The leading 2D texture-painting tool is Adobe Photoshop because of its wide range of paintbrushes, layering and masking features, and comprehensive set of image-processing capabilities.

In the not-too-distant past, 3D models were exclusively painted in Photoshop or other 2D image-processing and painting tools; texture artists had to paint all the textures in 2D on UV maps. One of the challenges with this process is interpreting how a 2D image will wrap around a 3D object without stretching or compressing textures. Another challenge is painting uniform patterns across seams that are cut when creating UV shells, because these UV shells are not connected on the flattened 2D UV layout. A third challenge is that, when you paint in 2D, you constantly have to go back and forth between Photoshop and your 3D application to see whether the results are acceptable.

With the introduction of 3D painting tools, this process has become easier; users can paint directly onto the 3D models in 3D. Mudbox, in addition to being a powerful and versatile digital sculpting application, provides some unique and extremely useful digital painting and texturing capabilities.

These 3D painting capabilities remove most of the barriers of 2D painting applications because you can paint directly on your model in 3D, across UV seams, and see the results of your brush strokes in real time. This does not remove the need for a 2D compositing or image-processing application such as Photoshop, but it is a huge time-saver and gives you greater creative control over the texture-painting process. The introduction of versions 2010 and 2011 now further integrates Mudbox with Photoshop by enabling you to save to Photoshop's native .psd file format, preserving the layers that can be brought directly into Photoshop.

Another big advantage of Mudbox is that you can paint on your 3D model in layers. This is useful because you can have layers that represent different details on the surface of the model. One layer could be a base material of the model; the second could be a paint layer; the third could be weathering such as dust or dirt or rust; the fourth could be decals, emblems, or posters; the fifth could be glossiness or shine; and the sixth could be bumpiness of the surface. Just as you can in sculpting layers, you can adjust the opacity and blend modes of these paint layers to show as much or as little of each layer to get the desired look.

The new Flatten to UV Space feature introduced in Mudbox 2011, which is accessible through Mesh → Flatten to UV Space, enables you to paint or edit textures on a flattened representation of the original. You can also use the Pinch or Grab tools to perform basic UV position adjustments on the model in this mode. This is extremely useful to get an idea of how your textures are distributed on the UV map and make adjustments to the textures and the UVs themselves to accommodate the results you want.

I mentioned UVs in the preceding chapter and indicated that they represent coordinates on the surface of the model. You will revisit them in this chapter as they pertain to texture painting. All UV editing on a model needs to be finalized before painting and texturing your 3D model in Mudbox to avoid misaligning UVs and your painted areas. (Chapter 6, "Digital Sculpting Part II," covers how to lay out UVs for a sculpture by using Headus UVLayout.)

For the most part, UV maps for a model take up one square UV tile, or quadrant. Some applications, such as Mudbox and Maya, support UV shells distributed to multiple-square UV tiles. The benefit of distributing the UV shells across multiple UV tiles is that you get more room per shell to add detail to your texture. However, for the most part, you will do all your UV work in the first quadrant, or 0 to 1 tile.

Whether UV shells are laid out in one quadrant or many quadrants, the UV shells should not overlap. Overlapping UV shells will cause unpredictable results in 3D applications and in your final result. The only time UV shells can overlap is when you intend to take advantage of symmetry in your model and save texture space by overlapping symmetrically identical UV shells precisely on top of each other to reuse a part of the image that is symmetrical or repeated on the model.

Make sure to leave 4 to 6 pixels of space between UV shells, and between UV shells and the four borders of the UV tile. If you fail to do so, the textures will bleed from one shell to the other, producing streaking lines on the paint you apply to your model.

The shells in your UV map can be off scale. For example, if you are texture-painting a human being and you need more space to paint the details of a face, you would want to scale up the UV shell that corresponds to the head. However, if you do that, you will have broken the proportionality of the shells. For example, if you have all the shells in the same proportion and you apply a checkered pattern to a human model, the checkered pattern will be uniform throughout the body (see Figure 4.1).

However, if you apply the same checkered pattern to the model on which you scaled up the head, you will notice smaller checkered squares on the head than on the body (Figure 4.2), because you have more UV space allocated to the head.

If you need more UV space for the entire body because, for example, you need to paint intricate tattoos on the body in addition to details on the face, you can scale up all the shells and distribute them across multiple UV tiles. Note that even though you can distribute UV shells across multiple UV tiles, they should be laid out so they are contained within the square UV tile and do not cross any UV tile borders.

Figure 4.1. A uniform checkered pattern on proportional UVs

Figure 4.2. The checkered pattern appears smaller on the head because more UV space is allocated to it.

The way to determine the distribution of shells to UV tiles depends on the following factors:

Variation of detail on the model If you have a section with minimal variation in color and texture detail, and the UV shells corresponding to the section are receiving a uniform color, then you can give its UV shells less UV space. On the flip side, if you have sections that need more detail, such as a head with facial details or a mechanical part with intricate surface paneling and text, you want to give the shells corresponding to that section more UV space.

Game requirements For games, you will want to constrain all of your UV shells into one UV tile because games need to load textures really fast and game machines need to manipulate models in real time. Games might even cram UVs for multiple objects or characters on one UV tile.

Texture types UV shells might also be distributed among different tiles by type of texture—for example, all metal UVs on a metal texture UV tile, all leather UVs on a leather texture UV tile, skin textures on a skin tile, and so on.

UV space usage Rotate and move the UV shells to take up minimal room in the UV tile and maximize UV space usage. Try to have the least number of shells possible because this maximizes the amount of utilized UV space. This must be, of course, balanced by making sure you have the minimal stretching or compression of UVs, which is usually done by cutting more seams or dividing UVs into more shells.

Texture maps are 2D images that you generate either in 2D image-processing applications such as Photoshop, or in a 3D painting application such as Mudbox. These 2D images are then wrapped around your model much like wrapping paper around a present.

After you have laid out the UVs, you can output the UV map as a flat bitmap image. Most UV layout tools, such as Headus UVLayout or UV Texture Editor in Maya, allow you to output the UVs in any desired map size. Usually these are 1K (1024 × 1024), 2K (2048 × 2048), or 4K (4096 × 4096) bitmap images. This UV image will serve as a good base layer to use if you are going to do any 2D painting on your model or as a point of reference to see where the various paint regions align with the model.

All the texture painting you will be doing will be on the tiles on which you laid out your UVs. Texture maps are generally RGB or RGBA 2D images in one of the nonlossy, popular image formats such as TIF, TGA, or OpenEXR. You can create one or multiple texture maps to apply to your UV map. Multiple images are subsequently composited as layers and flattened to form the final texture map that will be mapped on your model for its final presentation.

We can generally tell what an object is made of by the way it looks. This is because our eyes and minds are trained to look for certain characteristics that distinguish one object from another. These characteristics help us discern glass from metal, wood from plastic, and rubber from rock. We are so good at this that we don't even think about it; we just know from years of experience.

When we are creating computer-generated imagery, we need to convince these well-trained eyes and minds, whose suspension of disbelief can be stretched only so far, that what they are seeing in the image looks like the object in the real world. To do this, we have to reverse-engineer what our eyes see to create believable or even stylistic objects that convey characteristics of the world around us. Even though software has the capability to generate these characteristics, we still need to know how to manipulate the features to get the desired appearance.

To do this, the first step is to start training our eyes to look at characteristics that differentiate the appearance of the objects around us. What is it about the characteristic appearance of an object that makes it metal, plastic, marble, or glass? How does it reflect, refract, or absorb light? What surface properties define rubber, velvet, brushed metal, rust, and water? Why can we still tell what something is made of even under varying lighting conditions such as candlelight, neon, or sunlight?

Two characteristics to take note of when looking at objects are their material and texture. People new to the field mistake one for the other or start out thinking they are one and the same, but material and texture are, in fact, two distinct characteristics of the way an object looks.

In 3D applications, both materials and textures can be represented with shaders, which are sets of instructions that tell the rendering software how to render a material with a set of properties that can be adjusted by the user to represent real-world objects.

Just as the grand masters put down layers of oil paint on canvas to paint realistic images, you need to layer textures on your image to achieve a believable look to your model.

These layers can be complementary to a specific pattern on the surface. In Chapter 1, "Getting Your Feet in the Mud: The Basics of the Mudbox Production Pipeline," for the egg we used a base coat layer that we painted on, added a projection-painted layer of the image of an egg on top of that layer, and then adjusted the opacity of the two to get a more realistic look. Layers can also represent different surface patterns such as drips, rust, patina, and decals.

Paint layers are different from sculpt layers because with sculpt layers, each object can have its own subdivision levels and one or multiple layers independent from the other objects in the scene. However, with paint layers, objects that are in the UV map together are all on the same paint layer. As you are painting on different parts of your model, you are changing only one paint layer.

Paint layers cannot be subdivided; to add finer detail to a paint layer, give the texture map associated with the layer higher pixel resolution. Even on a model with a few subdivisions, you can paint a very detailed and nonpixelated texture map if you chose a large texture map size such as 4K or 8K.

We control the results of layering texture maps in Mudbox by the order and opacity of the layers. The ordering is done top to bottom, with the topmost layer in front (as the outermost layer on the model), and the subsequent layers behind it in order (inside the model, like layers of an onion). The opacity is controlled by a slider that ranges from 0 to 100, where 100 is fully visible and 0 turns off the layer completely.

In 2D or compositing applications such as Adobe Photoshop or Corel Painter, you have long been able to add layering effects including multiplication, addition, hue, soft light, and so forth. Mudbox 2011 introduces the ability to apply some of those compositing effects to your Mudbox paint layers as layer blend modes. You are now able to use Multiply, Add, Screen, and Overlay paint layers to produce various blending or compositing functions.

Blending layers is extremely useful because certain effects need these blending layers to work effectively. For example, if we extract an ambient occlusion map, or generate one in another 3D application, the way to apply it to the model would be to Multiply the layer it is on to the overall paint layer stack. This will ensure that the occluded darker areas show up in the composite while the other areas do not. We will go into blend paint layers in depth, and see their effects in Chapter 8, "3D Painting."

It is important to know that the Mudbox Material shader has various channels that can have layers of their own. These channels are Diffuse, Specular, Gloss, Incandescence, Bump Map, Normal Map, and Reflection Mask for the Mudbox Material in Mudbox 2011, and Diffuse, Specular and Specular2, Gloss and Gloss2, Bump Value, and Reflection Mask for the 2009 and 2010 versions. The Mudbox Material used in versions 2009 and 2010 can still be used with later versions to maintain compatibility. As mentioned, all these channels can have multiple layers or maps of their own. The Diffuse channel contains the color maps. The Specular and Specular2 channels contain the highlight and fringe highlight maps. The Gloss and Gloss2 contain the size and fringe size maps for the specular reflection. The Bump Value channel contains the bump maps, and the Reflection Mask channel contains reflection maps that mask and restrict reflections.

One last important aspect to know about layers is transparency. For example, if you have a layer that has decals on it, you need the decals to show up on the surface of the model, but not the areas on the layer surrounding the decal.

The transparency is achieved with the image's alpha channel, which stores the transparency information for each pixel. Just as the RGB (red, green, and blue) channels store information about the amount of red, green, or blue each pixel has, the alpha channel stores information about how transparent each pixel is. If the alpha value is white, the pixel is opaque; if it is black, then it's fully transparent. The different levels of transparency are represented by 256 values of white to black.

By default, all the file formats for paint layers in Mudbox are created with an alpha channel. However, if you are importing images as texture maps into Mudbox, and you want details from inner paint layers to show through parts of your imported image, you need to make sure the image you are importing includes an alpha channel, that is, it is an RGBA image. Look at the file formats in Chapter 2, "The Mudbox User Interface," to determine which ones have an alpha channel.

Like texture maps, specular, gloss, and bump maps are 2D images that denote where the surface of the model is shiny, glossy, or bumpy. This is usually done with value, meaning that areas on the specular or gloss map that are lighter or white are the shiniest or glossiest, while the darker colors are duller and less glossy.

The surfaces of objects do not reflect light evenly. For example, if you look at someone's face, even though it has the same skin throughout, there are spots such as the nose or the forehead that might be shinier than other areas because of some surface oil, so they have a different specularity than the rest of the face. A car's surface might have a different shine or specular value in areas that are more prone to friction than those that are not. Even a shiny surface such as the cover of this book might have different specular values from fingerprints or scratches.

The different values of gray on a bump map denote relief on the surface. Darker areas with less value are indented, whereas lighter values protrude from the surface of your model.

In general, these maps are grayscale, but they could be in color. Of course, if they are in color, only the value of the image will be used.

Whether we are creating something that does or does not exist in our world, we still need to base it on something that has visual familiarity. The best way to do this is to find objects that represent that familiarity either in the form of images or in the real world.

With the advent of digital photography, Internet image searches, and photo-sharing sites such as Flickr, Picasa, and Photobucket, you can find thousands upon thousands of images of what you are looking for. It is a good idea to devote a sufficient amount of time to reference gathering before you sculpt or paint your model. Digital images are plentiful and extremely useful, but it's also helpful to go out and physically look at your subject matter. Images can never substitute for walking around your subject, touching it, smelling it, squinting your eyes at it, or viewing it in different lighting conditions.

All sorts of reference images need to be compiled. As you progress in your career, you will start putting together a reference library that you tap into again and again for your sculpting and texture-painting needs. For bigger projects with big pipelines, reference images will be given to you by an art director or the art department, but it's always helpful to augment that with reference material of your own if possible.

Reference images can have many aspects, and you should gather your information to cater to most if not all of them. The types of reference materials include the following:

Environmental setting Gather reference images of location and setting indicating where your model will be in your final output, the environment around it, the weather conditions, the age of the environment, similar objects in that setting, and so forth.

Color Color is a powerful communication vehicle that sets the mood of a scene and needs to be balanced delicately to achieve the results you need. Gather references that show the values and saturation of the colors you want to convey. The color of an object can diametrically vary based on its environment and lighting; it's indispensible to have images of the colors you want to convey in their natural setting so you can simulate these aspects in the environment to achieve the look you are going for in your reference.

Composition Another critical aspect of your final output is the juxtaposition of your model with respect to other models in its environment and surroundings. This adds visual interest to your image and guides the viewers' eyes when they view the scene. It's best to have some images that give you visual cues of how to balance the composition of your scene to achieve the impact you want to achieve with your final output.

Lighting Just like color, lighting sets the mood of your output. You need to have reference images of the lighting settings you are looking to achieve. When you look at your lighting images, you need to ascertain where the different lights are emanating from that are showing on the surface of the subject. What is the light source? Is it the sun, the moon, a lightbulb, or a candle? What angle is the light coming from? How many light sources are there? Are they direct light sources, or are they bouncing or reflecting off surfaces around it? All these questions will need to be answered, and they will be easier to answer if you have reference images of the lighting you want to achieve.

Pose References of figures in the pose of your model will also be helpful, especially if the images are from multiple angles, so you can achieve the same pose on the model.

Materials and textures and stencils Gather or take as many images as you can of materials that portray the same characteristics as what you want your model to look like. Get overall pictures as well as zoomed-in pictures that will give you more detail as to how the texture is formed. Not only are these references good to look at while you are working, but you can actually use them to either paint the texture on your model in a 2D image-processing application such as Photoshop or to projection-paint them as a stencil in Mudbox. You can also use desaturated and manipulated versions of these texture and material images as bump, specular, or gloss maps.

Stamps and Photoshop brushes Reference images that represent various textures can also be used as tools to paint and sculpt with. When you see a pattern that you like, you can desaturate it and use the 256 levels of gray in it to sculpt relief on the surface of your model, or you can use it as a color image to add texture to your model.

As mentioned before, Bertie is a robot from Ashley Wood's various comic books and paintings. Even though I have all the books that his images are in, an Internet search also found collectible action figures of Bertie and various images of those action figures. I first created a reference folder of all the images I found. I also spent a good amount of time looking at the comic books, reading the stories, and thinking up scenarios that Bertie would be in.

Then, I thought of real-world materials and the wear and tear a robot like Bertie would manifest. I did a search for images of WWII tanks, metal barrels, and mechanical robots in extreme conditions and came up with some surfaces that I liked. I then determined the number and nature of paint layers I would need. I envisioned three distinct materials on my model of Bertie: the metal that makes up his body, the leather pouches, and the metal that makes up his gun.

I also determined that I would be working with four diffuse paint layers and one specular layer. The four paint layers would be as follows:

Base coat layer This would be the metal base that Bertie is made from. Through my research, I found that big lumbering machines are made of galvanized metal sheets that are cut and bent into shape. Doing research on the Web and using Photoshop, I made a tileable galvanized metal texture (included as galvanized_steel_tileable.tga on the DVD in the Stencils folder). The Web has multiple tutorials on how to make tileable textures in Photoshop and other applications. As the paint would scratch off Bertie, this layer is what would show up underneath. This layer would mostly take care of the metal portions of Bertie, and I would paint the leather pouches a neutral leather color to set their base coat.

Paint layer The paint layer would include the basic color of Bertie's different parts. His body would be a dark greenish-gray, the leather pouches would be black, and the gun would be dark gray steel.

Weathering layer This would be the grime, rust, and wear-and-tear layer.

Decal This layer would include all the spray-painted or glued-on insignia and stickers.

I also planned to include a specular layer that would distinguish between the various specularity areas of Bertie's body, gun, and leather pouches.

The Paint Layer

The next layer to be painted on is the paint layer. This will include the paint that is applied to Bertie, his gun, and his leather pouches. You will also strip away some of the paint in areas that could be scratched to reveal the galvanized metal beneath it. You will again block in the major swaths of paint and then texture the gun and pouches afterward.

1. Click the Layers window drop down menu and select New Layer from the drop-down menu. Create a new diffuse paint layer with a size of 2048 × 2048 (2K) and use the Save As Targa [8 bit, RGBA] format (.tga). Call it paint (Figure 4.12).

   

   Figure 4.12. Create New Paint Layer dialog box settings

2. Right-click the layer you just created and click Export Selected. Export it as paint.tga. Now right-click it again and click Delete Selected. You could have skipped this step and just created your image in Photoshop, but this process automatically creates an alpha channel for you.

3. Load the paint.tga file into Photoshop or your favorite 2D painting application and fill it with a uniform military dark gray-green or any other color of your choice and save it back out as a 24-bit .tga file. You can alternately use the camo.tif camouflage pattern I have supplied in the Stencils folder on the DVD. However, you will have to use the same technique we used with the galvanized metal texture to projection-paint the camo pattern on the seams.

4. Click the Layers window drop down menu, click Import Layer, and re-import the paint.tga file. Make sure it imports on top of the base_coat layer.

5. Select the Paint Brush tool and then a black color. Color the two holes on Bertie's head black.

6. Select the Paint Erase tool from the Paint Tools tray, and reduce the size to a small size such as 2 or 4, and select a strength of around 10. Choose the bw_bristol2.tif stamp, and start erasing areas that you feel might have paint chipped or scratched off, such as the borders of the hatch, the insides of the scratched and gouged areas on the model's surface, the edges where Bertie's head and torso would turn, and the sharp edges of the arms, legs, and fingers (Figure 4.13).

   Look at references of paint chipping off tanks, tractors, and construction vehicles to get some ideas of how paint would chip or scratch off a painted surface. If you make mistakes, you can use Undo to go back a few steps, or you can delete the paint layer and re-import a pristine version of it. You may also want to save different versions of your paint layer by exporting texture map versions such as paint_slight.tga or paint_extreme.tga.

7. When you are happy with your results, it is time to move on to the gun. Select all the gun parts in the Object List (Figure 4.14) and click Display → Hide Unselected. From the Select/Move Tools tray, click the Objects icon and click anywhere except on the gun in the 3D view to deselect the gun.

8. 

9. From the Paint Tools tray, select the Projection tool and start painting on the gun. Notice that as you paint on a surface that faces you, and your brush goes over the edges, you get streaks when you rotate the model (Figure 4.15).

   

   Figure 4.13. Paint chips and scratches

   

   Figure 4.14. Select gun parts.

   

   Figure 4.15. Streaks on the edge of a model while projection painting

   This is overcome with two easy steps. First, make sure your brush fits on the surface you are painting and does not go over the edge. Then rotate the gun so that the top edge of the gun is facing you at an angle, as in Figure 4.15, and projection-paint on the angle. This is a good exercise for you to practice projection painting and not have any streaks because you will be working on flat, angular shapes as well as the cylindrical shape of the silencer.

10. Remember that you can move, rotate, and scale the stencil by holding down the S key and clicking and dragging with the middle, left, and right buttons of the mouse. You can also press the Q key to hide the stencil and view your model; another press of the Q key toggles the stencil back into view.

11. After you are finished applying the stencil as evenly as you can, it is time to paint over it with a dark gunmetal gray while letting some of the underlying details show. Pick the Paint Brush tool and a very dark gray color. Set your strength to about 2. Gently paint on top of the stencil pattern, letting bits and pieces of it show through to get a realistic metal finish (Figure 4.16).

12. Use the same process to color the leather pouches. Find a leather reference or shoot your own. Isolate the pouches by selecting them and hiding the rest of the model. Use your images as stencils and projection-paint the pouches.

13. Save your scene, and then right-click the paint layer in the list of Paint Layers and click Export Selected from the drop-down menu. Save it as paint.tga.

In addition to using the Paint Brush and Projection tools, you also used the Paint Erase tool in this section.

Paint Erase is used to remove paint from the currently active paint layer. If a model has multiple paint layers, erasing paint on one layer reveals the visible layers below it.

Figure 4.16. Gunmetal finish

Compositing the Layers in Photoshop

Now that you are finished texture-painting the model, you can take the textures into Photoshop for further manipulation. Load the Bertie_map_composite.psd file (Figure 4.21) into Photoshop and notice the various layers and their opacity. This would be a good place to paint details that you were not able to do in Mudbox, because Photoshop has numerous brushes and layer-compositing tools that you can manipulate to get the exact look you are going for.

Figure 4.21. Texture layers manipulated in Photoshop

In this chapter, you have gone through texture-painting an imported model of Bertie the robot. You have seen how critical a good UV layout and efficient UV shell distribution in UV space are to the texture-painting process. You have learned the difference between materials and textures and how you need to reverse-engineer the look of objects around you to determine their material properties and their texture layers.

You also have learned about paint and texture layers and their functions in Mudbox. You have seen how in Mudbox you can use opacity levels and layer order to control how layers are composited on top of one another. You have learned the importance of gathering and compiling reference material to aid in the modeling and texture-painting process. And finally, you texture-painted Bertie with various layers to get the final result.