Hour 8. Character Modeling

Nearly all modern visual-effects-laden films and shows contain at least some character animation. From talking fuzzy animals to drooling insects, the television and film VFX markets rely heavily on modelers who can bring designs to life in 3D. This means that for your purposes as a CG artist, familiarity with the character-creation process is crucial. If you are planning on moving further down the character-modeling path, this Hour will be helpful for you to start getting accustomed to common considerations with CG character models.

In this Hour, you will encounter the most common issues surrounding character modeling and learn how to overcome them. You will set up a scene in a way that allows for easy character creation and accurate modeling from a design. You will also learn about modeling considerations as they pertain to a production. Like most of the CG disciplines, character modeling is subject to personal workflows. Therefore, as you read this Hour, take note of the concepts that seem to come naturally to you, and those that are a little more difficult, so that in your personal exercises later on you can focus on finding your own way to achieve the goals explained here.

To begin, we must first create planes that will serve as guides in 3D space as to where our geometry is supposed to be placed. This is commonly achieved in one of two ways: using camera image planes or using polygonal planes with the designs as a texture. The first way involves importing the image as a background in your panel—otherwise known as an “image plane” (more in Hour 22, “Working with Film”). The advantages to this method are that Maya automatically retains the aspect ratio of your drawing, so an extra step is removed. The downside is that to move, rotate, or scale these planes, you must do so through the image plane’s attributes in the Attribute Editor, which can be cumbersome when you are in the middle of modeling the character. Even with this drawback, the camera image planes are quicker and easier to create than polygonal planes, so we will use them.

First, we must locate our designs and make sure they have been created correctly. We just need to make sure that the front and side views’ images line up perfectly. Navigate to this Hour’s source files directory and open up design_Ortho.jpg (see Figure 8.2).

You can clearly see that the front and side views line up correctly. So in order to use these designs, we must load them as image planes on their respective cameras. Open Maya, and in any panel, go to Layouts, Saved Layouts, Four Panes. This will reset your panel layout to the well-known standard layout shown in Figure 8.3.

Next, we load the images as the image planes for the front and side cameras. In the front panel, go to View, Image Plane, Import Image.... Navigate to this Hour’s source files and choose image_Front.jpg. The image has been cropped for the front view. In the Attribute Editor (which should open automatically when you create the image plane), one of the very first sliders is the Alpha Gain. This controls how “see-through” your image planes are. Set the Alpha Gain to .5 for the front image plane as well as the side image plane when you are done importing it. The properly loaded image is shown in Figure 8.4.

Now do the same for the side panel. Go to View, Image Plane, Import Image... one more time and find the image_Side.jpg file. Once it is loaded, you can see in the Persp panel that both images are placed at the world origin, ready for your use, as shown in Figure 8.5.

We will want to move the image planes so they are not in the way of the character when we start to model him. Select the front image plane and press Ctrl+A to bring up the Attribute Editor. If you scroll down the attributes, you will see a section called Placement Extras. You can resize your image plane and move it around the scene using these attributes. The Image Center attribute controls where the plane is placed. We want to move the image plane backward so it will be behind the model, so change the value in the far-right box next to Image Center (this is the Z value, even though it is not labeled) to –8. Select the side image plane and make the X value (first box) in its Image Center attribute –12. Figure 8.6 shows the planes with their correct positions.

Once that is completed, we are almost ready to start creating geometry the same way we did with the hammer in Hour 4, “Modeling with Polygonal Geometry.” There are just a few more steps to take to make sure we are going to have the most efficient workflow when modeling. Create a polygon sphere by going to Create, Polygons, Sphere. Click anywhere in the front panel and press the 5 key to shade the objects. As you can see, the sphere is opaque and we cannot see our image planes through it. To see through objects in a panel, click on (in your panel) Shading, X-ray. RMB drag on the sphere and choose vertex. With the vertices highlighted, it’s easy to see how intuitive and simple it will be to line up the polygons with details in the designs.

To create a mirrored instance, we are going to start with a single polygon. Create a polygon plane by clicking Create, Polygon Primitives, Plane and then the options box . In the options box that appears, make sure the width divisions and height divisions are both set to 1 and then click Create. Now in the Persp panel, RMB drag and select the two vertices on the left side of the side image plane (the two vertices in negative X space) so that you can move them to the axis, as in Figure 8.7.

We are going to move these two vertices to the center of the world axis. Press W to switch to the Move tool. Now, holding down the X key makes it so that Maya will snap a vertex to the gridlines (the V key snaps to vertices, and the C key snaps to curves). With the X key pressed, move these two vertices along the X axis until they snap to the center (they will be aligned with the side panel’s image plane). Figure 8.8 shows the vertices in the correct position.

Now we will instance this geometry. Press F8 to return to object mode or RMB drag on the plane and choose “object.” Go to Edit, Duplicate Special and then the options box . In the options box, change “Geometry type” to “Instance,” and in the first value box (the X value) for “Scale,” enter –1 (remember that when you see three boxes, they are X, Y, and Z). Leave the rest of the settings at default and hit “Duplicate Special”. A new plane will mirror across the X axis (the result of scaling it by –1), and it is an instance of our first plane. The two planes are now linked together, as shown in Figure 8.9, and all of the edits and modifications you do to the geometry of one will be applied to both simultaneously.

Start by selecting the vertices and lining them up with the forehead in the side panel (see Figure 8.10).

Next, select the top edge and extrude it once (in the Polygons menu set, click Edit Mesh, Extrude), working in the side panel to line up the geometry with the design (see Figure 8.11).

Continue extruding the geometry until you have gone around the entire circumference of the character in the side panel. To connect the last two faces, line up the vertices as best you can in the side panel and then switch to the Persp panel. Select two vertices that are supposed to be connected and then select Edit Mesh, Merge. They should “weld” together.

Let’s create the edge loop that encircles the eye. One way is to create some polygons in the area of the eye and then use the Interactive Split tool and extrude to edit the polygons into shape. Instead, we will purposefully create the edge loops explicitly so that we know where they will be. Extrude the edges down the nose until your polygon is lined up with the edge loops drawn on the front image plane. Select the edge that is closest to the inside of the eye, as shown in Figure 8.13.

Extrude the edge in the front panel two times toward the eye socket (see Figure 8.14).

Move the vertices of the extruded edges to be in line with the eye socket, like I did in Figure 8.15.

Now extrude the edge all the way around the eye socket and then weld the vertices at the end when you are done (see Figure 8.16). Eight to ten extrusions is enough.

Now we must work simultaneously in the front and the side panels to move the vertices in the Z direction and line them up with the design in the side panel. Move them into place, as shown in Figure 8.17.

Remember from Hour 4 we talked about how it is ideal to have three- and four-sided faces in our models? One way to reduce the number of poles you have from the outset is to try as hard as possible to make sure you only have four-sided faces, and that your edge loops are as long as possible. This will help ensure you are not creating tangled nests of polygons. Chances are your edge loops are terminating at poles; therefore, if you double-click on an edge, follow the selected edge loop to its ends to discover poles.

A place where poles are guaranteed to appear is anywhere a face is extruded from a plane. The vertices already have four edges intersecting at them, and you are essentially adding one more. If you think of the most low-poly bipedal shape possible, you will see that poles are unavoidable wherever appendages are extruded (see Figure 8.19).

This character was designed with “mitten hands” so that it would not be too challenging to try to get the hands in place. Keep extruding, appending, and adjusting vertices to get the hands in place, as shown in Figure 8.20.

Working “low poly” means that you aren’t necessarily trying to model detailed parts of the character into the geometry. This is particularly true for details such as buttons on a shirt, belts, arm bands, and so on. These small details would be better suited to be separate objects that you either constrain or rig to move along with the character geometry rather than being built into the single character mesh. A good rule of thumb is that anything that is “skin tight” can be a part of the character model, and anything that is not should be a separate object. Just imagine how hard it would be to have good edge loops if you had to add really tiny details to a low-poly object.

The last advantage to a low-poly workflow is the fact that you are simply dealing with fewer polygons. When we get to rigging and skinning this character, you will see that having fewer polygons to deal with is extremely advantageous. Particularly when we are creating shapes for the eyelids and mouth for animation, our low-poly character will be very simple to set up to take on some very dynamic shapes.

Besides areas that are going to deform, you also want more density in areas that are going to get the most screen time. For this reason, the face and head are typically the densest regions on your character model. Taking a step back and looking at our character model proves this is true, even with cartoony characters, as you can see in Figure 8.25.

The one other consideration for density is for any effect that relies on polygonal detail. Some particle systems generate particles per face, and if your mesh is too lopsided in detail, it can create some odd effects. Some third-party fur plugins generate hairs based on polygons as well; again, having far too much detail in one section and far too little in another can mean unpredictable results.

When you are getting close to making your final adjustments to a model, previewing the smoothing is important. Keep a close eye on areas where the geometry is protruding or receding, because smoothing will reduce the contrast in these areas the most.

Under most circumstances, if you are using Mental Ray to render (and not the Maya software renderer or a third-party choice), smooth your geometry with Mental Ray’s smooth mesh preview (by pressing 3 with geometry selected). This is when smoothing can happen as a last step before rendering.

In some rare cases, certain deformers, constraints, or miscellaneous nodes depend highly on the geometry and you will need to smooth your mode using the Mesh, Smooth command. Things such as toon lines depend on topology, and in order for them to follow the contour of the model, you will have to smooth the mesh using the Smooth command and not Mental Ray smooth mesh preview.

With the W key pressed and held, LMB click in the panel and make sure your Move tool is set to “World” space (see Figure 8.28).

Go into the Move tool’s settings (Windows, Settings/Preferences, Tool Settings—or double-click the Move tool in the toolbox). Scroll down to the section called “Move Snap Settings” and uncheck “Retain component spacing.” This setting controls whether components you move are going to stay locked in their relative positions when you move them, or whether they will move absolutely. Now, with the center vertices still selected, hold the X key down and move the vertices on the X axis until they snap to the next gridline, and then move them back so they snap back to the center gridline. What you just did is moved all of the vertices to an adjacent gridline and snapped them to the grid and then moved them back into place. Now we know all of the center vertices are in the right position. Go to Edit Mesh, Merge, and the vertices will merge and close the center seam.

We will make a seam on each of his arms in the socket where it connects to the body, and a seam along the back of the arm so it can unfold flat. In the Persp panel, select the edges around the connecting point of both arms, select the edge loop at the wrist, and select the edge loop that runs through the center of the back of his arm, as you can see in Figure 8.30. In the UV Texture Editor, go to Polygons, Cut UV Edges.

Create another seam through the center of the back of his head, and seams around the top horns that protrude outward, using the Cut UV Edges tool. Also, create a seam around the edge loop just inside his lips.

With a little more cutting and sewing, you are ready to Unfold. Select each UV shell by clicking (in the UV Texture Editor) on a vertex and then Ctrl+RMB clicking until you see the option to convert “To Shell.” To unfold the UVs all at once, in the UV Texture Editor, go to Polygons, Unfold. With a little moving, rotating, and scaling, the UVs are complete (see Figure 8.31). Open character_Model_Final.ma to see what the finished product looks like.

A. The workflow does not change drastically between highly realistic characters and cartoony characters. The only thing that changes is that most realistic characters are exported to a high-polygon sculpting software after they are finished in low-poly form.

Q. My image planes are not lining up with each other. What is wrong?

A. When this happens, it is because your image planes are not the same size. It is always easiest to take advantage of letting Maya size and position your image planes automatically by making the side and front images have the same dimensions.

Q. When I combine the two halves, and I select all of the center vertices to merge them, they do not all merge. Why?

A. You might have accidentally moved the models, or the center vertices are not exactly lining up with each other. To increase the distance that two vertices will travel to merge together, increase the Threshold value in the Merge command’s option box.

Q. I noticed your character’s eye socket is empty. Shouldn’t we model an eyeball?

A. Remember any object that is not skin tight should be a separate object and not modeled into the character’s model. That goes for eyes and teeth as well. Best bet is to have a sphere in place as you are creating the eye socket to give you a sense of what the eye shape is going to be, and remove it before you are done. Alternatively, you can use that sphere as a good starting point for your eye once the body mesh is complete.

2. What is the attribute called that adjusts the transparency of an image plane?

3. A copy of a polygon that is mirrored but takes on all of the changes to the original is called what?

4. Name a few areas where good edge loops are important.

5. What is a pole?

2. The attribute is called Alpha Gain, and it can be adjusted in the Attribute Editor for an image plane. Remember that to see the transparency in the Persp panel, the “renderer” must be set to Viewport 2.0.

3. A “mirrored instance.” This “link” between the polygons is preserved even if you delete history, but not if you combine the two halves.

4. Eye sockets, cheeks, mouths, muscles, and any area that has well-defined, curved detail.

5. A pole is any vertex that is at the intersection of more than four edges. This happens most often when you extrude faces, or have non-four-sided faces in your mesh.