Modeling a POLYGON Head in maya

There are many approaches to modeling the human head. Most of these methods were designed to accomplish specific goals such as easier sculpting of facial features, better facial animation, easier texturing, and so on. Few of these modeling techniques have the scope to accomplish all the necessary goals that are necessary for a realistic head and its subsequent facial expressions.

Modeling a Polygon Head

Although polygons have some disadvantages, they also have some compelling advantages. One of the main ones is that extra polys can be inserted into areas that require more detail. NURBS and splines, on the other hand, run throughout the mesh. When one inserts extra isoparms, these will appear from the beginning to the end of the mesh. In other words, if an isoparm is inserted at the eyelid, then this extra curve will flow all the way down to the base of the neck. If your software is capable of modeling with hierarchical b-splines (h-splines), then this will not be a problem. H-splines allow you to have one mesh with varying levels of detail in it.

Another plus with polygons is that isolated areas can be selected and worked on while hiding all the rest. This is usually not possible with spline or NURBS meshes, since they are one connected unit.

Some people may find it easier to texture polygon surfaces. Rather than trying to work with complex parametric mapping, they can simply select specific polygons and assign a name and texture to them.

Welding points to connect polygon surfaces can yield seamless models that do not have to update like NURBS models with fillet blends.

Although there are many techniques for polygon modeling, one of the best and also favorite methods for this author is to use the same approach as was discussed previously for modeling a NURBS head. Just as before, vertical curves that begin on the inside of the mouth and radiate outward to follow the contours of the features and finally end at the base of the neck are used (Figure 7-28). The lines again follow the direction of the muscles.



Figure 7-28 All the splines originate on the inside of the mouth and radiate outward to end at the base of the neck. The dashed line indicates the beginning of each spline.


The Spline Cage Method

To start, you may want to load several image files of the front and side of the head. An image of the top of the head is also useful but not as important. You might also decide to use an existing model of a head and then shape the curves around it. Some head templates and models of heads can be found on the accompanying CD.

Beginning with the first curve, start drawing it from the inside of the mouth, curve it over the top lip, around the nose, forehead and skull and end at the base of the neck (Figure 7-29).



Figure 7-29 The first curve defines the profile and middle of the head. It starts on the inside of the mouth and ends at the base of the neck.

This curve is located in the middle of the head and forms the outline of the head's profile. The side view is used most often for this outline. Once you are finished with this curve, you can either delete extra points or add some around the mouth, nose and eye areas. In total, you should you should have approximately twenty four to thirty points on the curve. Most of these should be clustered around the lip, nose and mouth. Since you might duplicate this curve to make all the rest, you do not want too many points to push and pull around. On the other hand, it is best not to have to worry about inserting points later on for extra detail. Most modeling packages that implement lofting or skinning work best when each curve has the same amount of points. This is also true when making a spline cage. All the points on each of the splines will be manually selected by you in the proper order and then connected as cross sectional spans. Therefore, if any curves have more points than other ones, it would be difficult to find a way to connect these extra points.

The total amount of curves for one half of the face should not be greater than thirteen. The face in Figure 7-30 uses eleven splines and twenty four points on each spline. If you can manage to keep the points and curves to a minimum, you will find it much easier to animate the face later on. Surfaces also remain smoother when they have less points and curves.



Figure 7-30 The curves that define the facial features generally follow the direction of the muscles.

If you plan to use copies of the original curves, then the next step is to duplicate the first. Rather than moving the entire curve away from the first, it is better to select individual points. These should be moved parallel to the first curve's points. Copy the second curve and move its points to continue shaping the outline of the facial features and skull. If your software has layers, then it might be easier to copy the first curve, paste it into another layer, make the first layer visible underneath it and then move the points on this second spline. Once the second curve is completed, you can then paste it back into the first layer. You can get an idea of how most of the curves and points will be placed by studying Figure 7-30.

If you are using a three dimensional head to model from, use the wire preview window to move individual points according to the wireframe's surface. The wire preview can be rotated until the selected point is visible against the edge of the template's surface. This will let you see the relationship of your points to curves on the mesh.

Duplicate and move the new splines. Detailed areas such as those around the wings of the nose and the nostril can saved for later. For now, just make the basic outlines of each section. Later on, polygons can be subdivided and their points moved for all the minute parts.

Check the flow of the curves so that they flow evenly in both directions. The connecting spans can be seen by temporarily selecting points in the right order and making a connecting curve between the spline. Figure 7-31 illustrates how one can select points in order and then connect them. Smooth out any severe angles.



Figure 7-31 The splines are joined by selecting the cross sectional points in the right order and making them into a curve. The connecting spans will create a spline cage.

After a while, you should notice that the duplicated curves become shorter since they no longer have to cover as great an area. Subsequently, the number of points on these curves squeeze together into a tighter space. Points that are closer together can make the surface bumpier. This type of problem can occur around the cheeks, mouth, and the chin. Even though correcting individual points can be tedious, in the long run it will save you time.

Once you finish making all the curves for half of the face, then connect each set of points (Figure 7-31). Be sure to select each point in the right order. Once they are selected, make a curve out of them. This spline will connect the original curves. Continue selecting the cross sectional points and making curves out of them. Once you are finished making all the connecting splines, you should have a spline cage that looks somewhat like Figure 7-32. At this point, it would be a good idea to tweak the points on the cage to smooth out any bumps. Compare the spline cage to your template(s) to see how the points match up.



Figure 7-32 The resulting spline cage. Notice how the cross sections follow the contours of the facial muscles.

When you are satisfied with all the curves and the placement of their points, then patch the cage with a polygon mesh (Figure 7-33). .



Figure 7-33 Once the spline cage is patched, the form becomes a solid polygon model.

This would make a polygon mesh on top of the spline cage. Software like Lightwave 3DTM uses an Auto Patcher plug-in. The subdivision level should be low so that you can edit parts without worrying too much about detail. A subdivision level of one will give you a basic form that can be edited. Higher subdivision levels would make it too difficult to shape details since it would generate many small polygons. One can always subdivide polys later on. It is easier to do that than to go back to a simpler shape.

The polygon model should now allow you to see a shaded preview. This should make it easier to fix any trouble spots. At this point you are no longer working with the spline cage, but rather directly with the polygon mesh. You can begin to add points and split polys for details such as the nostrils, lips, and so on. It would also be a good idea to save the spline cage separately from the model. In the future, you can use it to create other types of faces by changing its basic structure.


Creating the Eye Area

Since the original curves start in the mouth, they are not really facing in the right direction for modeling the eye lids. In order for the lines to follow the direction of the Orbicularis Oculi, the circular muscle circumscribing the eye, it makes more sense to cut a hole for the eye. The eye lids are modeled separately and then welded into the eye hole.

Figure 7-34 shows the area around the eye that should now be modeled. It can be made from a spline cage or a series of concentric points that are connected to form polygons. Be sure to add an extra row of points for the thickness of the eyelid. These will form a shape that extends into the eyeball.



Figure 7-34 The area around the eyeball is modeled separately.

The next step is to cut holes for the eye sockets. Select the points around the perimeter of the eyelid shape that you just modeled (Figure 7-35).



Figure 7-35 Points along the perimeter of the eyelid mesh are selected and copied.

Copy and paste the points into another layer. Select the points in order and make a polygon out of them. Make the eye lid model visible underneath this layer. Enlarge the polygon somewhat like the one in Figure 7-36.



Figure 7-36 After copying and pasting the points into another layer, they are made into a polygon which is then enlarged.

Make the head mesh visible in the front layer and hide all the polygons except for those around the eye socket. Make the eye socket polygon visible underneath the layer with the face (Figure 7-37). Use a 2D drilling operation to cut a hole out for the eye socket.



Figure 7-37 The outline of the eye socket is drilled onto the face mesh and then cut away.

After drilling the eye socket hole, weld any extra stray points around the hole opening. Paste the previously modeled eye lid object into the layer containing the head (Figure 7-38).



Figure 7-38 Once the eye socket hole is cut out, then paste the eyelid model into the face mesh layer.

Select the points around the perimeter of the eye lid object and the edge of the eye socket hole. Create connecting polygons between these two objects so that the eye lids are now part of the same mesh as the face (Figure 7-39). Connect or weld any extra points so that you only have three or four-sided polygons.



Figure 7-39 The upper eye lid folds under creating a line that should be modeled by splitting the polygons with extra points.

Figure 7-39 also shows how the eye lid folds under creating a thick line. It is very important to model this part of the eye lid since it gives a lot of character to the face. Insert extra points in an oval direction around the eye lid, select them in order, and then split the polygons. Once you have several extra half oval lines on the lid, then you can move them forward and back to form the eye lid fold. Approximately three or four sets of semi-oval lines make up this form.

The next step is to model the eyeball.


Making an Eyeball



The eyeball can be a simple primitive ball or sphere. The labeled iris surface is inverted to form a concave shape on the white eyeball (Figure 7-40).



Figure 7-40 Making the iris concave adds to the illusion of the iris having depth.

The pupil is simply a hole in the middle of the iris. The pupil appears black because the inner surface of the white eyeball is colored black. You can select the back half of the eyeball and flip the polygons inward, name the surface and make it black. Model a transparent cover for the entire eyeball (Figure 7-41).



Figure 7-41 A transparent cover makes the eye appear glossy and wet.

An eye socket should also be modeled. By selecting the points around the inside lid, copying and pasting them into another layer, and then selecting and making a polygon out of these, one can proceed to either extrude and reshape the socket or attach polygons to a larger version of a partial eyeball (Figure 7-42).



Figure 7-42 The eye socket shares the same points as the inside of the eyelid. It is also slightly larger than the eyeball.

Another small detail that can be added is the pink membrane in the corner of the eye. This can be modeled from a squashed half sphere. If you want to show a little liquid at the bottom of the eye, then model the shape and attach it to the points along the inside of the eyelid (Figure 7-43).



Figure 7-43 A long thin shape that runs along the bottom eyelid can simulate liquid in the eye.

Through the skillful use of shaders and textures, you should be able to get a fairly realistic eye. Be sure to give the transparent eye cover a high specular level and glossiness. The liquid in the eye object could have about a 70% transparency and a fractal bump map. The iris could have a real iris texture such as the ones found on the accompanying CD. Figure 7-44 shows an example of a finished eye.



Figure 7-44 The finished eye.


Extrude Method



The extrude method subdivides the face into vertical slices. The head is built one section at a time. Compared to other methods which require one to model a face one point or polygon at a time, it is much quicker. This technique is also more accurate and less haphazard than modeling a head out of cube or sphere by pushing and pulling points.

Before starting, you might want to use some 2-D or a 3-D template in another layer as a guide. Since it is always a good idea to model in real world measurements, size the template head to approximately 8.7" or 22 cm from the top of the head to the bottom of the chin.

In the side view, draw a series of points to outline the profile of the face (Figure 7-45).



Figure 7-45 A series of dots are created to form the outline of the face.

Select the dots in order and make a polygon out of them (Figure 7-46).



Figure 7-46 The dots that outline the profile are connected.

Select the large polygon and extrude it to about the width of half of the nose (Figure 7-47). The extrusion should be about .26" or 6.6 mm.



Figure 7-47 Extruding the first polygon.

Select the points along the newly extruded profile and drag them in to follow the contours of the face. Depending on which direction that you extruded, you are shaping the profile of the features and the shape of the skull that is located to the right or left of the original profile (Figure 7-48).



Figure 7-48 The extruded form against a shaded view of the template. The points are dragged in to form the second profile.

Select the second large polygon that outlines the inner part of the head's profile. This is the one formed from the points that you dragged inward. Extrude this polygon about the same distance as the last one.

Select the points along this newly formed profile and drag them in to outline the features that fall along this axis (Figure 7-49). You are now shaping the third profile.



Figure 7-49 The points along the second extrusion are moved to outline the inner contours of the face.

Once you are finished shaping these new contours, select the new large polygon. This is the one formed from the third contour. Extrude this polygon to make the next slice of the face. Move the points along this new contour so that they are placed against the features of the face. At this point, you are most likely forming the side of the nose, middle to end of the lips, chin, and so on.

Continue to select the newly formed large polygon profiles, extrude them and shape their points (Figure 7-50). Once you are past the features of the face such as the eye, nose and lips, you can make larger extrusions since you only need smaller polys for details. Try not to fuss too much with detail. You only need to form the general features. Later, you can come back to work on the minute parts such as eyelids, nostrils, and so on.



Figure 7-50 As each section is extruded, its points are dragged to form the shape of the head.

As the extrusions approach the ends of the face, around the ear, you should notice a compression of the polygons. Since the same number of polys are moved into a more limited area, they become smaller in scale. You can now start to merge some of these little polygons. Once they are unified, you can delete their extra points. Other polygons have a tendency to become stretched during the point dragging process. These will have to be split into smaller ones.

Once you make it to about the same point as Figure 7-51, delete all the large profile polygons. Since you will no longer extrude these, they no longer have any use. Do an automatic merge of all the points. This will weld all the points that occupy the same spaces and insure that the face is one seamless mesh of polygons.



Figure 7-51 The hole at the side of the face (white area) will be subdivided into smaller polygons.

You can now connect all the points that form the perimeter around the empty space near the ear (white area in Figure 7-48). Once the points are connected to make a large polygon, subdivide this into smaller ones and move their points to finish the side of the face (Figure 7-52).



Figure 7-52 Once the hole is subdivided, its points are pulled out for the final overall shape.

The head is completed by adding details around the eyes, nose, lips and so on. The surface can then be named and a smoothing algorithm is then applied to it (Figure 7-53). Later, you will be shown how to model the ear. If you want to see what the final head will look like, then duplicate, mirror and merge the middle of the head points.



Figure 7-53 The final head with details and smoothing applied. The ears will be added later.


The Single Polygon Method

Those who want the ultimate degree of control, when modeling, will find that using single polygons is the preferred choice. This method requires that one polygon at a time should be drawn until the entire face is completed. The approach is similar to creating a mosaic in three-dimensional virtual space. It is also the most time consuming technique.

If you are not familiar with using a single polygon creation tool, then it is probably best to start modeling a broad area such as the forehead. Using a pen tool, you can start clicking to form a rectangle shaped polygon. Make your polygons fairly large (Figure 7-54) and do not worry if the polygons are non-planar. Later, they can be subdivided into smaller triangle shaped ones.



Figure 7-54 Single polys form the forehead. The arrows indicate adjoining points that are welded.

Using two or three dimensional templates as guides, move the newly created polygon, into the right spot next to the previously created one. After you draw each poly, select both of the rectangle's overlapping points and those of its neighbor and weld them (Figure 7-54). The idea is to create one continuous mesh of polygons that share corner points.

Moving across the forehead create the polygons for the top, the skull, sides and back. Since the face will require more detail and thus smaller polygons, you might want to save it for last.

Rather than constantly creating new polys from scratch, you can select three points from adjoining polygons and make a triangle out of them. Figure 7-55 illustrates how one can select points from two polygons. A fourth point is inserted on this three-sided poly.



Figure 7-55 Individual points are selected from neighboring polygons.

This extra point is then moved into the right position (Figure 7-56). This will save you time since you do not have to weld points nor move more than one point into the right spot.



Figure 7-56 Once the triangle is formed, a point is inserted and moved to the right location.

Facial characteristics should be modeled last since they require more attention to detail. Sometimes it is best to copy points or create new ones, position them to follow the muscles of the face, and then connect them to make polygons (Figure 7-57). Working with points allows one to work at the minutest level. Specific forms such as the features require this type of attention.



Figure 7-57 Single points are created for the facial details and then connected to make polygons.


This tutorial is an excerpt from Mastering 3D Animation
Peter Ratner is the author of 3-D Human Modeling and Animation.
He is the founder of the Computer Animation program at James Madison University where he is currently teaching.