

really yes ! i was strictly thinking about this ! that we need modelers to do these, thanks mr A. D. ;) :thumbsup:

[Zickar]

I tried hard to understand what you are trying to achieve because it seems an important question ... Can anyone really explaing the poblem ? Is it about Smoothness ... I'm a newbie so my answers might not be that accurate but I use Cinema 4D (Which isn't exactly Maya) and if you have enough segments on a model you can then add a Hyper NURBS object to add smoothing

[Nyro]

Can anyone really explaing the poblem ? Is it about Smoothness ...

No, the "problem" here is this:

You create a primitive torus. This is parametrically defined by the program to be represented geometrically correct given an arbitrary parameter value; in the case of a torus, you can define radius, inner radius (which is used to calculate the "thickness") and segments (which for simplicity's sake we'll assume is provided only for the U-direction, i.e. the "ring", not the "tube"). The last parameter is the important one here. It defines how many edges are used to to create the ring-like shape; the more edges, the smoother the ring and thus the more accurately it will resemble a torus.
The algorithm used to create this (parametrical) primitive will make sure that the segments are placed so that the geometry matches a ring like shape. Basically it simply places as many edges as specified into an evenly distributed, circular pattern.
Smoothing this will result in a perfect torus. Why? Because as stated above, the edges are placed in a circular shape. Imagine a circular rail on the ground. Edges may only be placed on that rail and are bound to that rail. To retain the circular shape, edges must be moved along that rail.
What Sky4LOS did was insert a new edge inbetween those provided by the parametric calculation (for the C4D user: after hitting "C" and making the torus an Editable Mesh). This new edgeloop does not lie on the rail! That is why, when the geometry is smoothed, it does not look like a perfect torus anymore.

Now, Sky4LOS' question was: is there a program that can determine automatically where that rail is and place a new edge on it.

The answer is: no. The program provides the initial parametrical primitive, but it's called a 'primitive' for a reason. If you want to further refine it, you'll have to make sure yourself that your actions do not change the shape of the geometry.
In this specific case, he started with a primitive torus. The edges were well placed (on our imaginary circular rail), so the smoothed result looked like a perfect torus.
Then he added an edge. The original, unsmoothed geometry looked the same (because the edges are interconnected by straight lines/edges, so adding edges implicitly adds vertices along these lines, without changing their shape). But the smoothed result looked different. That's because the smoothing produces a perfect torus only when the original edges all lie on the rail. The new edge does not lie on the rail, so the smoothed result strays from the perfect ring.
This is why you manually have to move the new edge so that it lies on our imaginary rail.

But even then: since you only added one edge between a specific segment of original edges, the mesh has more density here. The smoothing algorithm will take this into account (as illustrated in my previous post about "averaging").
Let me illustrate:

You created a primitive torus with 8 segments. So you have 8 edgeloops on our "rail", we'll call them A, B, C ... H.
You insert a new edge between B and C. If you now apply a smooth (C4D: hypernurbs) algorithm, it will interpolate (depending on the algorithm; we'll just say "average") between neighbouring pairs of edges, i.e. [A B], [B C], [C D] etc. Since the region between B and C is now more densely defined (by adding our new edge), the resulting smoothed curve between [B C] will look differnt from all the other segments ([A B], [C D], [D E] etc). This is simply the way smoothing works. If the new edge is perfectly placed on the rail and in the middle between [B C] then the result may look virtually the same.

I've written another 100 pages just to get a simply point across:
Plan ahead when working with Smoothing (Hypernurbs, Turbosmooth, polySmooth etc.). Your topology (that is, the way your geometrical mesh is constructed, how individual vertices are connected with edges) plays a huge role in defining the visual result after smoothing. Altering the topology will alter the result.
In Sky4LOS' case: if you need more segments (because you want to extrude a segment, cut a piece out or whatever), then add them in the parametric stage. Or else add them manually later but you'll have to push the vertices around to retain the perfect shape.

I strongly encourage you to study the basics of subdivision algorithms (which are used in smoothing). Create simple planar geometry, and a duplicate/instance of the geometry next to it which has smoothing applied. In C4D, create the geometry, create an instance and place the instance in a Hypernurbs. Then modify the original mesh, moving and adding vertices. Study the result of your manipulations. Also, if you really want to know what's going on, here are a couple of names you might want to google: Ed Catmull, Leif Kobbelt, Jos Stam.


Wow this must be my longest post so far

:beer:

[Zickar]

I AM GOD

Maybe I'm just putting words into your mouth but yes you are a god .. at least IMO ... I initially didn't understand your post but your enthusiasm has encouraged me to go on and in the final example I actually understood what was going on ... I hope I haven't derailed the topic but in all truth thanks for explaining it mate ... a real help

[Nyro]

my fault if I wasn't making myself clear enough. Glad you understood some of it after all :)

And there is no 'derailing', only 'adding more to the knowledge pool'.

[RageOfAges]

Seems to me like all you've got to do is go into local transform and move that edge along its axis until it sits in the right spot.

Anyway like Nyro said: When working with sub-d's it is important to plan ahead. Especially when you are new to them. Once you get more experience with them it will become second nature.

-rage

[CGTalk Moderation]

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.