Next Gen Character Chapter 1
Take a sneaky peak into the first chapter of one of 3DTotal's groundbreaking eBooks in this short sample tutorial. If you would like to build upon what you learn in the following article or if you would like to continue to follow this tutorial you can purchase the full eBook in the 3DTotal shop.
Please note: This eBook includes free resources, which can be obtained by purchasing the full product by clicking the banner above.
There has never been a better time to create a next-gen creature. Tools like ZBrush, Mudbox, and 3ds Max are all weapons in our arsenal, allowing us to create believable, realistic, and detailed creatures that were previously too complex to even be considered.
We will begin this tutorial series by looking at creating a base mesh for our creature, designed by Richard Tilbury at 3DTotal (Fig.00). Using a combination of ZBrush and a variety of software packages, we will create a base mesh from ZSpheres, then retopologise and perfect our mesh in 3ds Max, Maya, LightWave and modo.
Since coming onto the market, ZBrush has revolutionised the video games industry. Detail that was previously only possible through meticulous hand painting can now be done quickly and easily because of this programme. Organic characters are now able to look just as realistic and stunning as their high-poly counterparts. ZSpheres are a wonderful way to create a quick, simple base mesh that can be retopologised (the process of creating new topology on a mesh) and refined in a general 3D application.
1. To create a mesh using ZSpheres, we'll first drag one out onto the canvas, and then proceed to add more ZSpheres off of the original. This will build up the character in the same way we would with bones. It's important with ZSpheres to plan out the construction of your character before you start. The original ZSphere should act as the pelvis or waist, with additional ZSpheres coming off to form the limbs and midsection.
First we click on the ZSphere icon in the tool palette, then left-click and drag to place a ZSphere onto the canvas. Release the left-click, and then hit the Edit button [Q]. Now our ZSphere is placed and editable, and we can go on to add new ZSpheres off of this original (Fig.01).
2. To save time and make our job easier we can use symmetry to automatically create the opposite side of the model as we work on only one side. To enable this in ZBrush simply press [X] or go to Tool > Symmetry > Activate Symmetry, and make sure the X button is highlighted. If you now hover over the model, you will see another cursor on the opposite side that mirrors the actions of the original cursor that we're manipulating (Fig.02).
3. Click and drag on the right side of the ZSphere to create what will be the hip section (Fig.03).
4. Create a new ZSphere off of the last one by using the same click and drag technique. This new ZSphere will act as the first knee joint (Fig.04).
5. We'll now use the brush move tools to position the section we just created. At the top of the window (providing you are using the default ZBrush interface), there are buttons labelled, Edit, Draw, Move, Scale, and Rotate. Click on the Move button and right-click over the model. Make sure the draw size is small, as a large draw size acts as a soft selection and will move other parts of the model as well. Decrease the draw size to the minimum and move your mouse away from the menu. Select the ZSphere we just created and drag it downwards and into position (Fig.05).
6. Now create a new ZSphere on the bottom of the knee joint, remembering to create it on the ZSphere itself and not on the intersection. Position this as shown and go on to create another two ZSpheres to act as the end of the leg and foot bones. We are going to move on to creating the mid-spinal sections and the chest. Create another three ZSpheres on top of the pelvis and position them as shown, using the Move tool as before (Fig.06).
7. Create new ZSpheres on the side of the upmost ZSphere to act as the clavicle, or shoulder area. I find that creating these sections, as opposed to just using that last sphere as the elbow, gives a nicer curvature for the underarm and shoulder areas once the model is meshed (Fig.07).
8. Now we go on to create two ZSpheres on top of the upmost middle sphere to act as the neck and head. It's important to position the head section straight up, as the topology of the top of the sphere lends itself to the top of the head, with the face being built on the front (Fig.08).
9. It's now time to create the arms, which will be built using the same methods as before. We create three sections to form the elbow, wrist, and hand (Fig.09a & Fig.09b).
10. The last ZSphere to be placed is the one on the back of the lower knee joint, duplicating the bone that sticks out in the concept drawing (Fig.10a & Fig.10b). Once all of the ZSpheres are placed, we can move into the refining stage, part of which is creating the hands.
11. Surprisingly quickly, we now have the whole character built and blocked out, and it's time to go over it with a tweaking pass. Use the Scale and Move tools to try and match the concept art in proportions, pose and scale. The closer we get now, the easier the retopology work will be.
A ZSphere model is nothing until it's meshed, and that's where Adaptive Skin comes in. Adaptive Skin is a method for creating a usable model from ZSpheres. It also comes with the handy option of previewing the model before creation. Open the Tool menu and find the Adaptive Skin palette.
There are plenty of confusing options in the palette, so here is a quick explanation of their functions (explanations of almost any tool in ZBrush can also be found by hovering over the tool and holding down Ctrl). Under the Adaptive Skin menu (Fig.11a) we have the:
• Preview button, which turns on and off
our meshing. You can use the hot-key [A] to enable and disable this. It's very useful to keep checking the mesh to see how your ZSpheres are looking.
• Density, which calculates the number of
subdivisions that the mesh should have. A density setting of one will be very low poly; a setting of two will be four times that, and so on.
• Ires (Intersection resolution), which is the number of consecutive ZSpheres that will be used to create a single object when the mesh is converted into polygons. This can be visualised by thinking of one sphere with a two-segment chain on either side. With an Ires of 0, each sphere will turn into individual spheres with a few joining polygons to connect them together. An Ires setting of one will turn the three middle spheres into one flowing cylindrical mesh, with the outer two spheres still maintaining their spherical shape.
• Mbr (Membrane Curvature), which adjusts the smoothness in the T- and L-shaped intersections. The higher the setting, the smoother the curve will be, whilst a low setting will maintain the shape of the right angle.
• MC (Minimal skin to child), which ignores the child* of the section of the object, using it to connect two parts together but not contributing any polygons. This allows smooth transitions between new branches and can help eliminate twisting and collapsing.
• MP (Minimal skin to parent), which ignores the parent** of the section of the object, using it to connect two parts together but not contributing any polygons. This allows smooth transitions between new branches and can help eliminate twisting and collapsing.
• PD (Pre-divide), which lets you specify a pre-division resolution to match objects which are attached using connector mesh or local mesh features.
* A ZSphere created off another ZSphere is known as a "child" object.
** The original ZSphere is known as the parent (think of it as a mother and daughter holding hands!)
Once you understand the features, Adaptive Skin is not too complicated, and it's a very quick way to block out a character, animal, alien - or even a car! So now that we have a good set of Adaptive Skin settings, we need to modify the mesh a little to make it optimal for animation and speed up the next optimisation section.
Place new ZSpheres at the join areas of the model, namely the shoulder, the crotch, and optionally the neck. Placing a sphere on either side of the knee and elbow joints gives us three edge loops close together - the optimal amount for bending in animation. You can add a ZSphere to an existing section by holding control and clicking on the section while in edit mode (Fig.11b - Fig.11c).
12. The next - and final - stage is to create the hand. The process is roughly the same as with the whole character but it needs to be a little more precise in the layout of the ZSpheres. We are creating mid-sections that will serve to create the palm and finger joints (Fig.12a - Fig.12d).
13. Follow along using the video and screenshots as references to lay down the ZSpheres. Once all are in place, it's important to check the adaptive skin [A] and adjust the hand to anatomically correct proportions as much as possible. Creating a natural, relaxed pose here will be of great benefit to us later (Fig.13).
14. Our character base mesh is now built and ready to be exported into our 3D application. No UVs have been set up yet, as we will do that inside our 3D application later. Before we can export we must turn our preview mesh into a usable one by simply clicking the Make ... (Fig.14a).
This just created a duplicate model, which ZBrush swapped the active tool to and began using as the primary model. This new model needs to be converted to a PolyMesh3D object which we can export. Under the Tool menu, click Make Polymesh3D (Fig.14b).
Now all that is left is to click Export (in the Tool menu), give it a name, and save into your project location ready for the next step.
15. Congratulations on creating your base mesh so far (Fig.15). With the knowledge learnt from this section you will be able to create a base mesh for almost any organic object and know how to create an optimal mesh using Adaptive Skin! In the next step we will take the exported model into our 3D application, give it new topology, and perfect the shapes and forms ready for sculpting and animation. Please continuing reading for Part 1B.
Part 1b - 3ds Max
Optimisation and topology are vitally important in real-time character creation. Whether it's for video games, the web, or interactive media, polygon count and topology should be high on your priority list. If you have a budget of 10,000 tris for a character (most game studios work with tri count as opposed to poly count), plus clothes and accessories, there are a number of considerations to decide on while building. Things like: How am I going to spend those tris wisely? Which areas should have the most tris? How can I go about creating animation-friendly topology with as few tris as possible and how can I keep the silhouette as smooth and edge-free as possible with the minimum amount of tris? These questions are vital, and should be asked all throughout the process of modelling a character. They should be considered at each stage and should be present in all your decisions during this part.
I am using 3ds Max 2009, but older and newer versions will not be too different. On the left of the screenshot I have a modelling toolbar with very useful scripts and features by a genius Maxscript artist, Remus Juncu (http://313designstudio.com/rappatools) (Fig.00). Common selection shortcuts, view shortcuts, edge loop tools, wire colouring and much more are all part of this absolutely free toolset, and I find them extremely useful while modelling or retopologising models. Feel free to go and try them out and see if they can improve your workflow. I won't be using any of the functions during the tutorial except the view and selection shortcuts, so don't worry if you haven't got them installed.
1. We'll start off by importing the model into 3ds Max. Open 3ds Max and go to File > Import. In the File Type dialogue, find Wavefront OBJ, and select the exported model that we created in Part 1A (Fig.01).
2. When I import with the default options the model application is flipped 180 degrees, upside down. In the co-ordinates boxes at the bottom of the screen, enter 180 into the Y box so the model will rotate right-side up (Fig.02).
3. Now it's important to convert our model from editable mesh to editable poly. There are a few reasons for doing this: editable poly is a newer and more complete modelling solution, providing us with access to the newer 3ds Max modelling tools; It also eliminates the possibility of double-sided polys and extruded edges, which in most game and real-time engines will cause errors or fail to export. Right-click over our viewport and click Convert To > Convert to Editable Poly (Fig.03).
4. Unlike clay or wood sculpting, 3D artists have the luxury of symmetry modelling, with one side automatically updating as we work. To do this in 3ds Max we need to remove one side of the model and apply a symmetry modifier. Select half of the mesh after switching to the front view (default hotkey [F]). Make sure that the Ignore Backfacing option is turned off. Selecting half of the model by using a rectangular selection will select both the polygons facing and not facing our camera (Fig.04a - Fig.04b).
5. On the right of our Max screen we have the modifier stack. This is a collection of all the modifiers that can be applied to our currently selected object. Click on the down arrow to open the dropdown box, and select Symmetry. You can also drop down the box and start typing the name of the modifier to quickly select it. The modifier is applied to the object and we can adjust the settings.
In the settings for the symmetry modifier, the correct axis to use is X, as we want the model to be mirrored in the X plane. We want Weld Seam to be enabled, as this will weld each vertex along the symmetry plane together. The threshold setting decides the distance a vert has to be from its mirrored version to be welded together: a too high threshold results in collapsing of too many verts together; a too low setting results in un-welded vertices. Reduce this number to zero, and while holding Alt, drag the spinner upwards until you notice all the vertices have been welded (holding Alt while manipulating a spinner in Max gives you a finer control of the increments adjusted) (Fig.05).
6. Click Editable Poly underneath the newly added modifier. We will work underneath the symmetry modifier for the remainder of this section, collapsing again when ready to export, with no asymmetry being put into the model pre-high poly.
We can move onto cleaning our model now, ready for adding and modifying the topology. The area I always start with is underneath the arms. I want all areas of the mesh to be visible, with no intersecting polygons. Go ahead and enter Vertex sub-selection mode  and select vertices in the arm region. Select all that are intersecting, plus a few more in the surrounding area to ensure a smooth transition (Fig.06).
7. Now, as we did in the earlier step, find the Relax modifier from our modifier stack. Apply this to the model by selecting it from the list and adjust the settings to match the ones in the image (Fig.07). As mentioned before, we are trying to eliminate any overlaps or intersections to assess the topology properly.
8. With the shoulder area relaxed, collapse the Relax modifier by right-clicking on it in the list and choosing Collapse To. We now have a collapsed object, without affecting the Symmetry modifier. Go on to relax and collapse any other areas of the mesh that are intersecting. This might include - but is not limited to - the elbows, crotch, backs of knees, and the hand.
In this step I have added edges running from the lower back to the neck. The head simply doesn't have enough detail to provide a clean and smooth silhouette. The broadness of the back dictates our need for more edges in that area (Fig.08).
9. Here I am shaping the new edges into place while in Vertex mode. If you enter Screen mode for translation of objects (Ctrl + Alt + Right-click and select Screen) then you can simply drag vertices around one by one without selecting and using the manipulator (Fig.09).
10. I've added in an edge in the neck to show the sternomastoid muscle. This will start off the flow of topology I want in this area. I'm not worrying too much about keep everything quads (four-sided polygons) at this point as I want to rough out the flow of the edges and then tidy up once that is in place (Fig.10).
11. I'm adding a row of edges in to mark the underside of the chest. The edge loop also runs into the shoulder area to mark the insertion point of the main shoulder muscles. This gives great deformation when the character lifts up his arm. I'm adding in these edges using the Cut tool, which I have on a shortcut key as it's a commonly used tool (Fig.11).
A neat and timesaving trick is to enable snapping when cutting (default hotkey [S]). Right-click on the snap icon in the top toolbar (it will highlight on and off when you press S), then turn all options off, except vertex and edge.
When you normally cut, it can be difficult to see if you're on a vertex or an edge, and many loose vertices get created due to this, leaving the model quite messy after a while. It then requires a lot of welding verts and deleting edges to clean it up. An alternative is to try the above method technique of snapping, which will lock your cut along edges and snap to vertices.
12. I go on to further relax the elbow and wrist sections. As you can see, the topology of the hand leaves a little to be desired. The wrist will not animate well this way and the fingers need a lot of work to get them to a good stage for sculpting and animation. The crotch area topology came out almost perfect, showing the power of ZSpheres (Fig.12)!
13. I add an edge here to define the wrist and topside of the hand. I remove a few edges on the hand by right-clicking, holding down Ctrl and clicking Remove. By holding Ctrl and clicking Remove, it removes edges and vertices at once - a clean remove. In general, you should always be removing this way to keep the whole model clean, as we did with our snap cutting (Fig.13).
14. I'm working on the underside of the hand, re-flowing the topology to form the two major padded sections of the palm. It's important to note here the flow of the edges around the thumb, as this will later be very efficient in pivoting the thumb around. All of this is done in the same manner as before, adding and removing edges to optimize the flow (Fig.14a & Fig.14b).
15. The top of the hand topology should flow out of the fingers and into the wrist, just as the bones and sinews do anatomically. I reduce the edge loops around the fingers to adjust the joints and a connector loop near the knuckle. I'm also making an attempt to clean up the hand by removing and adding edges. The aim of this is to end up with all quads. Of course, as this is for real-time, quads are not as essential as for film or other models. The main issue is that triangles and 5+ sided polygons do not smooth well. As real-time models are not smoothed with turbo or meshsmooth we do not have that consideration. In fact, for real-time models, triangles are a tool that we can use to reduce the overall poly count or create points like the elbow or ankle bones without using too many polygons (Fig.15).
16. To further work on the hand I need to hide a large portion of the mesh. Sometimes it can be difficult to see the area you're working on when the rest of the mesh is in the way. Select everything but the hand up to the elbow and in the modifier stack, scroll down, and press the Hide Selected button (Fig.16a & Fig.16b).
17. I create points on each knuckle to reduce the poly count and allow the hand to flow into the wrist without adding additional polygons (Fig.17a - Fig.17b). Hands are notoriously difficult to model as they can bend and move into a number of awkward positions. As people are so aware subconsciously of the anatomy of hands, as we are of faces, it becomes easy for observers to spot mistakes automatically. Unfortunately, without repeated study we don't have enough information in our heads to fix those mistakes by eye. For that reason I advocate dedicating a portion of your time to studying the anatomy of the hand, the muscles and bones that drive it, and how they interact with each other. After a number of drawings and sculptures the information is burned into your brain for use in the future. The more information we store, the more we can recall at a later date, allowing us to work more creatively, efficiently and with more confidence.
18. I work further on the palm and thumb connection point. As this will be the model to be sculpted in ZBrush later, it's important to maintain an equal size of polygons. The only areas of the hand that need more numerous and smaller polygons are the joints, wrist and perhaps the fingertips (Fig.18a - Fig.18b).
19. Select the edge loops that make up the joints of the fingers and thumb - two around each finger, and one on the thumb; right-click over the model and hit the little icon next to the Chamfer button. Selecting the Chamfer text will use the default settings without popping up a menu box, whereas we want to click the icon so we can dictate those settings ourselves.
Increase the size of the Chamfer to approximately the size in the second screenshot here (Fig19a - Fig.19b), and increase the number of segments to 2. Now we have created optimal topology for our finger joints that will also deform well.
20. After doing small tweaks on the model throughout this entire process, some of the centre line vertices have been moved off centre. The welding no longer matches up and a clear seam is visible. To get this back to an accurate state go to the Editable Poly level of your model and press  on your keyboard, or select the Edge Borders sub-selection mode. Click once on one of the edges on the centreline and Max will select the whole loop.
With the loop selected, in the modifier panel find the Make Planar button. On the right of the button are three letters corresponding to the various axes. These buttons take the current selection and line every vertex up in the chosen axis. As the symmetry modifier is in the X plane, we can go ahead and select the X icon. You should see the edges line up as you click. We are not quite done, because the lining up averaged all the vertex locations and now they all lie in that position. Again in the co-ordinates input boxes along the bottom of the screen, type zero in the X box. This puts all the vertices back along the X axis and our symmetry modifier works once again (Fig.20a & Fig.20b).
21. Finally, it's time to work on the head again, and we start by blocking in the eye areas. If the eye holes are in the right place it makes it very easy to locate the nose, mouth, and ear (Fig.21).
22. I apply a new material to the object to see how it looks in a silhouette form. During the whole process of building the model we should be looking out to make sure the silhouette looks interesting, proportionally correct, and as close to the concept as possible. To apply a new material to check this there are two ways, as follows (Fig.22a - Fig.22b):
• The first is to simply press [M] to open the Material Editor and change the diffuse colour of the applied material to pure black. Then deselect the model and view the silhouette.
• The other - and perhaps better - way is to select the model, and above the modifier panel where the model's name is written, click on the colour swatch next to it. This is the colour of the wireframe, the object's default colour without material. If you apply a pure black colour here you can then go to the Display tab (the fifth tab from the left), and click on the checkbox for Object Colour. The model will now be displayed simply with the colour of the object and no lighting. This is very useful for working on the silhouette of the model and versatile in that you can simply switch back to material colour when you want to go back to a lit model view.
23. The legs are going to need some attention before proceeding, and I'm going to start by simplifying the connection between the sticking out bone and that second joint - it's more twisted than I would like it to be for a simple base mesh. I go about reworking the area; I remove a number of edge loops in the bone section, and in the foot also. It's important to think which areas will be deformed during walking or action with the character and put more polygons into those sections. Areas that will always remain static, and especially areas that will remain straight, need far less polygons (Fig.23).
24. Flatten out the bottom of the foot and shape it so it resembles the concept art. It's important to note the foot is basically a hoof, and is solid. There are no toe areas, so the foot will not bend when walking. This means we can reduce the polygons on the topside of the foot by collapsing some edges and forming triangles (Fig.24).
25. Continuing the head, I'm going to put in edge loops that range from the bridge of the nose to the bottom of the chin. Even if it's not perfectly tidy, it's important to block out all the basic topology loops in the face so that we can later alter the shape and quickly add detail by inserting new edge loops (Fig.25). Just as with a nurbs model we might create these sections separately, stitch them together, and then add detail through isoparms.
26. Now we're going to create the base for the eyes. Creating circular loops around the eyes makes it easy to add additional loops later, but also creates the most optimal topology to animate the eyes. Select the polygons around the eye and right-click over the model, selecting Inset. The cursor will change, and click-and-dragging over the selected polygons will create a loop inside of the selected one (Fig.26).
27. Like the eyes, I'm going to create more loops inside of the mouth section. I create circular loops around the mouth, making sure to position them at key points that will affect the silhouette; for example, the top of the upper lip, which in its circular form will create the bottom part of the lower lip (Fig.27).
28. I've selected the newly created eye polygons and created another inset. The size of the inset does not matter here as we will right-click again on our object and hit Collapse. All selected polygons will now have been collapsed into a single vertex. I'm also insetting polygons on the side of the head to form the ear. We want the topology of the top and side of the head to flow into the ear naturally, as without the benefit of mesh/Turbosmooth, any ugly topology connecting to the ear at harsh angles will appear darkly coloured due to the way realtime engines handle smoothing groups (Fig.28).
29. Add further loops to form the chin and jaw, making sure it follows the jaw line all the way up to the ear. As the nose does not really stick out of the character in the concept, we will not add topology for it. Rather, we will use the normal map from the high-poly sculpt alongside a good texture to achieve a realistic result. Not all detail needs to be in the base mesh, only detail that effects the silhouette, or will do when the character deforms (Fig.29).
30. Let's not forget the back of the character. It's great to work on a specific area for a long time, but it's easy to get too focused on a particular area and go into too much detail before it's time. It's important to work the overall shape of the model and try to work in stages. Each stage should add another level of detail to your model, as if you were subdividing in ZBrush, or adding another level of smoothing. That means detail like the eyes, small muscles, or veins should all come at the end of the whole process (Fig.30).
31. I add a few more edges into the ear and continue to tidy up the area. As the image shows, the newly created polygons of the ear are not smoothing in the same way as the rest of the model. The reason for this is smoothing groups, an important part of real-time modelling as it gives us the opportunity to create hard edges and smooth surfaces without using more polygons. Smoothing groups tell the application how to display the model. A six-sided box with a smoothing group on each of the six sides looks like a standard box. If we apply only one smoothing group to the entire object, the programme will attempt to smooth it into one surface, instead of six, making the box appear somewhat spherical. As this is only a lighting effect and does not alter the polygon count or the actual model, it is simply an illusion. Because a silhouette is a shadow and lacks all forms of light, smoothing groups won't affect the appearance of the model's silhouette, which is just one more reason why it is so important to use silhouettes and smoothing groups in conjunction throughout the entire modelling process (Fig.31).
Setting up smoothing groups is not something we will do often with an organic character. It's much more common to use them for accessories, weapons, cars, and most game-ready models.
32. Enter Polygon sub-selection mode and hit [Ctrl] + [A] to select all polygons of the model. Scroll down in the modifier panel to Polygon Smoothing Groups. The number one is invisible, which means that not every polygon is part of group one. We can therefore deduce that, because no other smoothing groups are invisible, the polygons that are not in group one are also not part of their own second smoothing group. In order to make sure the entire model is all part of the same smoothing group, simply click on the invisible 1 button and see the ear and eye areas smooth over (Fig.32a & Fig.32b).
33. Applying the same principles we've used throughout the tutorial so far, I'm working on the neck area and shaping the back of the skull. I want the large neck muscles to be prominent in the base mesh and follow the muscles' direction so the area will animate well when the character turns his head (Fig.33).
34. Continue to work on the model by tidying up all areas until you are left with mostly quads. If you do have triangles, ask yourself why they are there. They should only be there if they are performing a function, whether it be for the silhouette, polygon reduction, or to help deformation (Fig.34). It's good to ask yourself these questions throughout. If you don't, the programmers will later on.
35. For the final section of this part, I've used a combination of moving vertices using soft selection and hand pulling the model around to work further on matching the pose and proportions of the concept. What we have at the end is a base mesh that will not only work well for sculpting our high poly, but also animate well, run efficiently in our real-time engine, and look good from all distances due to our care and attention to the silhouette (Fig.35).
Next up is Chapter 02 in which we'll be sculpting the real meat of the character, muscles, veins and all!
Please note that this series is also available in Maya, LightWave and modo, all available in the 3DTotal Shop.