Before I go any further, I really want to get an idea of where I’m at, from a SL standpoint. Working in 3rd party applications is all fine and good, but the end result has to work in SL, so it’s always good from time to time to check that what you’re doing in your application, is going to work for SL too. I actually do this semi-regularly, but today, I’ll take you along since we’re at a pretty big milestone…
I went ahead and converted all my Bezier and NURB parts to mesh, and then dumped the whole thing into SL, just to see how it looks. If I do say so myself, it’s looking good:
Some parts may appear strange in the picture, remember on many pieces we only made 3 sides, so when viewed from the back side, they appear transparent. That’s completely normal. When worn, that side won’t be visible, and would just be wasted polygons.
I checked my polygon count, and It wasn’t bad at all. But, i still have a way to go on this design, and it was higher than I’d like. So I looked closer to see what I could see.
A great thing to do when evaluating your work (and other’s work) is to go into wire frame mode with the SL viewer. There’s an option for that on the develop menu, and a handy shortcut (Ctrl-Shift-R).
It can be hard to see with everything rendered as triangles, but with some squinting you can see just how everything is put together. Looking at the harness, I immediately see something I didn’t expect.
The straps are quite a bit denser than I had originally planned, and so many polygons are neither wanted nor needed. Since there are so many straps, that can add up quickly. Looking back in Blender can probably give some answers.
Thinking about how the strap profile was designed, can see now what happened. I only used a few control points, but to get them to create the surface features, I had the number of divisions set to 3. This really isn’t good because that creates a lot of faces between each set of control points, and i have my control points pretty close together.
Looking at the strap in Blender you can see just how many faces were created along the cross-section:
Though at first it may seem counter-intuitive, the answer is actually to add *more* control points, so we can lower the number of divisions. When lowering the divisions to 1, you can then control exactly where your faces divide, as it will be *only* wherever you put a control point. This is exactly what I want for the profile.
Now our cross-section for the straps looks *far* better. And once I get busy making this change across all the straps, it will reduce the poly count significantly. Back on track!
Best of all, we didn’t lose any detail doing this. I still have the molded channels and raised center that gives the strap its distinctive look. There’s other ways to reduce poly counts as well, as I use them, I’ll be sure to point them out. Until tomorrow…