3D BASICS: Surfaces: A re-write

Surfaces

Very basically, there are three surface properties whose appearance is due to inter-reaction with light:
1.)Constant-Also called luminance, self-shading, ambient glow. This parameter controls how much of a base coat of illumination an object has. Most often, a slight amount of Constant shading goes into the overall appearnace of an object to make it look real (because except for my closet upstairs, nothing is totally without illumination in the universe.
2.)Diffuse-Also called roughness. This parameter controls how much illumination scatters when it hits the surface of an object. Indirect lighting in a house is the product of diffuse lighting, as is something such as a tennis ball.
3.)Specularity. Also called shineness, reflectivity. This parameter controls something called the mirror angle of light; how much light accumulates on a smooth surface into a highlight and reflects back to the camera (the viewer).By mixing amounts of diffuse and specular surface properties, you can create a plastic look (rough+specular), and metal (smooth+specular). As smoothness approaches 100%, the object takes on the color of the scene's illumination, and loses its native color, and generally will not take on shadows.

Transparency is a separate shading function which denotes the amount of light that passes through an object. Do not mistake Transparency for invisibility, however; a transparenct object can still show specular highlights.

Surfaces

Very basically, there are three surface properties whose appearance is due to inter-reaction with light:
1.)Constant-Also called luminance, self-shading, ambient glow. This parameter controls how much of a base coat of illumination an object has. Most often, a slight amount of Constant shading goes into the overall appearnace of an object to make it look real (because except for my closet upstairs, nothing is totally without illumination in the universe.
2.)Diffuse-Also called roughness. This parameter controls how much illumination scatters when it hits the surface of an object. Indirect lighting in a house is the product of diffuse lighting, as is something such as a tennis ball.
3.)Specularity. Also called shineness, reflectivity. This parameter controls something called the mirror angle of light; how much light accumulates on a smooth surface into a highlight and reflects back to the camera (the viewer).By mixing amounts of diffuse and specular surface properties, you can create a plastic look (rough+specular), and metal (smooth+specular). As smoothness approaches 100%, the object takes on the color of the scene's illumination, and loses its native color, and generally will not take on shadows.

Transparency is a separate shading function which denotes the amount of light that passes through an object. Do not mistake Transparency for invisibility, however; a transparenct object can still show specular highlights.

More surface properties:

Color) This properties depends upon the global lighting in the scene, so it isn't truly inherent to a surface. For example, if a blue ball is illuminated by a red light, it is black to the viewer.
Decal)Also called image mapping. It's the most very basic of surface mapping types.
Bump) James Blinn developed a very simple way to create the appearance of a surface being perturbed (bumped). Today, bump mapping, the use of an image to tell the rendering application what parts of a surface go up and go down, is still a very economical way to add reality and visual interest to an object. The drawback is that the silhouette, the outline of the object is not bumpy--it doesn't go out and in.

Displacement) A more sophisticated and processor-intensive way to add visual interest to an object, displacement actually changes the surface appearance of an object as it goes throuygh the rendering pipeline in image synthesis. IOW, the model you build does not take on displacement detail, but the rendering of the object does.

There are two kinds of displacements (as there are bumps) : An image controls the displacement or bump, or a procedure is called. A procedural texture is a "recipe" made out of math; you can see the results, but unlike an image file, a procedural texture can only be seen as the recipe is applied to a surface. Procedures can have fractals within them, and generally can be scaled smoothly, and are darned hard to write when you need something realistic. PIXAr totally subscribes to procedural shaders, and that's why PIXAR movies have this "less than real, dreamlike" quality. When you use an image to make displacements and bumps, the size of the bitmap is finite, and scaling the resulting surfaces sometimes yields artifacts (noise). But a surface that uses a bitmap for a decal, a reflection (sorry, reflection mapping can be done three ways: through a procedure, through ray tracing, or through an image)---is always as realistic looking as the image is.

Hi Gare... listen bud, how about doing a tut on mapping... like say turning a low poly slightly tweaked sphere into a head through your basic image mapping techniques... this is one area I really need to get a handle on and you seem to have this down real pat like... Please http://www.talkgraphics.com/images/smilies/biggrin.gif

Please do not minmg my artwork here; I'm trying to create examples, and not win any contests!

Okay, this is the way I'd approach creating a head with image mapping. Let me define image mapping first: image mapping in modeling is the use of a bitmap image as a resource for surface properties on a model. For example, if you create a human face as a painting, you can then map the painting onto a distorted sphere to create a facsimile of a human head. There also exists bump mapping (the use of a greyscale image to create the appearance of elevations and depressions along the face of an object ("appearance" is a key term here, 'cause bump mapping, although it can create spectral highlights and stuff, the object is still smooth), displacement mapping (where a greyscale bitmap tells the rendering program to actually create elevations and recesses along the face of an object), and refelction mapping, where an image is used as a reflection on a shiney object.

The rule is generally: the lighter the image area, the higher the corresponding elevation on the object to which the map is applied. Usually, there are different types of mappings for an object, just as we use different types of maps of the world to better see the location of countries. the basic mapping types are:

•Planar-the image map is simply pushed onto the object from an angle parallel to camera view. This type of mapping works great for flat objects, but can also be used for other objects as long as you do not rotate and animate the object. Planar is also a mapping type that is straightforward to create. In my example here, the bump and decal (color map, image map) map were created as planar maps and "pushed" onto the spheroid shape.

•Cylindrical-This mapping type is good for cylinders, vases, any type of surface that is revolved in its construction. It's also harder to paint one of these maps.

•Spherical--To map an image across the face of a sphere is wicked hard! There's a lot of distortion you have to take into account, but it's also the only type of map, for example, that is effectively (realistically) used on objects such as the pits on the face of an orange.

There are also mapping types that are unique to different applications. Shrinkwrap is a weird but useful mapping type that does what the name implies-the image is smooshed around the surface so the front face is a tight fit, and the other sides of the object stretch and pull on the image map. "Natural" is also an interesting, useful mapping type, where every face of the object starts a new image mapping. The disadvantage to this is that an object's faces are not always the same size, so you need to resize the number of repititions for each face of the object.

Here's a crude example of creating a face using image and bump mapping. Note the the nose appears to protrude because the corresponding area on the bump map is lighter than its surroundings.

Gary David Bouton
Gary@GaryDavidBouton.com
Free education! The Writings Web site
and the updated GaryWorld Gallery is pretty okay, too.

Creating a head with displacement mapping.

I claim no credit for this whatsoever: John Knoll, the creator of Photoshop, created a freeware plug-in called CyberMash that translates brightness in an image map to actual DXF elevations, similar to what you do with a displacement map (and a cylindrical object in this example).

See how the brightnesses in the head image translate to actual perturbations along the surface of the cylinder? This would be one way, probably the best way, to create a head using only image maps (then you apply a color map to the head object).

But why are we getting into creating heads when there's a program such as Poser?

Hi Gare http://www.talkgraphics.com/images/smilies/smile.gif

I am trying to wrap my feeble head aroung this mapping stuff so as to better understand just how far to take some aspects of modeling mainly re poly counts, and then where to introduce mapping thus having a lighter model for animation purposes...

For all intensive purposes the head example was but the first thing off the top of my head...hehe...

As for the Poser bit, well, I suppose that one could always go that route though in my own mind I really would rather learn from scratch just how to box model the human, and then apply my newly found knowledge with one of these fine mapping procedures you have nicely supplied here...


I really have to figure this stuff out, as I know it will very much assist through the whole process, as well I would imagine this kind of info will assist any and all who are looking to learn same...

Hey thanks for this info Gare, and should you have any more info to add or demonstrate through further examples, well, I am your humble student http://www.talkgraphics.com/images/smilies/biggrin.gif

Cheers

Gary, thank you for such informative/understandable information.

I have only dappled in 3D. This very subject matter has been very hard for me to get a grasp on because of my ignorance of the concepts (and terminology) to begin with. I believe your efforts here will make it easier to understand the less than ideal "help" files and manuals which the software manufactures provide.

Regards, John

Let me see if I can't come up with a good link to further info about image mapping.

Thanks for the votes of confidence on this thread. IMO, you're never too late to learn, and 3D is a wicked hard topic.

But the rewards are also obvious.

Here's an underdocumented example of how 2D paths become 3D surfaces, if anyone has questions. I'll write about extruding and lathing (rotating a path around an axis) real soon.


My Best,
Gare

Extruding and lathing are two of the most basic operations for making a 3D object out of a 2D path.

Extruding is the action of pushing a 2D path (sometimes called a profile) back in space. This depth direction creates the 3D dimension. If you goof with Adobe Dimensions, you'll see that you can also apply what is called a bevel edge to the corner where the depth dimension and the 2D path meet.

Lathing is like turning a piece of clay on a potter's wheel. Basically, if you take a path that looks, for example, like a vase, and divide it vertically through the center, you're left with a vertical edge that is straight. From there, you can command a 3D application to spin the path in space around the left edge. The result? a 3D vase.

Now, no one says that you have to use half a vase to make a lathe object. For example, you can spin a circle 360 degrees in space, with the axis of rotation not being on the circle. The result is a donut (technically, it's called a "torus").



My Best,
Gare