Chapter 7

Polygon Modeling

page  *1  2  3  4  5  6  7  8  9  10  11

    A subset of Softimage’s Modeling tools, the Polygonal Modeling tools enable you to manipulate objects made of polygons, as well as individual polygons themselves. Sometimes using the Polygon Modeling tools is just the easiest way to make the objects you need for your rendered scenes, but other times Polygonal Modeling is the only way to precisely control what you get. Using the Polygon Modeling tools you can draw each polygon precisely where you want it, control the exact location of every vertex, and keep total control over the number of polygons in each model (see Figure 7.1).

    Figure 7.1 Several models with different levels of polygonal detail.

    Every polygon in Softimage can be transformed separately, colored and textured individually, and even animated individually. The POL mode restricts all the Transformation menus to work only on polygons, and the Polygon menu cells add even more polygon-specific functionality.

    In this chapter, you will

  • Create and edit polygons, point by point and edge by edge
  • Model from poly primitives
  • Copy, duplicate, and transform polygons
  • Extrude polygons
  • Make polygonal terrain models
  • Round and bevel poly objects
  • Perform Polygon reduction
  • Making an entire scene by drawing and connecting individual polygons is like building yourself a house out of toothpicks, or knitting a car out of steel wool. Sometimes that's what it takes!

    Polygon Modeling for Games

    For real-time 3D scenarios, however, all this possible detail can be too much of a good thing. Polygonal modeling has become increasingly important over the last few years with the popularity of real-time 3D games and VRML (Virtual Reality Markup Language). When a game or VR simulator needs to display a 3D environment interactively to the user, so the user can move freely about the 3D space, the computer rendering the scene must be able to redraw the scene very rapidly, typically 24 to 30 times each second. Because the computer must evaluate and perhaps draw each polygon in the scene every single time the scene is refreshed, keeping the polygon count low is critical to smooth gameplay and good computer performance (see Figure 7.2).

    Figure 7.2 A real-time image and the geometry behind it.

    In game development, the programmers who build the code engine typically perform tests to determine exactly how many polygons the engine can draw per second, on a reference machine that the game players are likely to have. Then the game designer sets the frame rate at which the game must refresh for smooth action, and the total engine performance number is divided by that frame rate to deliver the polygon budget. The Polygon budget is how many polygons can be on screen in each frame, before the game engine begins to choke and can’t draw them all in time. After the polygon budget has been determined, the game designers list all the elements that need to be onscreen at one time, and divide the polygon budget up into a budget for each element.

    For example, if my game engine can draw 90,000 lit, textured triangles per second, and I want the game to play at 30 frames per second, the polygon budget is 3000 triangles per each frame. If I assign 1000 triangles to the terrain, 200 to the backdrop, and decide that five enemies can be onscreen at once, each with 200 triangles for the enemy and 100 triangles for weapons, shots, and so on, that leaves me with 300 triangles for the main character.

    Modeling an attractive, convincing main character in less than 300 triangles isn't easy, but it can be done by creating your character from scratch with the Polygon Modeling tools.


    For those of you interested in video game technology, there is usually a texture budget as well. A typical game engine can draw a polygon without a texture in about 10% of the processor cycles required for the same polygon with a texture, so reducing texture usage can help performance.

    Also, as the textures get larger, the time required to map them onto polygons and the memory used to store them increases. Texture maps in video games are often very small, ranging from 32 pixels wide by 32 pixels tall to 128 pixels square for really big ones (See Figures 7.4 and 7.5). Some game engines even store multiple sizes of each texture to assist in drawing the textures at different scales, called MIP mapping (see Figure 7.3).

    Figure 7.3 A real-time texture library with many textures in a single file.


page  1  2  3  4  5  6  7  8  9  10  11