Hello,
I wanted to ask some of you about subd modeling. I have watched the gnomon organic modeling videos in which they use image planes and NURBS to model separate surfaces (eyes, nose, mouth, and head) and unify them using patch modeling. This can be very tedious and take a long time.
So I have tried doing organic modeling through subd. This method is easier for me to rig because you have just convert the model to poly for an instant wrap deformer and takes less time to model since you don't have to make patches. But you do have to do UV layouts which is cool with me.
The main problem I have with subd is that when using an image plane as a reference, I model using one singular surface. For instance you start with a poly box and keep splitting faces to rough out the head and then further onto the eyes. More and more detail but less intuitive than nurbs modeling since you can just work on separate details and just unify them to make one smooth surface.
So then I saw this tutorial. It is pretty simple but catches my interest:
http://www.geocities.com/susantio/head01.htm
The artist starts off just like the gnomon video. He uses different nurbs surfaces to do the eyes, ears and head. Then he converts them to poly using control points. He deletes some faces and combines edges to unify all the detailed parts into one poly model and voila! He converts to subd to make a pretty smooth subd model.
Now I tried this technique on a much more detailed model (much more nurbs spans than his). The main problem is when I convert to poly and try to unify the edges, there are some messy tangency issues that need to be fixed and even if you do it doesnt look that smooth. You can see some ridges of where the detailed parts were attached to the head.
Basically I wanted to ask about what you guys think of the method and what your preferences are and your techniques. I have seen many of the "model a subd hand" and hate it when they do refinements on levels 1 and 2. The reason why is because when you create the convert to poly wrap deformer, it doesnt have that one to one vertex as a model with no refinements. I have also notice that subd models I have rigged with refinements on levels 1 and 2 lead to the spike effect on certain poses.