This strays from the typical tutorial in that I don’t go step by step through the creation of a model. I wanted to quickly show some of the useful ways the XSI 1.5 polygon tools and other features can be used in the building process.
It’s all about the edges:
Edge operations are easily forgotten or unused especially for those of us who are used to modeling in 3.x. Edges are extremely useful and fast for certain modeling tasks. Look at the following figures. When using the standard approach to poly modeling through duplication, trying to scale polys down a curved or uneven line usually gives undesired results. If you select the edges and scale in local mode, you can scale each segment along its local axis. Notice that the icon shows how the local "x" cartesian coordinate follows the length of the segment. This local behavior is different than local behavior for points.
Note: The transform icon can be misleading at first because it looks like it represents a common center for all selected components. When actively scaling you’ll see that the icon is a visual indicator for only one of the edges so that the screen doesn’t get cluttered with an icon for every selected component.
Point operations:
Sometimes local mode isn't what you want. In 3.x, a handy way to align a group of points to a straight line was to align the object center to some tagged points by using the "center2vertices" plug-in, rotating the center if necessary, and then scaling in tag mode. XSI has a quicker solution for many situations. Tag your points and use global mode with COG enabled for scaling. The reason I don't use local is that all the points in a selection will contribute to a local center. This often skews the center to an awkward angle. Global mode gives you dependable world coordinates and COG adds the necessary offset.
Mixed mode modeling:
Being able to mix your display modes in a viewport is very useful. Take a look at the following figures. Often, it can be very hard to visualize the junction between 2 surfaces you want to merge. My process involves snapping the points of one surface to another and adding or deleting polygons on both surfaces to get the boundaries ready for the merge. The edges can get very confusing on complicated meshes. Mixing the display helps to differentiate your surfaces. In the images below, using a wireframe and a shaded object makes visualizing the edges very easy. Also, you can see through one of your surfaces to get at those tough angles. The image on the right is taken from inside the head looking down through the roof of the mouth.
Snapping power:
For 3.x veterans, snapping to a surface (Drop points and Slide points) was an incredibly useful plug-in set that appeared in the later versions of Softimage. Despite the usefulness of these effects they could get a bit heavy and were somewhat limited by the plug-in architecture. Surface snapping in XSI is just another snapping option. Not only is it many times faster and trouble free, you can also snap on any number of surfaces. When you combine this feature with other operators like the improved "extrude along curve", the workflow is very fluid and powerful. In the left image, I have snapped a nurbs curve along the wing bones. In the right image, I use this nurbs curve to extrude a polygon to start the wing's membranous surface.
Snapping for layout:
Another useful application of snapping is for laying out the flow of your poly edges. In the image below I’ve got mixed viewing mode on, some simple underlying bones for structural reference, and a basic body model for the dragon. I've created linear curves for the trapezius muscle by snapping the points to the surfaces underneath. These curves can now act as snapping guides for polygon edge refinement or as skinning curves for new mesh surfaces.
D - indicates the "blocked in"
deltoid muscle.
C - is the clavicle
T - indicates the lower edge of the trapezius muscle curve. It has the red(tagged)
points and extends up into the neck.
S - is the scapula
Body building:
I've used several modeling approaches on the dragon. Because of XSI's polygon modeling strengths, I also refined one that I couldn't use efficiently in 3.x. For building the wing muscles, I dropped the standard surface refinement techniques in favor of the following method.
I tend to think in terms of underlying muscle and bone structures when defining the skin surface. In the following images, I started with very simple cubes and blocked in several muscles of the wing appendage. Boolean unions were then used to join the muscles together.
A lot of cleanup is necessary and the process was tedious at times, but I got a much better result than I would have otherwise. In brief, cleanup involves getting rid of extra triangulating edges resulting from the boolean, unifying the flow lines of the edges across the different muscle regions, deletion of internal polygons, and softening of some of the muscle junctions to accommodate for a fat layer and the skin thickness.
The following image shows a quick look at how the muscle subdivisions look before they are merged together. I've added all the highlighted muscles to a group so that I can quickly adjust them all at the same time. Simple scripts and hot keys can also be defined for faster workflow.
The unlimited uses of groups:
My first impression of groups was that they were a very convenient addition to my old 3.x workflow. After using them for several months I've realized that this "little" feature is much more than it seems. You can group objects for material sharing between objects and clusters, defining subdivision surface properties, setting discontinuity angles for bevel effects, defining one quickstretch operator for several clusters, and mixing display modes to name just a few possibilities. There are a lot of properties that can be set through groups. In 3.x we were limited to hierarchical inheritance for effects and deformations. Groups get rid of this limitation in a big way. It's a huge feature and easily underestimated. In the image below, I've grouped objects for a localized lattice deformation.
Rendered images:
Here are some rendered images of the dragon. Unfortunately it's just a project for fun. He'll probably be a work in progress for a long time.
Well, that's it for this round. I hope these tips prove to be useful.