Subdiv Modelling

Seamless3d has supported "polygon subdivision" (subdiv) modelling for nearly the same amount of time it has supported nurbs but very little work has been done on subdiv in comparison to nurbs modelling and so very few features have been developed for it. However polygon subdivision offers a considerable amount of power with only a bare minimum of features and if the user begins with a pre-made structure all the user has to do is tug at the control points to modify the shape. If anyone wants to play with a polygon subdiv model using Seamless3d they can. Just download the subdiv_av.smls file that is based on the same model used for the nurbs patch avatar tutorial.

Dragging Control points

To be able to drag a polygon control point the Subdivision node must be selected in the scene tree:



Any of the corners in the control polygons (the cyan coloured polygons) can be clicked on and dragged.

Selecting a polygon

Whenever a control point is selected a control polygon is also selected:



The red and green dots mark the selected edge of the selected polygon.
A neighbouring polygon that shares the slected point can be slected by pressing W or E

Deleting a polygon

The selected polygon can be deleted by pressing delete.

Extrude Poly

5 polygons can be inserted in the selected polygon by pressing S.
After this command the extra polygons inserted will not be revealed untill the 4 new control points are dragged:

Adding an ajoining polygon to the selected edge

Press Ctrl + M

Joining a control point to another control point

Press J. This command joins the selected control point to the nearest control point from the perspective of the view. Care must be taken not to accidently join to a control point hidden by the mesh.

Changing the density of the mesh

The Subdivision's cycle field can be changed to change the nunmber of subdivision cycles performed. Each cycle increases the number of vertices by about 4 times.

Why I use Nurbs not Subdiv to make my Avatars

I understand there have been hot debates when comparing nurbs to subdiv on forums so let me say first I find subdiv very exciting and I love its simplicity and that I would like to experiment more with it. I have not had enough subdiv modelling experience to be able to make a fair assessment of the 2 methods but I will explain from my own perspective why most of my focus has centred on nurbs not subdiv to make avatars so far.

For high density modelling Subdiv and Nurbs modelling are generally preferred over "poly modelling" (not to be confused with polygon subdivision modelling) because you only need a few control points to control many polygons. Both nurbs and subdiv methods are used for creating organic models by professionals but subdiv is said to be the most popular for organic models. For real time animation however, poly modelling is said to be the most popular method. This is because poly optimization is very important for real time graphics and poly modelling lets the user have complete control over the final structure of the mesh. For skinning it is even more important to have control over the poly structure because vertices should be denser in the joints than in the rest of the model. Because of Seamless3d's NSPE (NURBS surface polygon editing) feature, the nurbs modeller can also have complete control over the poly structure. NSPE has the great advantage over poly modelling in that it lets the user edit the poly structure much more quickly. Because nurbs surfaces are polygon independent, NSPE lets the user edit the structure without the concern of accidentally modifying the shape in the process. Nurbs modelling concepts may be a little more complex than subdiv for making organic models but once a little time has been invested in learning how to use Seamless3d's FuseSurface node, creating organic models that have limbs and fingers that branch out is easy. In comparing my subdiv av example to my nurbs patch avatar (made from 3 patches), the subdiv model has more control points and it creates more polygons for the final mesh than the nurbs model and yet it does not look as rounded as the nurbs model One of the reasons for this is because the nurbs patch av uses weighted control points that makes the mesh naturally form circular curves. Considering that the robot only takes about 15 minutes to create the static model using nurbs, nurbs modelling without a doubt can be a very fast way to model.

To develop 3d modelling skills one has to invest time. To invest time one must be confident in the software and the methods used. I have heard people say they wish modelling could be as easy as modelling clay using one's hands (and I can remember wishing this too) but imagine making a shape as complex and as smooth as my nurbs patch model out of clay in only 15 minutes and then imagine trying to animate it.

Examples of well known organic models made by professional artists using nurbs:

Shreck
Flushed Away

---

[3d Modelling Software] [Tutorials] [Forum] [Features] [Download] [Gallery] [FAQ] [Worlds] [Avatars] [Links] [Thyme]