Rendering à la Sechman
The more these applications develop, the less obvious becomes the point at which we switch from one to another. Our general approach is to use one application until it is no longer efficient and then to move on to the next stage.
When the architects give us design changes mid-process, we often modify the geometric model within the rendering application rather than return to the 3D CAD modeler, though this varies between projects.
Controlling the Building Model
Our projects are often large, so the rendering process can be slow and unwieldy if we do not plan ahead. Rather than use the architects' digital model, which is usually far too detailed for our purposes, we create our own digital model, working to keep the polygon count low. We have learned through experience how to accomplish the most with the fewest objects. For instance, we only model those sides of the building that will be visible in the final rendering.
How we import the AutoCAD files into 3ds max is a factor in determining model size and accuracy. Although AutoCAD has the capacity to export 3ds-formatted files, we find it best to import AutoCAD's normal DWG files into 3ds max because that way they retain all the attributes of the original CAD model.
In AutoCAD, the basic polyline entity delineates objects very efficiently. The simple underlying model is geometrically accurate and it arrives in 3ds max with all surface normals facing out. This means, essentially, that the planes face the correct direction, so only one side needs to be rendered. In cases where both sides of some polygons will be seen, you can assign those surfaces two-sided materials with the material library.
One technique that is invaluable to our work is to render the overall building scene reasonably well, with basic light and material, and then to isolate and work separately on particularly important objects. For this we use alpha mattes, or masks, which allow us to work on certain portions of the building independent of the rest of the rendering. The images must be rendered with an alpha channel and saved as 32-bit TIFF files.
These masked areas are then moved into postproduction as a series of renderings. They appear as layers in Photoshop, where you have the most control over the rendering of surfaces for light quality, color, and a host of other controls. [Note: This manually manipulated rendering does not provide physically accurate lighting, which can be generated by advanced rendering systems like Lightscape or Radiance. — Editor]
When compositing rendered layers, one of the most valuable tools in 3ds max is the "Box Select" found in the rendering dialog box located on the top right menu bar. This tool allows you to precisely align renderings for later compositing in Photoshop. It also allows you to compose your rendering as a photographer would compose a photograph.
Box Select is much like a view camera in that you can shift your composition while maintaining a constant size and aspect ratio. This eases the process of setting a viewpoint close to the subject while avoiding the sense of distortion given by looser three-point perspectives. The tool is useful, for instance, in composing views of tall buildings.
To take advantage of this tool, when importing rendered TIFF files into Photoshop, make sure that each is composed within a Box Select of identical size and location within the scene and of identical aspect ratio as the final image.
Then, after you open the TIFF file in Photoshop, select the transparency (or alpha) channel. You should see the active alpha channel represented by the moving line; you might have to inverse the selection. Next create a duplicate layer in the Layer Task Bar. The copy will appear as background. Then, from the Layer Menu, select Add Layer Mask and choose Reveal Selection.
An alpha mask now appears in the Layer Menu. You can drag and drop the rendering onto your base image in which all the composites will be combined. To place the new rendering precisely, drag the rendering close to where it should go and it will snap into position.
This procedure may not be obvious at first, but after a little practice you'll be able to get bring all the parts of the rendering into Photoshop in perfect correspondence.
Taken together, these techniques will give you greater artistic control over light, materials, and entourage.
Michael Sechman graduated from the School of Environmental Design at the University of California, Berkeley. His company, MSA, was first based on hand renderings. Their evolution into computer rendering began in 1986.