This tutorial explains some techniques for creating
a 3D human in a generic fashion,
i.e.without referring to any specific 3D program.
Building your portfolio for employment
Character creation is one skill that every employer
looks for in a portfolio. This is because it provides
a good reference to judge a potential employee’s skills.
If you can model and texture convincing characters then the chances are that you can model/texture almost anything. Of course this means that everyone expects high standards of character models — they are, after all, our emotional and visual link to the game world in many games.
Not every portfolio actually requires character models in it; if you do not have the skills to create good characters it is better to leave them out and concentrate on well constructed and textured buildings and vehicles. It is easier to get a job with a portfolio made up of excellent buildings and vehicles, than it is with a portfolio that also contains not-so-good character models. In short, if you can model/texture something well, leave it in the portfolio. If in doubt, leave it out.
Concept drawings
The example shown here would not look out of place in a fantasy game world. It was based on a variety of Dutch Old Master paintings (a wonderful source of imagery for fantasy characters) and a couple of modern fantasy artist’s work
Most good character modeling starts with a concept drawing. This traditional method of design is still the fastest way to work out the look of the character and may also convey his/her personality and also a colour scheme. Good character design relies upon more than the technical skills required to ‘draw people’. Before drawing a character, you should research their physical appearance, hairstyle, clothing, equipment etc that should fit in with the style of the game.
I like to draw my characters in a pose that relates to their personality, but many artists prefer to draw characters in a neutral pose, which can be helpful if the final model is to be created by another artist. I drew the concept sketch using pencil on paper, scanned this and painted over it in Photoshop and experimented with a range of colours until I found a scheme I liked.
Counting your polygons
The next step onwards is to model the character. When modeling in 3D in your 3D program, you may find that it allows you to replace the background of your viewport with an image. If so, you may find it useful to put a copy of the concept sketch there. If this function is not supported, you could make a box with the same dimensions as the concept scan and apply the scan to the box as a texture. Move that to the rear of your viewport and you can consult your concept without having to look away from your monitor. When modeling a character for a computer game you need to know a few pieces of information that would normally be supplied by the lead artist or a programmer.
- How many polygons do I have to model the character with?
- Is it a seamless skin or a segmented model?
(read the book for further explanation on this).
- How many/how large/what colour depth textures can I use? (again, read the book for further explanation on this).
Each game has it’s own particular variables, but the above points are true of every game. For the example that will be shown here we will make the character a seamless skin 3000 polygon figure with a 512×512 pixel texture in 24 bit colour, with perhaps an additional 256×256 texture for the head and maybe a 64×64 pixel texture with an alpha channel for the hair. These limits would create a model suitable for the Xbox, Gamecube or high end PC. The PSX2 could also handle this amount of detail, but it would require smaller 8 bit textures because it has much less VRAM than the other platforms.
The first step to modeling a character in 3D is to decide which method to work with. There are many different ways of modeling characters, but the easiest way for the beginner is extrusion modeling.
Extrusion Modeling
Extrusion modeling should be available in every 3D modeling package. The basic concept is that you start with a primitive object, usually a box, subdivide it (a process that increases the number of polygons on each side of the box) and then select particular polygons and extrude them out or into the mesh to form general shapes. The extrusions/intrusions are then adjusted and refined to form the details of the model. In many respects, this style of modeling is similar to real world sculptural techniques with clay
To create a character, first start with a box that is approximately the height, width and depth of your character’s torso if he was standing upright. Now set the amount of subdivisions on the front to at least four, on the side to at least two and the top to at least two.
It is important at this stage to use even numbers because we want to ensure that the model has a line of geometry down its center, which will help later.
The legs. Now select the faces that are where the legs would start from at the base of the box. You should be able to select both leg tops and extrude them at the same time. Use a face extrude tool to extrude the legs out to mid thigh. Extrude again to the top of the knee, mid knee, and bottom of the knee. Then extrude to mid calf, ankle and the base of the foot. Select the quad face at the front
of the bottom of the leg and extrude the foot from it, giving it about three segments.
Use the same technique on the faces of the box where the arms should start and extrude the arms to the wrists, again selecting both arm start positions at the same time will make the job quicker and more symmetrical.
Extrusion modeling often works well with a type of polygon called a quad. In basic terms, a quad is two triangles that lie next to each other and share an edge down their center. This edge is often hidden in 3D programs to designate that the two faces can be selected in a single selection (assuming the package has quad selection). They are very useful because they are often rectangular in shape and easy to extrude from and then refine to create more of a curved form.
For the hands, extrude as before up to the end of the palm, allowing two segments to create the palm. This gives you a quad face on the side of the hand to extrude the thumb from. To create the fingers you will need to use a tool to subdivide the end quad of the hand.
This may be called edge or face divide, or cut or slice or a similar sounding term. Divide the edges that make up the top and bottom of this quad until you have four quad segments. You may have to use a turn edges tool to ensure that the quad face is now divided into four square quads. Now select each quad in turn and extrude it to form the segments of each finger. You may wish to move or scale each finger slightly as you extrude them so that they can be selected easily later on an individual basis.
The next task is to extrude the head. Select the faces at the top of the torso where the neck would be and extrude a neck, giving it two segments. Now extrude the head from the neck: it’s hard to estimate at this point how much detail you will
need for the head, so allow around 4 segments to give a basic shape to it.
The reason why I said it was important to have a symmetrical look to the character will now become clear.
The next task is to select half of the body in the front view and delete that half of the body, leaving you with half a torso, one arm, one leg and half a head. Now you have deleted half of the model, use a copy function in your 3D package to copy a ‘reference’ or ‘instance’ version of the half figure. Mirror this model so that it appears to rebuild the other half of the character.
Because the new model is an instance or reference of the original mesh, any modifications you know perform on the original half figure will be replicated automatically on the instance/reference model. This is a huge time saver in creating living creatures, which tend to be symmetrical. Depending on what program you are using, it may be important to remember to only work on the original mesh, as operations performed on the instance reference may not be passed back to the original. To make sure you do not forget which model is which, put some 3D text next to the original mesh saying ‘Original’.
Smoothing your model
The next part of the tutorial is the longest and hardest. Now you are ready to begin to transform the mass of ugly hard polygons into a curved, organic form. Using the mesh you have created you will need to start to refine and adjust the mesh to fit the design that you previously drew. At this point you will discover that the more time you spent on the concept drawing, the less questions you will have about the actual structure of the model.
Each modeling package has its own tools for modifying the surface of a mesh, but I would advise looking out for tessellate or subdivide functions to increase the amount of polygons in an
area of the mesh to give you more vertices to move about. Cut and slice tools will help you break long edges up so that the newly formed vertices can be moved to increase the curvature of the edge.
I would advise working over the entire mesh to increase the level of detail in it to an equal level. This means you should not overly increase the amount of detail in any one area over any other at this stage in the modeling process. The reason for this is it is very easy to become stuck trying to model the perfect face and then discover that you do not have enough polygons left in your polygon budget to refine the rest of the model. This is a common mistake and results in very chunky models with highly modeled and smoothed off faces, a combination which jars the eye.
If your 3D package has a polygon counter, keep it on screen to show you how many polygons you have left to work with. Once you have used 60 – 75% of your polygon budget on rounding the overall form of the model so that it looks good from a distance (it may be a good idea to set up a camera view that sees the whole model at a resolution of 640×480 pixels so that you can tell where you should add more detail, and also where it is not needed), you can use the remaining budget to increase the level of detail of the face of the character, as well as any additional details that define the nature of its personality. This could be more detail In the clothing to show the style of it, or the state it is in. Extra modeling could be put into the hair, especially if the character has a flamboyant hairstyle. Equipment for the character to have about itself may be a good addition: backpacks, weapon holsters, armour etc. A basic rule of thumb is anything that would look strange on the model if it was just represented by a texture on a flat mesh, should be modeled in the detailing part of the modeling process.
Once you are happy with the look of your model, you can join the two halves together and then fix any mesh artifacts that arise from this
process and assign or correct any smoothing groups that are required. You may discover at this point that parts of the mesh are not quite in proportion. It is a good idea to have someone else look at your model for these types of errors because you have been looking at it for so long that your eye will begin to ignore such problems.
Texturing your model
When you have finished the modeling process you must begin the texturing process for the character. As this is a games model we have already decided that all of the body of the character will be textured using one texture, with the head using another texture and the hair using one final texture. Because the body must fit into one texture, you will have to create a texture that looks like a bit of a puzzle, or a mass of severed body parts! If you look at the texture for a games character (Quake3 models are an excellent source as their bitmaps and models are easy to find in the installed game) you will see how the artist has packed every detail possible into every pixel available to him.
This process usually starts by applying different forms of UVW mapping to the parts of the character mesh (for further explanation about different forms of mapping, read the book). Thus an arm is often given cylinder mapping, the palm and back of the hand uses planar mapping etc. Then each part of the mesh that requires a texture is selected in turn and then ‘unwrapped’ to a flat surface which can be saved out as a bitmap containing an image of all the wireframe of the model within it. These wireframe images can be combined in Photoshop to form a collage of all the parts of the model. Pack them as close to each other as possible (although it is often good practice to leave a 1-2 pixel gap around the edges of each body area, as this prevents the pixels from one area of the model from ‘bleeding’ through into another area when the model has bi-linear filtering applied to it in the game engine). Now you can see where you should start to paint your textures.
If you want professional services for your next project, click here.