general Overview

Setup And Preparation

Polygon Modelling Tools

Understanding UV's

Subdivision Surfaces

Texturing And Hypershade

Rendering And Final Output

More Topics:





Part 1: Gereral Overview


In the following lessons we will be examining the workflow of creating a head in MAYA. The purpose of focusing on the workflow is to give you an idea of what a director might expect from a Maya operator when working towards fulfilling the goal of creating an animation within a team of Maya operators. This is the most realistic approach to learning an application such as Maya,

  • First consider the application as a whole, and gain a general understanding of how all the aspects of Maya fit together.
  • Secondly consider how Maya’s workflows can be broken up into smaller segments (such as modeling, texturing, rigging, animation etc).
  • Finally consider which aspects of Maya appeal to you most and specialize in those fields, taking into consideration what specialties are most sought after in the market place you're aiming for.

Starting off with modeling is the most logical first step, because as an animator, texture artist, technician etc you'd first need to understand the principles of how a 3D model is generated from the base up. Understanding this will help you create more predictable geometric deformations (the process of modifying a 3D model, such as bending a character's arm).

The order of this course's workflow follows:

  • Planning & preparation
    Project setup
  • Polygonal modeling
    UV texture mapping
  • Subdivision surface refinement
    Polygonal refinement
  • Texture mapping

I've written these lessons as though we were working within a 3D production environment, where you are given the task of creating a character's head and texture mapping it. Upon completion of your task you would then hand your model, with its textures to a rigging team. This team based on previous discussions with you, as a modeler, will prepare the model for animation, then hand it over to the animation team who will animate it and subsequently hand it over to a compositing and lighting crew. This team will render out the animation to produce the final result. The discussions between the different departments that dictate what your model will ultimately look like are key to producing an efficient model that plugs seamlessly into the various workflows. Planning and preparation will outline the key points that need to be addressed before you can start modeling your character's head.

Planning and Preparation

Before you begin modeling your character's head you need to establish what it will be used for. Because we are modeling a head we can pretty much rest assured that our character will be used for something like a close-up or ECU (extreme close-up) shot. We can pretty much rule out a Wide Shot (or full body shot).

Wide Shot

Close Up

ECU (extreme close-up)

For the purpose of our exercise we'll be modeling a head for a close-up to ECU shot. We need to establish what kind of shot our character is involved in, what kind of lighting will be used in the scene and what the final render will be output as. This information can be sourced from the various Maya operator departments involved with the project.
Regarding the "shot type", we don't want to spend hours modeling a character's lower body when it won't appear in the final shot, so we'll focus on the most visible parts of our character, then thin our workload out towards the not so visible parts of our character. Creating a more efficient workable model that hasn't taken us forever to create.
How the character is lit relates to what sort of colours you'll be using to texture map your character. It wouldn't be advisable to create a bright yellow base-textured character when you're going to be using it in a very moody, dimly light scene. It just wouldn't make sense to the viewer, unless it was your intention to elicit a humourous response!
Finally, in finding out what the final output for the rendered version of the character will be, you'll have a better idea of what size to make your textures maps. 3D models are made up of vectors (vertices, faces, edges) all of which are resolution independent. Meaning that once you've created your character, and assuming you have not applied a bitmap texture to your model, you can zoom in as close as you want on your character, render it, and not see any pixilation. However as soon as you apply a bitmap texture to your character, and most 3D characters, if they do not use procedual textures, do have bitmap textures, the output rendering of your character becomes resolution dependant. Meaning that if you applied a bitmap texture of 256 pixels by 256 pixels to your characters head, but you render your character at 1024px by 1024px, Maya is going to have to stretch (or more acturately interpolate) your characters bitmap texture to four times its original size causing the output rendering to become highly pixilated. Not good!
To avoid this problem find out before hand what your model is going to be used for. Is it going to be used for Cinema, TV, multimedia, the web etc? Each one of these output formats has a specific target resolution, so for example if your final product was for television, in South Africa we use a system called PAL which has an output resolution of 720 x 576(non-square pixels). Meaning when the compositing team renders out a frame of your character they will set Maya's Software/ Vector / Mental Ray render to output a frame to 720px by 576px. It is therefore safe to deduce that, if you had a full body shot of your character, your entire character would not take up anymore than a maximum of 720 pixels in width by 576 pixels in height; so your texture map, for your entire character in this shot, should not be much higher than this given resolution. If you had a close-up shot of your character's head for the same output (PAL) then the texture map for your characters head alone, would not exceed much more than 720px x 576px bearing in mind that in the previous example your entire character utilized the same sized bitmap for its entire body. This also means that one should never feel limited to using one texture map on a character or character component. Texture maps should be created for the character on a per shot basis, this also means that the model should be versitile enough to reflect as wide a range of differing resolutions of texture maps as possible. A table follows of suggested bitmap resolutions for certain outputs on a per shot basis (remember that one character is often made up of several texture maps).

HD 1920 x 1080 16:9 2048 x 2048
PAL TV 720 x 576 4:3 1024 x 1024

Multimedia and Web outputs cannot be standardized at present as they are dependant on the end users computer capabilities. But a good starting point to create a texture for this platform would yield a resolution of about 512 x 512 per characters' texture map, this suggestion is based on a maximum compatibility standard output res of 640 x 480

Bear in mind also that although your character might be used in a wide shot, your character might not take up more than one tenth of the full frame size. This situation is particularly common when the focus of the shot is on the environment that the character is in, or if there is a crowd of several hundreds/thousands of characters. in this case the entire texture resolution of your character need not be larger than the percentage of the region that you character encompasses. for example if your character only takes up 10 percent of the entire frame shot at HD. then the entire resolution of your characters bitmap need not be larger than 256px * 256px. to arrive at this figure, 10 percent of 2048 (which is the suggested resolution for HD) is 204. however computers perfer working in figures divisible by 2, 4, 8, 16 etc. and it's generally a better idea to round up than down, so the next logical number would be 256.

PAL has an almost squarish shaped frame, like your TV

HD has a much wider shaped frame, like at the cinema

There are no hard and fast rules to creating the "proper" sized texture map only certain guidelines one can follow. Firstly make sure your texture map is to the power of 2. 3D applications tend to handle even aspect ratios more efficiently. Create multiple texture maps, one for your characters to be used in close-up shots. Another for your entire character to be used in full body shots and finally a high resolution texture map at least twice the suggested resolution for your character to be used in any ECU shots. Use the intended output resolution to determine a starting point of how big to make your texture maps. Not all maps need to be 2048px x 2048px; this is not an efficient solution as it will dramatically slow down the rendering process.

This character renders perfectly for a Close-up or Wide shot...

However when we zoom into the region between the eyes and the snout and re-render the image we can see some unpredicted pixelation. Clearly this characters texture map was made for a wide to close-up shot.

Once you have determined the criteria for creating your character's texture map, you will need to consider how your character is going to be animated. The best thing to do would be to obtain a storyboard of the animation from your director, and speak to the rigging and animation crew about what areas of your characters head will be manipulated most extensively in animation. In obtaining this information you will be able to decide what needs to be modeled on your character's head and to what degree of realism this needs to be achieved. Also what areas on your characters head can be faked with a texture instead of actual geometric modeling? An efficient model will not have every pore on its skin, every wrinkle around eyes and every blemish modeled, this is overkill! It would be near impossible for the other departments to work effeciently with a model having such a high polygon count (geometric resolution). Consider when you are working with Maya that you are working towards a goal not necessarily on that goal. As every product that comes from Maya has to go through some refinement process before it gets to the end user, for example a composting application to combine all the different renderings into a single frame for TV, cinema, print etc, or even a games engine to optimize and make the animations you created interactive, a web graphics optimization toolkit to reduce the size of the image with better compression algorithms for an internet browser.

Some areas of a character need to modeled, particularly if those areas are going to appear in ECU shots, or if they are going to be deformed through animation.

Other areas of a character should be assimilated using appropriate texture mapping and lighting.

Finally you will need to assess whether this character has any special systems interacting with it, such as particles, hair, fluid or fur. Knowing this information will help you assess what kind of geometry your final model will be output as. Not all Maya’s geometry types are compatible with some of Maya’s dynamic systems. Maya has three types of geometry NURBS (non-uniform rational B-splines), Polygons and Subdivision Surfaces. Of these three types polygons is the oldest technology (and therefore most compatible across workflows and platforms) subDivs is the latest technology but least compatible. Each technology has its own fields of excellence, for example nurbs for their ease of use and precision. Polygons for its versatility and highly interactive feedback make it a great choice for experimentation through to final output, and finally SubDivs have a specialized toolkit for organic modeling that makes it the best choice for high definition quick and effecient character modeling. So the question is which one to use? In actual fact you do not have to limit yourself to one geometric type as Maya allows you to convert between all three geometric types. This allows you to choose the aspects of each geometric modeling environment that suite your workflow, and benefit from the best aspects of each geometric type. However be aware that the conversion process between each of these different geometries does not happen without a little tweaking. The conversions between the different geometries can be as accurate as you want them to be but, for example, the more accurately you try to match your subDiv surface to your polygon output surface the more complex your model becomes and the more processing power you need to work with the higher resolution model. It’s ultimately, a balancing act as with most progressive technologies in 3D. It is also not advisable to try to create seamless blends between different geometry types, as these seams are prone to becoming evident in the animation process. However, for example, if you were to output your characters head to a single geometry type such as subDivs but he's wearing a top hat, there is no harm in constructing the hat entirely out of NURBS and his eyes out of polygons, if this yields the desired results.

Each Geometry type has special tools that enhance the best features of that type, and allow you to create very specific, distinguishable features in your characters.

Now that we have all the technical specifications of our character down we can start to sketch out perspective and orthographic projections of our characters head. If you have come this far in the process you would have already established what the character is going to look like. Sketch out a few views of the character. They don't have to be master pieces, but rather something that will convey an atmosphere or presence about the character. Then proceed to sketch orthographic projections of your character. Often a front and profile view will suffice but if the character has specific details on his crown you might need to sketch a top view as well.
Make sure your character's proportions align correctly in the different orthographic projections, use a TD (technical drawing) board with a T-square if you have one. If not you'll have to align the proportions in Photoshop. Scan the images in at a low res no more than 100dpi, save the images to your projects directory (see following section). The images will only be used as a reference for modeling and not be evident in the final output. Modify the images in Photoshop so that their dimensions are perfect squares and make sure that the subject matter is positioned correctly so that when the images are aligned if you were to draw a horizontal line extending across one images eyebrow, eyes, nose. Mouth etc the line would intersect the corresponding point on the adjacent image. You’re almost ready to start modeling.

Part 2: Setup And Preperation