Applying an Image Using Planar Projection

We apply an image by selecting Planar from the Projection list in the Texture Editor, and then deciding which axis to project the image along by clicking on it in the Texture Axis option.

Figure 15-3

This is demonstrated in Figure 15-4. As you can see, we take an image and simply slap it straight onto the object along the same axis that the front of the object is facing.

Figure 15-4

So basically all you do is consider the direction in which the surface is oriented, and then apply the image along that axis. In this case, the surface of the object is facing sideways, so we basically apply the planar image along the x-axis.

Likewise, if the surface faces upward or downward, we would apply it along the y-axis, whereas if it were facing toward the front or back, we would project along the z-axis.

Planar Stretching/Dragging

Now if we look closely at the object with the texture applied, we notice a slight problem. Do you see how the texture stretches through the length of the object, leaving those rather unsightly lines? This is an unfortunate problem that we always face when using planar projections.

Figure 15-5

Think of it like this. Say we have a block, and we have a nice little piece of paper with a little design on it that is the size of one of the block’s sides. To apply that paper design onto the block, we would simply take the piece of paper and stick it onto one side of the block.

However, because the piece of paper is flat and only the size of one side, we cannot bend it around the edges of the block, so instead it just stays stuck on the side that we have glued it onto. Similarly, a planar projection cannot bend around edges; it simply stays stuck on the face that you have applied it to, along the parts of the face that are oriented in the same direction as the axis that you have projected it along. Planar mapping obviously works like a plane (hence the name), and as you probably know from modeling, you cannot bend planes, because in reality that is impossible.

Figure 15-6

So how does this explain the nasty stretching shown in Figure 15-7? Well, consider this. If the design on this piece of paper that you have stuck on your block has been painted onto the paper, and the paint is still wet, you could try to get the design that is painted onto it onto the other sides of the box by smearing the wet paint along the edges of the paper along the surrounding sides of the box. This would look pretty hideous.

Figure 15-7

Think of the pixels at the edge of an image being like the wet edges of that piece of paper. Now does that make sense? Obviously the best way to get the design onto each side of the box would be to stick a piece of paper with the design on it onto each side of the block, not by smearing the paint.

In LightWave, you could do this by using a combination of a couple of different planar projections along different axes, so that the image could be projected onto each side correctly. The easiest way to do this would be to assign different surfaces to the block according to their orientation and then simply project the images accordingly, as in Figure 15-8.

Figure 15-8

In this example, I have assigned three different surfaces to the box, one for each projection axis. So basically the top and bottom polygons have one surface assigned to them with the image projected along the y-axis, the side polygons have one surface applied to them with the image projected along the x-axis, and the front and back polygons share a surface with the image projected along the z-axis.

Figure 15-9

This, however, is not the only method. Instead of assigning separate surfaces, you could also apply a single surface to the entire box and simply set up three different layers within the Texture Editor, each with its appropriate planar projection. You could then use falloff (covered in Chapter 13) to ensure that none of that nasty planar stretching is visible.

So now we understand how planar mapping works on objects like blocks, but what about uneven surfaces? Well, you can use planar mapping on some irregular surfaces, but surfaces that are very uneven cause problems, because that horrible stretching that happened in the block example becomes visible. In Figure 15-10, I have projected the dirt texture along the y-axis, and as you can see, this works fine with the fairly flat surface, but when applied to the very irregular surface, we get stretching.

Figure 15-10

Although you could probably get away with this sort of thing in a distance shot, it will not suffice for objects that are close to the camera.

As you can see, the more the surface slopes away along the projection axis, the worse this problem becomes, as more stretching appears.

Figure 15-11

This makes planar mapping really mostly ideal for flattish objects such as walls, doors, floors, or any polygonal surface that is predominantly facing toward one axis and does not have too much visible depth where stretching would occur.

Blending with Falloff

The most logical way of blending planar projections would be by using falloff on each layer that needs to be blended, and then carefully positioning each image so that it all works nicely.

Using the same irregular ground object from the previous example, all I have to do is create two separate layers in the Texture Editor, each with the dirt image in it. One of the layers is projected along the y-axis, as before, while the other is projected along the z-axis. The image that is projected along the y-axis will not have to have any falloff settings applied to it, because in this instance, it will be on the bottom layer, while the other layer will blend with it from above.

But let’s do this step by step so that everything is very clear.

So, starting right at the beginning, I create one layer in the Texture Editor. Figure 15-12 shows the surface once again, with the image applied along the y-axis only.

Figure 15-12: Texture applied to y-axis

Take a look at where the stretching occurs, and remember that this stretching is happening along the surfaces that slope downward, along the same axis the image is projected along, which is the y-axis. In order to cover up these areas, we need to apply something along the z-axis so that these areas will no longer be visible.

Now let’s add a second layer to the Texture Editor above the bottom layer, this time with the same image projected along the z-axis. This step of the process looks like Figure 15-13.

Figure 15-13: Texture applied to z-axis

Yes, I know this looks wrong, but don’t worry! Once blended correctly, this image will cover up the stretching along the z-axis.

So how do we get these two layers to blend together correctly? Well, all I need to do is add some falloff to the top layer (the one that is projected along the z-axis) and reposition it slightly.

So what I do is set up some falloff to the layer as shown in Figure 15-14.

Figure 15-14

The percent of falloff refers to percent per meter of falloff, which means that 100% falloff creates a linear falloff to 0% at 1 meter.

The falloff on the y-axis prevents it from stretching along the bottom of the object, while the falloff on the x-axis prevents the stretching from appearing too much along the sides.

I also shift the actual position of the image slightly upward (to make absolutely sure that the image does not drag along the bottom of the object), and remove all tiling options so that the image does not repeat itself at all by selecting the Reset option in my tiling options (see Chapter 13 for more information on tiling images).

For the sake of this example, I have hidden the bottom layer so that you can see the way in which the falloff and repositioning of this image has faded the image out along the y-axis.

Figure 15-15

As you can see, we have a nice fading that will blend well with the bottom image. Notice that I did not fade it out completely along the side and bottom, as this would basically make the entire image disappear.

When I switch the bottom layer back on and render it, the stretching is covered up by the top layer.

Pretty nifty.

Figure 15-16

Blending with Gradients

The next blending method we can try out is using gradients in conjunction with weight maps. This method is not always as effective as the falloff method, depending on the model, but is nevertheless worth a mention. Although it requires a little more work, it is, in many ways, easier than the falloff method, which can sometimes be a little confusing.

For this method you simply create weight maps that will act as alphas to place the images where you want them to be, and then create gradient layers over those image layers in the Texture Editor, with the appropriate weight map set up as the alpha for that particular image. Let’s examine this process.

Figure 15-17 shows the object and the three images that we wish to apply to the object. For the sake of simplicity I am using a very simple object, a cube, but this principle could be applied to any type of shape. Image 1 will be projected along the x-axis, Image 2 will be projected along the z-axis, and Image 3 will be projected along the y-axis.

First, I need to create a weight map for the object to determine where the different images will be visible. Now remember, once we create the weight map and use it with the gradient, it will act as an alpha layer for the image that we wish to place onto the model. This means that we must create the weight map with a nice solid area of one particular value to determine the area where the image will be visible. So, I start off by creating a new weight map and calling it “texture image 1,” and selecting the areas where I want the image to be visible. I then assign a single value, 100%, to this selection.

Figure 15-17

This area, which has now been assigned a value of 100% in this weight map, can now be used as an alpha layer with Image 1, and the image will only show through this particular part of the weight map.

Figure 15-18

I now create a second weight map called “texture image 2” that will act as the alpha for the second image. I do exactly the same thing as for the first one, except that for Image 2 I set the weight map’s value to 100% on the areas that lie along the y-axis.

Figure 15-19

And once again, I create yet another weight map, “texture image 3,” that has the 100% area located on the areas of the object that lie along the z-axis.

Figure 15-20

I now have three weight maps, each of which will act as an alpha layer for its respective image when applied to the model. I now switch to Layout and begin to set up my surface.

I first apply Image 1 to the color channel of the cube. I set the projection to Planar along the x-axis.

As you can see, the projection is working as per normal planar manner. However, since we only want the image to show up on the parts of the cube that lie along the x-axis, I now create a gradient layer above the image to act as an alpha for it. I set the gradient’s Input Parameter to Weight Map, and select the “texture image 1” weight map. I then create a key in the gradient at 100%, and set the color of that key to white. I also create keys at 0% and 99%, and set the colors of those keys to black. I set the gradient’s Blending Mode to Alpha.

Figure 15-21

This will now allow the image to only show through the areas of the weight map that are 100%, as we can see when the cube is rendered again. See Figure 15-23.

Figure 15-22

I now repeat these steps for the other two images, creating for each of them a gradient that acts as an alpha layer with its corresponding weight map.

Once all of that is done, I have a cube with all three images placed on their correct sides, as shown in Figure 15-24.

Figure 15-23

Figure 15-24

So, as you can see, this is actually a very easy way of placing images and controlling them, even though it requires a bit of extra work.

Blending with Alpha Channels

While the weight map and gradient method provides a nice hands-on and visual solution to creating alpha channels within Modeler for your textures, you can, of course, also just create alpha channels within the images themselves. Certain file formats, such as TGA, can support 32-bit image depth that includes an alpha channel embedded in the image itself.

To do this, all you really need to do is to create a falloff along the edges of the image within the alpha channel, as shown in Figure 15-25.

As you can see, this alpha channel in the image would show only the areas that are within the white part of the alpha, so the edges of the image would be invisible when applied to the object.

Figure 15-25

If we were to take three images and create an alpha channel like this for each one, and could stack them on top of each other in the Texture Editor on the surface of the cube from the previous example, they would automatically blend together without a problem. Figure 15-26 shows three different images, each applied to a separate axis and blended by using their own alpha channels.

Figure 15-26

The only tricky thing about using this method on an object like this is that the images fade out along the edges of the cube, showing the surface’s color beneath it. So just be sure to make the surface’s color similar to the overall colors of the images.

Applying an Image with Cylindrical Projection

To use a cylindrical projection for a texture, select the Cylindrical option from the Projection list in the Texture Editor.

Figure 15-27

Once you have selected this type, you need to choose an appropriate projection axis from the Texture Axis options. Just like with the planar projections we looked at in the previous section, the axis that you choose for the texture to project along depends on the orientation of the object to which you are applying the texture. Upright cylinders would have the texture applied along the y-axis, whereas cylinders that are lying on their sides would have the texture projected along the x-axis or z-axis, depending on whether they are facing from side to side or front to back respectively.

Figure 15-28

Think of it as taking a piece of paper and wrapping it around a tube, because that is essentially what cylindrical mapping does, as demonstrated in Figure 15-29.

Figure 15-29

“Capping” Cylindrical Objects

As you see, the actual use of this projection type is extremely logical. The only problem that we really face with it, even when using it correctly, is the pinching that happens at either end of the cylinder, on the cylinder “caps.” This really cannot be avoided if the ends of the cylinder have the same surface applied.

Of course, we can use a variety of methods to cover it up, such as using falloff or any of the other methods discussed in the previous section, or we could simply assign a separate surface to the end faces.

Figure 15-30 shows a method using weight maps in combination with gradients to control the visibility of the image on the top. This is using a single surface with the gradient (together with the weight map) being used as an alpha layer on top of the image projected in a planar fashion along the y-axis onto the top part.

Figure 15-30

As you can see, it works quite nicely, and it was really easy to set up. All I did was create a new weight map on the cylinder that had an initial value of 0%, and then selected all the points along the top and bottom of the cylinder and set them to 100%, as shown in Figure 15-31.

Figure 15-31

I then use this weight map, together with a gradient, and set up the gradient so that the image projected onto this section will only show in the areas where the weight map has a value of 100% when the gradient layer is placed above the image layer. We obviously need to use this weight map to avoid the planar stretching that we looked at in the previous section, which would occur from the image on top being projected down the length of the cylinder.

Ensuring Seamless Mapping

It is important when working with cylindrical projections to bear in mind that somewhere along the object the ends of the texture are going to meet. Because of this, you might want to ensure that the image can tile correctly if you do not want any visible seam. Tiling images is placing repeating versions of an image on a surface. Ideally, in order to create a smooth look to the surface, we would not want to be able to see where each copy of the image ends within the pattern, because this would look rather odd. These visible edges are called seams and are generally to be avoided at all costs.

Obviously you are not always working with images that have to have all the seams hidden, as some surfaces, like a soda can, might actually have a visible seam on the label. However, when working with other types of surfaces, especially organic surfaces, you have to make sure that the images are going to meet without any noticeable seams.

Figure 15-32 demonstrates what happens when an image’s two sides do not match, resulting in a seam, and also shows a correctly made image that tiles the right way.

Figure 15-32

As you can see in the example on the left, the image used as a texture has sides that do not match up, resulting in an image that will not be seamless when applied to an object. Notice in the little block below the image itself that when the image is placed alongside itself, you can see the separations at the edge.

In the example on the right, no seams are visible when using this image, as this particular texture is tileable because the edges are matched.

Using Adobe Photoshop’s Offset Filter to Hide Seams

The easiest way to make sure that an image can be tiled is by using the Off-set filter in Adobe Photoshop. If you create an image that you wish to tile on a surface, all you have to do is offset it along the top and one of the sides, and then simply cover up the visible overlapping areas using airbrushing or the cloning tool. Let’s take a look at Figure 15-33, which shows the image used in the previous example that had unmatched sides.

Now, we go to the Offset filter in Photoshop (Filter>Other>Offset), and offset it from the top and the right, both by 100 pixels. We get the result shown in Figure 15-34.

Figure 15-33

Figure 15-34

As you can see, we now have a visible seam within the image itself. All we have to do now is hide it. In this example, I just use the Rubber Stamp (clone) tool. See Figure 15-35.

If we try tiling this image now, we will have no visible seams, as shown in Figure 15-36.

Figure 15-35

Figure 15-36

And if we were to wrap this around a cylinder, we would have no visible seam either. So that is a nice quick method of ensuring seamless textures!

Using the Width Wrap Amount Option

The most astute of you may have noticed that there is an option when using cylindrical projections in the Texture Editor called Width Wrap Amount.

Figure 15-37

The value you enter into this field determines the number of times the texture is wrapped around the object to which it is applied. So if, for example, you wanted to wrap an image around a cylinder five times, you would enter a value of 5 in this field.

Figure 15-38 shows the result of wrapping a texture a total of eight times around the cylinder.

Figure 15-38

Obviously, this is great for adding any repeating details without having to actually make one big image that includes the repeating, especially since smaller images use less memory. In a case like this, it is beneficial to use the simplest, smallest image that you can get away with. Of course this is also a quicker method than creating a long image with the stripes repeated eight times.

Setting your width wrap correctly is also important for preventing an ill-sized image from becoming hideously stretched if it wrapped 360° around an object that the image is technically too small for. Increasing this amount would therefore reduce this stretching by repeating the image more times along those 360°, thereby making each repeat of the image cover less space, and therefore distort less.

Applying an Image with Spherical Projection

To wrap an image spherically around an object, simply select the Spherical option from the Projection list in the Texture Editor.

You then select an appropriate axis along which you wish to project the image. Most spherical mapping situations use the y-axis.

Figure 15-48

Figure 15-49

You’ll notice that we also have options called Width Wrap Amount and Height Wrap Amount when using spherical projections. Like the Width Wrap Amount for cylindrical projections, here this value simply determines the number of times that the image is repeated as it is wrapped around the width of the model. The Height Wrap Amount value determines the number of times that the image is repeated along the length of the object (as opposed to its breadth). These values basically work as tiling options for this type of projection.

If we change each of these values to 4.0 using our world image, the way in which these values work is clearly illustrated. In Figure 15-50 you can see that the image is repeated four times around the width and four times along the height of the sphere. For most situations, you’ll find that the value of 1.0 is most appropriate.

Figure 15-50

Ensuring Seamless Mapping

As with the cylindrical projections that we looked at in the previous section, spherical projections also have the risk of visible seams on the axis along which we have projected the image.

This means that generally we should always try to ensure that the edges of the image will meet seamlessly.

Figure 15-51

Refer back to the previous section on cylindrical mapping and review the technique of using Adobe Photoshop’s Offset filter for correcting this.

However, when dealing with spherical mapping, not only do we have to watch for seams along one side of the object, we also have to be aware of how the image looks when it meets up at the poles.

Using Adobe Photoshop’s Polar Coordinates Filter to Check Seams

A quick and easy way of checking to see how the image will look when applied spherically is to use Photoshop’s Polar Coordinates filter.

This filter is found under the Distort filters in Photoshop. Use it only to check your image, as leaving the filter applied to the image will mess up the image, making it no longer useful for our purposes.

Basically what you do is open the filter’s panel and ensure that the Rectangular to Polar option is selected. The little preview window will show you a decent representation of how the image will more or less behave when wrapped spherically around your model in LightWave.

Figure 15-52

You can use this to ensure that the image is seamless at the poles. In Figure 15-52, we can see that the Earth image has no problems as the two poles (the two polar regions) meet up with no visible seams.

On the other hand, if we take a look at an image that has visible seams in it, these seams will show up when looking at the preview pane in the filter’s panel.

Figure 15-53

As you can see, the seam is clearly visible. Using this as a guide to check our progress, we can then use the Offset filter technique and lots of Clone Stamping and Healing Brush work to eliminate the seam.