Rendering

Spalter’s chapter on rendering clearly demonstrates the issues involved in rendering a 3D model. A rendered scene is a raster image derived from a three dimensional model. It is a two dimensional view or screen shot of the model. The two qualities that must be present for a rendered image are material properties for each object or surface and lighting.

Several elements affect the designation of materials in a rendered scene: light absorption, light reflection, light transmission and texture. Light absorption determines the color of the object or surface. There are two types of light reflection. Diffuse reflection is not dependent on the viewer’s angle of view. Diffuse reflections reflect light equally from any viewing angle. Specular reflections are dependent on the viewer’s viewing angle, and reflect light only in a “direction equal and opposite to the direction of the light shining on it.” (p.261) Light transmission determines the transparency of an object. Texture provides visual complexity to rendered objects. Texture mapping involves wrapping a 2D image around an object, creating the illusion of pattern. Solid textures are through body textures and are present even when an object is split in two. Bump mapping provides a way of achieving the look of a modulating surface without actually displacing the surface, while a displacement map is used to displace the surface and change the geometry of the object. (p.266)

Lighting is crucial for a successful render. Without lighting, none of the geometric or material qualities of a model will be visible. There are five main types of lights in 3D rendering programs: ambient light, point sources, spotlights, area sources and directional, remote lights. Ambient light is general non-directional lighting that lights all objects evenly. Point sources emit light from a single point, such as a candle flame or light bulb. Spotlights are shaped light, that mimic many everyday lighting sources. Area sources provide light from an entire surface, such as a florescent fixture. Directional remote lights simulate distant light sources such as the sun. These sources are certainly directional, but their distance from the model, allows them to light each object evenly from one direction.

Photorealistic rendering is complex and requires the computer to calculate many complex equations. For this reason, many rendered images can take quite sometime to render, depending on the complexity of geometry, materials and lighting. Raytracing and Radiosity are the two global rendering methods that Spalter discusses. Raytracing works by “considering the paths of light rays as they bounce around a scene illuminating objects.” (p. 279) Radiosity is more accurate and incorporates more surface interactions than raytracing. Radiosity is not view dependent and therefore does not render specular reflections. A combination of raytracing and radiosity are necessary to achieve a rich environment with reflections and specular highlights.

I work with rendering almost everyday, so I was very familiar with the concepts Spalter mentioned in her chapter. I was also interested in the research at Cornell Computer Graphics program. It was nice to gain a better understanding of the vocabulary that I tend to toss around lightly. I think reading this chapter and the web articles about Cornell and Greenburg, have given me a greater intellectual understanding of the tools that I have used intuitively by trial and error.

I think photorealistic rendering is just one of the delightful processes of modeling and presentation drawings. Certainly it can be used to understand a design intent, or to grasp how light will affect a space, but I don’t really think that it is necessary in every design case to have photorealistic images. There’s a real “wow” factor when presenting a photo image to a client or even in house to fellow designers, but given the amount of time it takes to render these images, it doesn’t seem to be useful as a design tool. Plus there are so many tricks to getting accurate lighting or even unreal lighting to make a scene more dramatic. This doesn’t inform design decisions as much as impress clients. However, the process of making them is fun! And that might be enough to keep photorealistic rendering in the realm of architects and designers and not just a job for illustrators or artists. In the near future, when it takes five seconds instead of five hours to create accurate renders, then it will be feasible to utilize these images in the earlier design process. But currently I think their rightful place is as a presentation and sales tools.