The Description of Texture Mapping

Texture mapping is available in L2 system. This function has powerful expression to render a scene by your appropriate loying. You should free from the natural meaning of a word `texture', then please try to map many kind of textures on your objects. You can create many kind of objects with using `wrinkle', `stonewall' and other textures according to mapping place, even if you have only a few objects.

We implemented, as the CG area people called, `Texture Mapping technique.' (We don't use bump mapping technique.) Although, we can see `concave' and `convex' on the `adobe' texture, it is only a picture of adobe. So, shadow on the adobe is not changed even if light source position is changed. And we can see some king of an artifact on the adobe.

Sometimes you may meet texture distortion phenomena. For example, a part of texture image on an object is stretched. When this distortion is appeared on an non-animated image, you can say that `this is the image what I want'. However in the case of animated image, a distortion on objects may shrinking and stretching on objects, and you can feel strangeness easily.

This phenomena is caused by unavoidable rounding error of perspective transformation software, not a bug. Especially, in the case of following three condition exist simultaneously, this effect is expanded more.

• A polygon size is too large.
• There is very large difference of distances between camera and each vertex of a polygon.
• Some vertices of a polygon are out of view.

After all, these cases may occur when the camera looks around the side of a huge battleship or a huge base, or you use one large polygon as ground.

In many cases, creating ground will be a big problem. Like the sample animation of L2, when you want to make the movie that scenary is on the ground, the ground may be winding.

To avoid this problem, you can divide a large polygon to many sub-polygons. Thus, you should make the ground by many squares like a checker board instead of one large polygon. If a texture image has continuous boundary at upper (right) and lower (left) edges, we can hardly recognize the boundaries on the scene.

You will use this method frequently. For your convenient, it is recommended to use an object constructed by many square parts created by the parts assembler, in place of you use the motion editor to locate number of square parts cut by cut.

If you are a computer graphics expert, you can easily expect that there is inconsistency between object color and texture color.

Because the texture image like `Camouflage1' has already color independence of an object color. When object color is purple and texture image is `Camouflage1', what will be happened? The answer is that the result color is created by multiplying two colors. Therefore, you want to use the color of a texture image as an object color, you must make the object color white.

More precisely, if you set an object color purple, G element of RGB value will be almost disappear, and if you want to map green texture image which has no R and B value on the purple object. Under this condition, all RGB value will be near to zero and result color may be almost black. In the same fashion, when purple object with red texture makes red, and purple object with blue texture will be blue. Please think about these effects with separating color as RGB value.

In the case of monochromatic texture image like `wrinkle' and `rugged', white area of the texture image has object color only. However black area of that will be black in spite of any object color. Therefore, mapping the texture image will make the object color darker only.