Computer Art Methods and Terminology, Demystified

This has been bugging me for a while. Back in June, I made a post about an art piece that was my longest render time ever. I've also mentioned the term "render time" several times since -- but I realized that I never defined this term, and that to people who aren't 3D artists or CAD designers, it's probably pretty unclear. For that matter, probably a lot of the terms I use in conjunction with my art are unclear. So here's a primer:

"Render time" doesn't have anything to do with how long it takes me to create an image. That's not something I really keep track of. Instead, the render time is how long it takes for my computer to generate the final, full-effects "snapshot" of the scene I've created. In other words, while the computer is rendering my image, I'm off doing something else. So a 35 hour render time is 35 straight hours where my computer is unusable, or at least very slow, because it's busy cranking out the pixels. If I had a faster computer, or used multiple computers to do the rendering together, it would take less time.

My part of the work all happens before that, and I work off of smaller, lower-quality preview images and partial-renderings. Essentially, my job is to size, position, and texture the terrains and objects, set up lighting and atmospherics, and position the virtual camera. Once I'm reasonably happy with my results, given the little previews, I set the program to "render to disk," and walk away.

If I'm doing actual object-modeling work, as well, that generally occurs in a program like Sketchup or Hexagon or Poser. When I have my model done, I then import it (in grouped pieces) into Bryce, and I texture it and render it there. Object-modeling work is tedious and not something I'm especially good at, however, so often I just use pre-fab components. Even when I use models that I didn't originally create, however, I put a lot of work into tweaking parts of them, and I usually have to break the model into individual slices that I can separately texture.

"Texturing" refers to applying some sort of material to objects. Basic objects are just rendered as flat gray -- it's the process of texturing that makes things look like sand, water, glass, leaves, snow, weeds, etc. I don't do any of my own texture-creation -- most 3D modelers don't, actually, since that's a completely different skill set (it involves photographing real materials and then carefully cropping them to a repeatable pattern). I use base textures from Bryce, texture sets that I've purchased from third-party texture authors, and textures that are free on the Internet (Mayang's Free Texture Library is a great resource).

But texturing isn't as simple as choosing from a palette and then walking away. Often you need to blend multiple textures together, or resize them for the proper scale/orientation of your model (this is difficult to explain, but it can be tricky), and also certain things like reflectivity, transparency, diffusion, ambiance, etc, need to be adjusted (or changed all together). Plus, the process of just selecting the right textures that will complement each other, and/or blend well is difficult in itself. Scenes that have a lot of organic textures (like landscapes) can take a fair amount of work to make look realistic (or properly surreal, if that's your goal).

You might be surprised by how different the same scene can look with different textures applied -- the same is true when it comes to lighting and atmospherics, actually. The right levels/colors/distributions/positioning of fog, haze, clouds (cumulus and stratus), stars, sun/moon, etc, can make a huge difference. As can such effects like volumetric atmosphere (which simulates particles all throughout the air at the density and reflectivity of my choosing), and a variety of other atmospheric tools.

One more clarification about object-modeling: that doesn't refer to trees or terrains in my case (most of the time). Some of my trees are pre-fab, but most are generated using algorithms built into Bryce. I set a variety of parameters (trunk size/angle/style, branch density/angle, leaf type/texture/color/distribution, etc), and that's that. All of my terrains start out by being generated via fractal algorithms (in either Terragen or Bryce, depending on which I'm using) because fractals are great for realistic terrains. Once I have the general shape of the terrain in place, I use a height-mapping tool to "paint" and thus raise and lower the terrain, add cracks and rivults, that sort of thing. In Bryce, I can also apply terrain-wide effects such as smoothing, erosion, spikes, etc. When I'm done with all that, the terrain is ready to be sized, positioned, and textured.

Hopefully this provides some insight into the process via which I produce most of my art (my abstract works are a completely different process and skill set, mostly just using Photoshop by itself, or in conjunction with programs like Apophysis). Working in a wide array of programs helps keep it interesting, though -- and when I'm a bit stagnant in one program, I can always switch tack and do something completely different in another program.