2-660-715-02(1)

http://www.sony.co.jp/support/pc-acc/

USB SMU-WC3

SMU-WCRC120

USBtPS/2

SMU-WC3

Microsoft Corporation

Windows

Microsoft

2005 Sony Corporation Printed in China
Apple, Macintosh, Mac, MacOS Apple Computer, Inc.
Pentium Intel Corporaiton Power PC IBM Corporation
USB PS/2 USB USB PS/counts/inch 27 MHz 1 ch 8190ID 500 V/m 1m 3 - USB A 2 PS/2
Windows XP Media Center Edition 2005 Windows XP Media Center Edition 2004 Windows XP Professional Windows XP Home Edition Windows 2000 Professional Windows Millennium Edition Windows 98 Second Edition Mac OS 9.1 OS OS OS

SET POWER

USB SMU-WC3 T11-1001A-3 OS
- USB A - IBM PC/AT CPU Pentium - Apple CPU Power PC G3/G4/G5 350MHz - Windows XP Media Center Edition 2005 Windows XP Media Center Edition 2004 Windows XP Professional Windows XP Home Edition Windows 2000 Professional Windows Millennium Edition Windows 98 Second Edition - Mac OS 9.1 Mac OS 9.x 1 OS OS

LED ON/OFF

64.32 12.5 99.7 mm 87.8 mm

USB USB USB

USB USB
115 g 14.7 g 3.USBtPS/8 10% POWER ON/OFF ON

POWER ON/OFF

ON POWER OFF

USB PS/2

OS OS

USB USB USB

USB 1 SMU-S1 SMU-C1 2

SMU-S1 SMU-C1

SET 1 1
USB USB USB USB 5 USB USB OS SONY

USBtPS/2 USB PS/2

POWER ON/OFF LED ON

USB USB USB USB 5 USB

OFF USBtPS/2 PS/2

USB USB USB USB 5

BIOS USB OFF ON 30 USB Enabled USB BIOS Disabled
POWER ON/OFF POWER ON/OFF
OFF USBtPS/2 PS/2 USBtPS/2

1 PS/2 PS/2 PS/2 PS/2

POWER ON/OFF POWER ON/OFF

POWER ON/OFF OFF

Macintosh
Mac OS 9.1 USB Mac OS 9.x 1

Mac OS 9.x 1

in the definition of a module that gives modular assets their utility and flexibility is frequently one that is interchangeable with others. By having alternate modules that may be substituted for an original we open up the possibility of producing a new asset at the cost of only making a new module. With a sufficient number of modules, this can lend itself to nearly limitless possibilities. A modular asset then can be more accurately thought of as a large group of modules which can be configured to produce a final asset. The final asset itself does not use all of the constituent modules of the entire group, but rather only a small selection. Also, the total number of possible final assets (based on the number of possible configurations) is quite large when compared to the number of constituent modules. As such, a modular asset does not replace a single asset in video game development, but instead an entire group or collection of related assets. As has already been discussed, one of the largest and most obvious benefits of using modular assets in place of more traditional non-modular assets is the sheer volume of final assets that may be produced from the relatively small amount of work necessary to produce the constituent modules of that modular asset. However, there are other benefits as well, and these benefits form the majority of the reasons modular assets have been used in the past. Probably the single largest contributing factor to the use of modular assets is the concept of player customization. If a large number of modules are produced and then the player of the game is given access to and control over the configuration of these modules he or she then gains some control over the appearance of a visual asset within the video game, making it unique and customized to that player. While this has been used across a
variety of genres, it has been traditionally used by Role-Playing Games (RPGs) and is currently seeing its largest growth of usage in Massively Multiplayer Role-Playing Games (MMORPGs). These games in particular benefit greatly from the visual variety afforded to them by using modular assets for player-character avatars and their equipment, either through the sheer volume of combinations (as is the case in more traditional single-player RPGs), the customization afforded to the player, or the differentiation possible between final assets (as is the case with MMORPGs in which it is very helpful to differentiate different player-character avatars, allowing them to identify each other visually, rather than with names or text). However, visual variation is not the sole reason modular assets have been used, or remain useful. While many games give players the opportunity to customize their characters or other assets on a purely visual level, often this customization runs much deeper, allowing the player to customize the gameplay to some degree as well. Oftentimes a single game may afford both types of customization. In the case of the popular MMORPG World of Warcraft players are given the opportunity to customize their avatars face and hairstyle which have absolutely no impact on gameplay whatsoever. However, players may also customize their characters clothing and weaponry, each of which carry a significant impact on gameplay in addition to the appearance of the avatar. This application helps give players visual feedback to their gameplay choices in addition to their control over the appearance of customizable assets. Despite their many benefits, modular assets are not equally suited to every application or video game. A certain set of criteria must be met for a modular asset to be advantageous over normal single asset production. The two player benefits mentioned

1.2.1 Catalogue of Visual Styles
With the large number of MMOs currently available in the market today, it is difficult to find a way to categorize even their content, much less the visual style of this large library of games. However, to simplify the process, I will discuss the visual styles of a few large categories of games, grouped roughly chronologically as they relate to a modern day setting. First, I will look at some of the MMOs set in future together as a scifi genre and discuss some of the differences and similarities between these games visual styles. Next, games set in the modern time frame will be examined. Finally, games which share a pre-modern setting, from actual historical settings to more fantasy inspired ones will be discussed. One of the more popular visual styles in video games in general is that of sci-fi, or a futuristic look. Several MMOs which take place in a sci-fi setting could better be described as earthy sci-fi. This style is characterized by a very realistic style, where everything has a lived-in appearance, caked in grime and mud, with very few assets appearing shiny or brand-new. This carries over to character styles, most notably in terms of character races. Races present in games of this visual style are widely varied and imaginative, often departing significantly from the base human model in both proportion and visual make-up. Color scheme is also an important aspect in this style, with colors tending towards warmer reds and browns in general, and drawing particular attention to colors which break with the general scheme, such as blues and purples. Star Wars Galaxiesi is a particularly good example of this visual style, as is, to a lesser degree, Anarchy Onlineii. On the other end of the sci-fi spectrum in terms of visual style is that of Eve Onlineiii. Whereas Star Wars Galaxies has an earthy and varied visual style, what could
be termed a clean sci-fi style is largely uniform and stereotypically futuristic. This visual style is characterized by a much more mechanical approach, with organic objects appearing rarely in any assets. The color scheme tends toward a cooler spectrum of dull purples and intense blues. Even when other colors are present, they are often cooled, giving everything somewhat of a sterilized look. This is particularly well presented by Eve Online as nearly all gameplay takes place in the vacuum of space, with other characters rarely being seen in person, save for small communiqu icons. Although technically set in the future, The Matrix Online takes place in a facsimile of the modern world, making its visual style more modern than sci-fi. However, with that in mind, The Matrix Onlines visual style has much in common with the clean sci-fi look, and can probably be best summed up in a single word: sleek. All character visual assets look brand-new and un-used, often too perfect or idealized to even exist in the real-world, even on a store shelf. Character clothing tends toward leather and vinyl, with the one major unifying factor being that nearly all assets are slick and shiny. Drawing from the visual style of the movies on which it is based, the color scheme is startlingly uniform green, further driving home the point that the world in which the game takes place is an idealized, too-perfect version of our own worldiv. Like The Matrix Online, City of Heroesv and City of Villainsvi are similarly set in a world loosely based on modern reality. However, while The Matrix Online draws on a specific series of movies for its visual vocabulary, the Cities games draw upon a set of stereotypes that have coalesced to form the comic book genre. At first glance this style seems largely chaotic and un-uniform, but upon closer inspection a few generalities can be gleaned which define it as a cohesive style. The most cohesive element is also that

which makes this style so tumultuous: exaggeration. This applies to nearly every aspect of character visuals, from proportions to color scheme. Any individual will be cohesive when considered by itself within this element of exaggeration. That is to say, if a character is meant to look strong or super heroic, his proportions will be exaggerated in accordance, with much more emphasis placed on the upper body musculature, broad shoulders, and commonly a large chin. However, when considered next to characters exaggerated in a different manner, dissimilarities quickly become apparent. Color scheme follows in the same patter, in that any given character will likely have a cohesive color scheme individually, only to clash wildly with the color scheme of other characters. However, almost uniformly, colors are very bright and very highly saturated. While not technically a game, Second Lifevii deserves mention and examination along with other semi-modern set MMOs as it shares, and in some cases carries to an extreme, the importance of character customization and individuation. However, unlike nearly every other MMO currently commercially available, the majority of visual content present in Second Life is user made and contributed. As such, defining a cohesive visual style is even more challenging than with the Cities games, though one is still present. Default character avatars, being one of the few developer-made assets, tend to follow a set rule of simplification, tending toward a cartoon feel with over-large eyes, mouths, heads, and hands while arms and legs are often represented as nothing more than simple jointed tubes. User-contributed assets tend toward a general scheme of cartoon-like simplification, largely attributable to the fact that the modeling tools included with the program are best adapted to producing simple shapes. Beyond these generalities,
however, there is next to no unifying factor among visual assets, though near limitless flexibility and customization with respect to character avatars. Beyond modern and sci-fi genres, a large percentage of MMOs tend towards premodern and fantasy visual motifs. Within this genre, a style rapidly growing in popularity is one easily recognizable in games such as Guild Warsviii , Lineage 2ix, and Rappelzx. Though semi-realistic, this style also has a healthy amount of idealization and differs from standard human proportions in some significant ways. In general, figures are slim and elongated, while features alternate sharply between smooth roundness and slightly subdued angularity. Of particular note in this visual style is the attention to detail. There are very few broad expanses of a flat color or texture. Instead, in an almost horror vacuui manner, any available areas are filled with small details and decoration, giving this style a signature look of high contrast in small areas. In nearly direct contrast to this style is that of one of the most popular MMOs of all time: World of Warcraftxi. Whereas the majority of other MMOs tend to stick pretty closely to real human proportions, World of Warcraft has its own unique set of proportions which, when they stray near realistic ones, do so completely by coincidence. In general, this visual style tends toward big a bulky, applying equally to figure proportions as well as costumes, weapons, and all visual assets. Most assets also tend towards a bottom-heavy hourglass shape, being narrower at the top than the bottom and thinnest in the middle. Like a cartoon or comic book, colors tend to be highly saturated, although not to the same degree as City of Heroes. Also, in direct contrast to games like Guild Wars or Lineage 2, World of Warcraft has few areas of high detail or extremely

high contrast. Details, like all other elements, tend to be big and bulky, and as such are spaced far apart, allowing for large areas of relatively flat texture. As much as World of Warcraft approaches a cartoon-like visual style, there are other games which have given over entirely to a cartoon look. Notable among these are Maple Storyxii and Ragnarok Onlinexiii. Despite the fact that both of these games are sprite-based 2-D games (which will be discussed in depth later), they follow a simplification of character to the most basic essentials. Heads become spheres or balls, fingers are an unnecessary detail, eyes are enlarged to take up half of the head, mouths are non-existent unless open. Colors and detail are likewise simplified, with most elements being composed of a few simply defined shapes of broad, flat color with little to no variation whatsoever. In addition to these three broad categories of sci-fi, modern, and fantasy, a number of games do not fall neatly into a single profile, though they often share much in common with games of these genres. Puzzle Piratesxiv, for instance, is not fantasy in nature, but is every bit as much a cartoon style as Ragnarok Online or Maple Story, possibly even more so as characters are more reminiscent of Playmobil figurines than actual people. Toontownxv draws on a series of conventions in a manner similar to City of Heroes. However, instead of comic books, Toontown looks to color cartoons from the 50s and 60s, particularly those of Warner Bros. and Disney. The visual style of MMOs does not exist in a vacuum and is heavily influenced and informed by a number of other factors. As technology plays a pivotal role in the look of a game, the target platform and overall rendering capabilities play an enormous role in defining the visual style of a game. Games which are 2-D, such as side-scrolling MMOs
like Maple Story, must by nature be much more simplified and cartoon-like. Isometric or 2-D games like Ragnarok or Westward Journeyxvi can allow for greater detail in characters or even blend 2-D and 3-D elements into a nearly seamless whole. Sidescrolling and isometric games, despite their limitations, allow for greater control of a unified visual style as they not only have complete control of what a player sees, but also of what viewing angles content will be seen from. While 3-D games allow for much more viewing freedom from the player perspective, there is still a great deal of variation in what a render engine can support and what the target hardware is capable of. On one end of the spectrum are games like World of Warcraft, targeted at machines with very low render capabilities. These games by necessity tend toward blockier forms and more simple textures with minimal detail or contrast from lighting. One the other end of the spectrum are the so-called next-gen games such as Lord of the Rings Onlinexvii or the anticipated Warhammer Onlinexviii. As these games are targeted at hardware capable of much more impressive visual feats, they can choose to include a wide variety of advanced render techniques, such as specular, normal, and sub-surface mapping. This allows for the potential of a much more rich and detailed visual environment, but can also overload the player with a high-contrast and constantly varying playing environment.

1.2.2 Parameters of Customization In a discussion of MMO character and avatar, an element rivaling visual style is character customization, and the parameters/ mechanics by which that is defined. Every MMO must have some way to vary the appearance of different character, if only out of a
necessity to differentiate between different players in the virtual environment. However, considering the degree to which this may logically affect gameplay, many games carry this to an extreme, driven primarily out of player desire. An overarching concept among all of the different aspects of a character which a player may customize is the manner in which they may customize it. As exhibited in games thus far, there are two major approaches to this question, with a potential third only beginning to form on the fringes. The most direct solution is to provide a player with a limited set of choices, which we may refer to as the component method. While this may apply at different levels, it rarely occurs in MMOs on a macro level. That is, often a character avatars face will have components of facial features, hair styles, and coloration to choose from, with a limited set of predetermined choices in each component. The second approach to the problem is the slider method. Instead of having a limited set of predetermined choices, a given feature for customization will be broken down into a set of parameters with minimum and maximum values defined for each parameter which can then be individually customized by the player. This is most often seen as a set of facial sliders which allow players to individually alter things such as mouth size, face width, distance between eyes, brow height, and so on. However, Second Life is notable in that nearly every conceivable parameter of character appearance, from overall build to hairstyle is broken into such a set of parameters, easily numbering in the hundreds, allowing for nearly infinite customization. The third solution, which is only beginning to appear and may never fully develop as a workable alternative, is that of player-driven content. This allows players with sufficient knowledge and skill to develop and upload their own custom assets. This

in nature, and have no impact whatsoever on how the game is played. Maple Story and Puzzle Pirates are excellent examples of the latter. Another interesting aspect of character appearance customization, especially as exemplified by both Maple Story and Puzzle Pirates, is the impact of character appearance on business model and economic considerations. Whereas customization options may have little effect on gameplay, they may sometimes form a cornerstone of the economic model of a given MMO. In Maple Story, for instance, players may purchase virtual items which affect and alter the appearance of their character directly from the developer for real money. Likewise, on some servers of Puzzle Pirates, nearly any item a player purchase to affect their characters appearance has a cost in real dollars, or some derivative thereof. Games such as these usual depart entirely from the subscription model of other MMOs, relying instead on these micro-transactions to generate revenue for the developer.
1.2.3 Summary Whereas avatar appearance and customization may have once been a minor consideration in video games, Massive Multiplayer Online video games have quickly advanced it to being a critical element of many video games. Characters form a critical part of a video games overall visual style. Additionally, many MMOs have myriad options available for a player to make their avatar unique and separate from the masses of other players within the video games virtual environment.

1.3 Proposal

1.3.1 Overview For my Masters project, I will concept, model, and texture two modular, customizable game assets: a character and a vehicle. To further define modular and customizable, each of these assets will be created from an amalgam of a series of individual pieces and assets. In the case of the character these assets would be elements of clothing, armor, and other pieces worn or carried. In the case of the vehicle, the breakdown is more along of the lines of the function, appearance, and purpose of the vehicle. Some examples modules of the vehicle would be chassis size and style, seating solution, method of locomotion, and so forth. The notion of customization is fairly wellcovered by the modular approach to asset creation, but beyond this individual textures will be customizable as well, in a modular fashion. Customizable textures includes such things as the material an asset is made from (independent of its shape), surface color decoration, and small details such as wear, dirt, scratches, and grunge.
1.3.2 Art Style The art style will take full use of current technology, using next-gen approximate polygon counts (5-8 thousand for a character other props scaled accordingly) as well as taking full advantage of the variety of mapping techniques available, most notably diffuse, specularity, opacity, and normal maps. In an effort to use these techniques to their fullest potential, the proposed art style could be described as worn. While not necessarily strict photo-realism, every object and model will be considered as the culmination of a story and include appropriate details to match, as opposed to producing perfect or pristine objects that look brand-new, shiny, and without imperfections.

only a single method of attachment, intending to discover new methods as I advanced on to the next stage of the project. I executed all of these concepts in Photoshop using a new layer for each module with the base chassis as the bottom-most layer. This allowed me at a very early stage to begin experimenting with different combinations of modules to see how they would affect the overall silhouette of the vehicle and also to determine which modules would fit together easily, which would require some additional design work to fit with other modules, and which modules would not work together at all. While it was very advantageous at this stage to begin to see some of the impact of specific design decisions, it was also limited by both my drawing skill in perspective and by the fact that these modules were only drawn from a single perspective.

2.2 Modeling

From this early concepting phase I prepared to model out each module in 3d Studio Max 8. I decided to produce each model in a separate max file and use the xreference tool in 3DS Max to bring all of the assets together in a single file to experiment with combinations and view the finished asset. The X-reference tool was particularly useful in that I could reference an object into the final scene before it was even finished and, provided I never changed the name of the asset, any changes I made in the modules specific scene file would be updated into any file I referenced the object into. This allowed me to see all of the modules together at a very early stage and to continue to see the effects of a change on a single model across the entire asset.
As I began modeling modules, I quickly ran into trouble while trying to determine a target polygon count to aim for while producing each module. At this point I still had not determined a target output level (what game engine or overall graphics level) my asset would be targeted for. Additionally, while I had a rough idea of what a single assets target polygon count might be for a given engine, being that I was producing a modular asset, I did not have complete control over the polygon count as this would be subject to the final configuration and so, at best, I could limit the polygon count to a given range. While a decided that my target graphics level would be next-generation in general using the Unreal 3 engine as a specific target, I decided to err on the side of caution and keep my polygon counts for each module relatively low, falling in a range of 200-700 for any given asset, varying by complexity. To ensure all of the modules fit together properly, I decided to model them as I had concepted them: from the inside out. I modeled the chassis first and saved it as its own scene file. I then modeled each asset that was intended to attach directly to the chassis inside the same scene file, directly on top of the chassis. Before I saved these additional scene files (with new names, so as not to over-write my original chassis model) I deleted the chassis models and moved the new model to the center of the scene file. This process continued out along the prescribed chains in such a way that I would begin modeling a given asset within the scene file of the module it was intended to attach to, delete the original asset, and save the scene file under a new name. After I had finished modeling each module I created a new scene file and xreference each of them into it. Once the entire asset was assembled and could be configured into all its possible variations (accomplished through placing each module in

lose the consistent grain I had among all the textures, though moving into my next phase I realized this was less important than I had originally thought. With all of the problems that had arisen from my chosen method of texturing, I decided to abandon the idea of a modular texture and instead focus on high-quality nonmodular textures for each of the models, thus simplifying the process. I first revisited my unwraps and organized all of the pieces into the 0-1 texture space. Where possible I preserved the initial grain as I knew this would speed up the process of producing the final textures (I was planning on using my current set of textures as a base), though in some cases this was impossible if I wanted to use the space efficiently (which I did). After this I set about producing the final textures for each of the models. This phase turned out to be the longest span in the project and was much more difficult than I had initially anticipated. I forewent the idea of producing alternate textures for each model and limited each to a single material type; whichever I found to be the most appropriate. By and large this turned out to be the metal material set though cloth, leather, and rubber were also used (stone and wood were entirely scrapped). My process changed as I continued producing textures which necessitated revisiting some of my earlier textures to bring them up to the same levels as my later textures, but eventually I used the same process on all of them. I gathered a large set of reference textures taken from relevant objects, mostly mechanical or vehicular, though there was a large variety, and adapted these to fit onto each texture, thus providing detail and interest where before there had been only flat space. I also applied a series of grunge maps both above and beneath these detail layers and tailored to bring out the highlight of the modules purpose and details. Once the base color map was complete I used it to produce a normal map
(within Photoshop using the NVIDIA normal map plug-in) and specular level and glossiness maps as well. Having all of the assets in a single file proved invaluable at this stage as it let me see not only how the texture was progressing, but to ensure that the textures remained relatively uniform and appeared to belong together when they were placed on the final asset. However, I noticed some problems as I had before as I neared completion of this stage. Probably the biggest flaw I realized was that my concepts were not sufficient for the types of models I was trying to produce, which created far more work at this stage than may have been necessary. While I knew the function of each module from the concept, the actual shape was left very rough with little attention to detail and function (beyond the concerns of the modularity). As such instead of merely producing textures, I spent more than half of my time in this phase still designing as I determined what details to add to the model and where to add them. Also, while I could add details and embellish the models at this stage, I was limited to the shape I already had without revisiting the earlier modeling stage which I was reluctant to do as it would also necessitate reunwrapping the model and throwing out any work I had already produced on the texture. In order to achieve the kind of detail I was looking for, I also found it necessary to work at a relatively high resolution for each texture, commonly 1024 x 1024, though some were higher still. The problem of this was two-fold. First, the amount of disk space and memory this asset took up increased dramatically, something I noticed quickly when I tried to open the final max scene file. Secondly, not every piece required a texture of this resolution. However, this is something I couldnt ascertain from the texture or the model alone. To determine the proper resolution I would have to texture all of the

modules and compare the resolutions between them to determine which ones need that high of a resolution and which ones did not. Once all of the textures were produced and set at their proper relative resolutions I realized I still had one significant flaw remaining in the asset. While the textures had been produced at a very high level, some of the initial models had not. While I was reluctant to altering any of the models a great deal (thus necessitating reworking the textures as well), some of the models were simply too square and blocky and could be improved upon without negating the texture. Once these assets were brought up to a satisfactory level, I was finally finished with the first pass at producing a modular asset.

3. Results and Analysis

3.1 Final Output
The primary objective of this project was to produce a modular vehicle composed of a series of modules which can then be configured into a far greater number of final assets. I have modeled and textured a total of 22 different modules which can be classified into 7 different groupings. In all, there is an estimable 1232 possible configurations of these assets. However, because several of these configurations may be invalidated upon comparison to the optimal final asset (several configurations may have overlapping modules, or not satisfy all of the necessary requirements to qualify as a vehicle), the actual estimated number of valid configurations is roughly one-quarter of this, or 308 configurations. Using these figures as a starting point, it is easy to mathematically calculate the benefit of using a modular approach to producing video game assets. In the time it has
taken me (a single developer) to produce 22 different actual assets, I have produced a potential of 308 assets, increasing my productivity by a factor of 14. Even though the figure of 308 is itself only a quarter of the total number of potential configurations, if we further consider that potentially only twenty percent of the valid configurations are truly different enough to be considered different assets for our purposes, the productivity of a single developer has still nearly tripled. From this point of view, the validity of modular assets is clearly a huge success. Designing assets in a modular fashion has the capability to increase productivity by a bare minimum of 2000%.

3.2 Analysis

Several aspects of this project have proven to be huge successes when compared with my original goals for the project. Chief among these, as previously discussed, is the proven validity of using a modular approach when designing assets for a video game, as mathematically the benefits are readily apparent. Additionally, the widely varied appearance and profile of one configuration to the next I consider to be a personal success, as many modular assets I have seen undertaken in the past have, in my opinion, failed to achieve the possible amount of variety inherent in their nature. Related to this is the large scope of the modular vehicle I undertook and succeeded at. Traditionally most modular assets have been very narrow in scope. In my instance a more traditional approach would have been to narrow the scope of possible configurations to a single type of vehicle, say a motorcycle or airplane. However, I have succeeded in making a modular asset which can be configured into a wide variety of types of vehicles, including cars (4-

uniform, being almost precisely the same shape and size. This offers the advantage of making overlap explicitly clear in the design. A given piece of suspension can have other modules attached to it, but only if those other modules are wheels. Likewise, if one wheel is being used on a piece of suspension, another wheel may not be used on the same piece as they occupy the exact same Point of Contact. This approach greatly clarifies confusion in the structure, but also becomes more rigid and less flexible, leading to a potential decrease in overall possible variety. Also, identifying Points of Contact allows related modules to be grouped together. Modules which all share a common Point of Contact and are mutually exclusive (if one of these modules is used, any of the others may not be) can be thought of as a Module Group. The notion of a Module Group helps to organize and clarify structure and design, especially when it comes time to produce a Modular Map.

4.3.2 Modular Flow

Related to the idea of understanding precisely where two modules attach to each and how is the notion of understanding what order they connect in. To clarify, in every modular asset there is always one inner-most, core module. This module itself does not attach to any other module, but rather one or more modules may attach to it. At first glance this concept may seem purely semantic in nature, but if the structure of a modular asset gets to be sufficiently complex, this becomes crucial to establish. Once the Modular Flow of a given asset is known when can then break down the generic Points of Contact into two separate classes: Inputs and Outputs. Inputs are receiving points of contact in that they allow other modules to attach to the module they belong to. In the case of the wheel/suspension example, the Point of Contact on the
suspension would be an Input. Similarly, Outputs are exactly the opposite; they allow the module they are on to be attached to other modules. Remaining with the wheel/suspension example, the Point of Contact on the wheel would be an Output. These terms may also seem arbitrary depending on your vantage point of the asset, but once we proceed to producing a complete Modular Map, they immediately become selfexplanatory.

4.3.3Types of Modules

After all of the Points of Contact have been identified and categorized as either Inputs or Outputs, it is then possible to begin categorizing modules themselves based on these characteristics. In examining the underlying structure of my own modular asset as well as examining several other assets, I have identified five basic types of modules, though not every modular asset will have all five present in its design. These basic types are Root, Sub-Root, Link, Span, and Terminator. 4.3.3.1Root A Root module is one whose Points of Contact are composed solely of Inputs. Root modules always form the core of a modular asset and are the inner-most, core modules previously referred to. They never attach to any other modules but often have several Points of Contact allowing for many different modules to be attached to them. The chassis is the Root module of the modular vehicle I designed. 4.3.3.2 Sub-Root A Sub-Root module is a module with a single Output, but more than one Input. In other words, this module can attach to another module and can have more than one module attached to it. When I later discuss Modular Mapping, these modules will be

points on the map past the root at which the Modular Flow forks outwards. An example of this in the modular vehicle would be that of the Front Mount which only connects to the chassis, but has two Input Points of Contact where wheels or weapons may be attached. 4.3.3.3 Link A Link module is one which has only a single Input and a Single Output. These modules can connect to one other module, and likewise can have only a single module attached to them. Despite this strict definition, these modules in no way limit the possible variety of an asset, as there can be more then one Link for a given Input on a module, and there can be multiple modules which can attach to its single Input. An example of a Link module would be that of the Rear Mount, which can only connect to the chassis and only has one Point of Contact (which may be used by either the Ballista or Blimp modules). 4.3.3.4 Span Span modules are unique in that they are the only type of module which has multiple Outputs. That is, Span modules connect to more than one other module simultaneously. These modules can be quite visually interesting, but run the risk of reducing overall variety as they require two different modules to be in place before they can be used, limiting the number of possible configurations. If a Span module has no Inputs this is not a great risk, but if it does then the usage of any module which extends from the Span is dependent upon both (or all) of the requisite modules that the Span connects to.
Span modules can also overlap with other types of modules. That is, there can be a Span Sub-Root module that has multiple Inputs and multiple Outputs. There can also be a Span Terminator that has multiple Outputs and no Inputs. 4.3.3.5 Terminator Terminator modules are modules which designate the end of a chain or branch on a Modular Map; they have no Inputs whatsoever. Once a Terminator module is attached to the configuration, no more modules may be connected onto it, though other modules may still be attached on open Inputs on other modules. Terminator modules are not strictly necessary in design as any Input can simply be left open in a given configuration, but they are easier to design as there are few Points of Contact to worry about accounting for.
4.3.4 Level of Modularity
The concept of Level of Modularity is one that applies to an entire modular asset. As with many of the other concepts, this one is easier to visualize in a Modular Map, but in short it signifies the possible length of a chain of modules, or a range thereof. That is, the number of how many modules can be attached one to the other in a continuous chain. This is commonly best expressed as a range of minimum and maximum values, where the minimum value designates how many modules must be strung together for a given configuration to be considered complete and the maximum designates how many modules it is possible to string together (either before requiring the use of a Terminator module, overlap between placed modules, or any other reason). As an example, a modular character asset which consisted of a root asset (comprised of the physical body of the character) and a series of Terminator modules

modules plus Ammo Wheel and Engine), and the Front group (Front Wheel Cover and Front Weapons Mount). Additionally, there are three other Module Groups, though they become somewhat more difficult to identify and do not all occupy the same Level of Modularity. These are the Weapons group (Side Mount Gun and Energy Rifle), the Wheel group (Wagon Wheel, Big Wheel, and Wheel Cover), and the Rear Mount group (Ballista and Blimp). Both the Weapons and Rear Mount groups are easy to identify and occupy the second Level of Modularity as they can only connect to Level one modules, but the Wheel group is a special case. Both the Big Wheel and Wagon Wheel can connect to Level one modules from either the Locomotion or Front groups, but are not wholly mutually exclusive as the Wagon Wheel can connect to some modules which the Big Wheel cannot and vice versa. The situation is even further complicated with the inclusion of the Wheel Cover. The Wheel cover can only connect to modules which the Big Wheel can also connect to, but cannot connect to all of them, and the Big Wheel can then in turn connect to the Wheel Cover. This means that the Wheel Cover is entirely a Level two module whereas the Big Wheel can alternately be a Level two or three module in terms of the Levels of Modularity. While this complexity is reflected in the Modular Map it is also the single most complicated portion of the map.

4.4 Granularity

A final concept that is helpful when designing a modular asset or refining a preexisting modular asset design is that of Granularity. This concept is most applicable when using the Outside-In method of asset design as it directly concerns the placement of Points of Contact and where to separate single modules into multiple modules or how to combine multiple modules into a single module. In short, Granularity is a comparison of
the relative complexity of different modules. This is particularly important in relation to the impact any given module will have on the appearance and profile of a final configuration. Using the modular vehicle as an example, there is a high variance in Granularity from one asset to another. For instance, the weapons group (Side Mount Gun and Energy Rifle) are each highly complex models with high polygon counts and very unique, complex profiles. By contrast, the Chassis is a very simple module with a plain, blocky profile and very low polygon count. There are also modules which fall between these two extremes, such as any of the suspension modules which are each more complex than the Chassis module yet simpler than the weapons modules. While this separation in the complexity of modules is not necessarily undesirable, it is something to be aware of. If all of the component modules are near the same Granularity it is easier to predict the impact a single module will have on the overall appearance and profile of a final configuration. When determining appropriate levels of Granularity, there is one factor to consider: If there is any benefit to the modular side of things to be gained by breaking a single module into several modules. That is, will there be alternate modules that can be substituted for either (or any) of the resulting component modules, or if any of those component modules can be used elsewhere in place of existing modules. If neither of these is true, there is little to be gained in dividing the module up. Likewise, if there are two existing modules which are only used in conjunction with each other and no other modules, there is little reason to have them as separate modules. Finally, if a single module is much more complex than other modules in the asset or even other modules of

the same Level of Modularity, it may also be advisable or even necessary to simplify the design of that module to help unify the modular asset as a whole.

5. References

Star Wars Galaxies. Retrieved from http://starwarsgalaxies.station.sony.com/ Anarchy Online. Retrieved from http://www.anarchyonline.com/ iii Eve Online. Retrieved from http://www.eve-online.com/ iv The Matrix Online. Retrieved from http://thematrixonline.station.sony.com/ v City of Heroes. Retrieved from http://www.cityofheroes.com/ vi City of Villains. Retrieved from http://www.cityofvillains.com/ vii Second Life. Retrieved from http://www.secondlife.com/ viii Guild Wars. Retrieved from http://www.guildwars.com/ ix Lineage 2. Retrieved from http://www.lineage2.com/ x Rappelz. Retrieved from http://rappelz.gpotato.com/ xi World of Warcraft. Retrieved from http://www.worldofwarcraft.com/index.xml xii Maple Story. Retrieved from http://www.maplestory.com/ xiii Ragnarok Online. Retrieved from http://iro.ragnarokonline.com/ xiv Puzzle Pirates. Retrieved from http://www.puzzlepirates.com/ xv Disneys Toontown Online. Retrieved from http://play.toontown.com/webHome.php?r=838673&r=172401&r=30757&r=865698 xvi Westward Journey. Retrieved from http://corp.163.com/eng/games/westward_journey.html xvii Lord of the Rings Online: Shadows of Angmar. Retrieved from http://www.lotro.com/ xviii Warhammer Online. Retrieved from http://www.warhammeronline.com/english/home/index.php