Polygon modeling vs. NURBS modeling. Reel Header render update.

I began working on Robopig and began using the box-modeling technique, which uses a simple polygon primitive and where detail is slowly added via extrusions and edge loops.

I didn’t add in much detail and will show post the early phase of the modeling here.  The snapshot below shows the un-subdivided model.  I attempted to model Robopig through a single mesh but now have decided to remove polygons and also incorporate parametric NURBS objects as it will allow for a much quicker turn-around on the models.

I believed that if I saved the integrity of a single mesh, this would give me a more robust model for later phases of the pipeline such as rigging and texturing, and this may be the case.  But it is hard to model hard-body geometric shapes through subdivision models as adding the necessary resolution will actually take away from the curvature that would naturally result from subdividing my polygon model.  Polygon modeling is better suited for organic models such as animals, people, etc.  Where NURBS objects are better for hard-body objects via tweaking control vertex points and such.  This is my understanding so far.

Below is the subdivided model.  Notice the edges are smoothed out (with the exception of the ear).  Like I mentioned above, I will remodel using NURBS objects as I am not happy with the results of polygon modeling.

Aside from this, I decided that I would render the latter half of my Reel Header sequences on a secondary computer to speed up the process.  I discovered that this does not work as the Pyrocluster behavior is stochastic and will not match up to my other render from a separate computer.  Perhaps I should have tried to cache the Pyrocluster emissions.  Anyhow, it seems that I will have to wait for the renders to finish which at this point has completed frame 170 (out of 260)  for a total of 575 hours.  I estimate that I am rendering 3 frames every 24 hours.  I really need a render-farm.