08/15/2008    |    Devlog    |    BSharp    |    Discuss

Texture Mapping

I’ve always thought of “Texturing Mapping” as a task with the worst name in computer art production history. The title itself makes the act of painting a model sound about as romantic and creative as throwing a bedspread over a mattress. Clearly, the task was named by an engineer whose knowledge of art fell short of understanding that texturing is, in fact, the closest that 3D production comes to the highly skillful, passionate, and personal act of painting. Suppose the process of composing music on the computer was called “note diagramming”? Do you see what I mean? It completely misrepresents the art form – not to mention that it sounds unimpressive and uninspiring. Further, it somewhat implies that you don’t need to have musical talent to write music, but just to learn a process – the process of diagramming notes. Think how different the job would sound if that engineer had dubbed it “3D Painting” instead of “Texturing Mapping.” IMO, the title contributes to a universal misunderstanding of the important role that painting plays in 3D production and the challenging art form that it is – which is tragic really. Texturing, we shall see, has little to do with the dry exercise that its title implies and is (along with lighting) the most impactful element of 3D image creation.

Textures

The wiki gives us the following definition: “A texture map is applied (mapped) to the surface of a shape, or polygon. This process is akin to applying patterned paper to a plain white box.”

“Patterned paper to a plain white box”...wow, that’s accurate but it makes me sad all the same. Their definition is like most definitions, a dry description of a process. So, what is it about texturing that brings more value than implied here, exactly? I mean, besides the obvious fact that without them wood looks like a gray block, what qualifies them to be more consequential than the other elements of 3D imagery? To answer this, I’d like to share whatever insights I have into image making by way of classical painting.

Painting

When people look at a painting by someone like Rembrandt, is it the line drawing beneath all the paint that they’re interested in? Or is it the painting? I’m no genius, but I’m going to go with “the painting”. Rembrandt’s work is about his unique ability to make lighting transcend the medium of oil paint. His paintings glow with a luminance so powerful that it reaches out from over the centuries and shapes one’s perception of the physical world around them – at least, that’s what it does for me. When I first attended a showing of Rembrandt’s paintings, I tuned into qualities of lighting that I hadn’t previously noticed; it was quite inspiring. He was clearly a master image maker.

As for the drawings at the base of his work…well, they should be accurate, of course. I mean, the subject shouldn’t have a giant left eye or whatever, but the sketch underneath all that paint is not nearly as important as what he could do with the layers of paint on top and isn’t what draws people to his work. A sketch and a painting are worlds apart and if you’ve ever been privy to seeing the pre-production sketches done by a master in preparation for his masterpiece, you understand how different they look and how much more powerful one is than the other. So, I’ll say it again: when looking at a Rembrandt, what were appreciating is the painting. It’s important to keep this, perhaps obvious, idea in mind because people seem confused about this with 3D when, in fact, it’s no different.

3D vs. Painting

If you think about it, a 3D image is very much like a painting: the wireframe of a 3D model is just like the line drawing on a canvas, awaiting paint to complete it (can you see where I’m going with this?); and just like the paint on a canvas, the textures are going to be what people see and what they appreciate. So in the end it is not the model (or drawing) that is going to make an image compelling, but the textures (or paint) layered over it, just as it is with a Rembrandt. This is an extremely important fact to have straight because most people have it exactly backwards and think that models are the s—- and textures are just something that prevents them from remaining gray. In the age of 3D technology, modeling is the star and texturing is just something you do to the model...and so the Leonardos and Rembrandts of the 3D era’s praises go unsung.

Why is Texturing Marginalized?

There are multiple reasons why texturing is pushed aside. First off, it’s hard and therefore people don’t want it to be more important than other, easier, tasks. Becoming a good painter (which is basically what a texture artist is) takes an enormous amount of time and dedication. One has to develop one’s understanding of color, saturation, contrast, texture, edge control, surface quality, lighting, and artistic “voice”, etc, to be good at it (understanding color alone is one of the most difficult challenges in art). Painting is a lot more complex than drawing (drawing, in fact, is just another thing on the list of things that you need to understand before you can be a good painter) and riskier, because after all the time, dedication, and effort it takes to learn to paint, there’s no guarantee that you won’t merely be mediocre (after all, not every artist from history is as brilliant as Rembrandt; save for a handful, most are mediocre).

Second, the amount of time it would take for someone to learn modeling is far shorter than the time it would take to learn to paint (and with less risk of failure in the end). Modeling is easier to learn and succeed at, which means a larger number of people can do it, which means a larger number of people want it to be more important than texturing. So, there are political reasons why people would want to hype up its worth. If enough people tout modeling and marginalize texturing, others are going to adapt this bias as gospel (for the most part, art is a mystery to people on the outside of it, which makes it easy to manipulate perception). Unlike painting, 3D modeling is, in fact, a process that can be learned by pretty much any random pedestrian off the street. Now, some will be quicker at picking it up, some more economic with their poly counts, some more clever with their vertex arrangement, some will be faster and cleaner at production, etc, but anyone can learn to do it.

BTW, when I write about modeling, I’m not talking about design. Design and modeling are different. Design is a creative expression and modeling is a process. For instance, making a model of the Chrysler Building is a learnable task (that’s modeling), but if the Chrysler Building never existed, creating something like it for the first time is something very few people can do (that’s design).

Third, the name (see first paragraph). While it was probably not intentionally misleadingly named, I’m certain that the title is part of the problem. I seriously doubt any programmers have evil in their hearts when they inadvertently diminish the importance of a feature by naming it inappropriately; it’s easy for a non-artist (or, for that matter, a bad artist) to overlook the importance of an artistic feature. Still, the teacher in me looks forward to a revolution that forever banishes all the programmer labels from art production and replaces them with more artistically appropriate and intuitive titles.

The Impact of Textures

If we took the same line drawing and handed it to both Rembrandt and Norman Rockwell, we would get a pair of fantastic but very different ensuing paintings. For instance, Rembrandt’s would be all about strong dramatic lighting while Rockwell’s would deal with light in a very different way – but I’m confident that both would be great. Now, if we handed that same drawing over to a mediocre painter, we would get mediocre results. Two of our paintings would fall in the “masterwork” category and the third would fall in the “valueless” category, even with the same line drawing at the base of them all.

In 3D it is no different, which is why I state at the top of this devlog that textures are the most impactful element of 3D imagery. No matter how nicely a character or scene is modeled, a mediocre texture is going to transform it into a mediocre result. So then, texturing is no more something that you merely do to a model than is painting merely something you do to a drawing.

Technology

Of course, in the world of 3D today we are not limited to just “paint” for the creation of surfaces. We have all kinds of techy sounding shaders that create effects beyond the constraints of bitmaps alone (bitmaps are simple, straight-forward textures with nothing fancy about them). Things like normal maps, specular maps, parallax maps, etc, have been invented by smart devs to create sophisticated tricks that can be layered on top of textures to make surfaces appear more complex, detailed, or light-reactive. These features, however, are not a quick recipe for success in textures (as many believe); artists cannot ignore the development of their skill as painters because success (even with these high-tech sounding features) still requires the same insight to image making. Technology does not think for artists, artists still have to think for themselves.

FLS

At Flying Lab we care very much about textures. Our team endeavors to improve on their ability with each new task. Many of us study painting in our spare time as a testament to the sincerity of our dedication to growth in this area. Our environment artists have developed a process in which they create textures first, then model to those textures (you could say that at FLS modeling is merely something we do to a texture). We all strive not just to be better technicians (modelers) but to be better image makers because we respect the power of good imagery.

Anyway, I just picked up a new music CD called Video Games Live that includes one track by my good friend Jason Hayes. Jason wrote the “note diagrams” for World of Warcraft. I think I’ll give it a listen.

08/15/2008    |    Devlog    |    BSharp    |    Discuss