Classic Postmortem: Obsidian's Knights of the Old Republic II: The Sith Lords

13 years ago this week, LucasArts published Obsidian Entertainment's role-playing game Star Wars Knights of the Old Republic II: The Sith Lords on PC and Xbox.

Since then it's earned a reputation as one of the most ambitious and morally complicated Star Wars games ever made, and so to celebrate its 13th birthday we've gone ahead and republished Obsidian's postmortem look at how it was made. The below postmortem, which is studded with images pulled from various versions of the game, was written by Kevin Saunders (then a senior designer at Obsidian) when it first appeared in the April 2005 issue of Game Developer Magazine.

Although the original Knights of the Old Republic (KOTOR) was developed by BioWare, Obsidian Entertainment developed the sequel to the LucasArts game, Star Wars Knights of the Old Republic II: The Sith Lords (KOTOR II). Upon release, the original was met by both industry acclaim and consumer praise. The challenge for Obsidian was to attain the high expectations of such a game's sequel, while adhering to an ambitious development schedule.

In tackling this project, we decided to build upon the proven success of the first game's design and were careful to not fix aspects of the game that had already proved their worth. Instead, we identified areas of KOTOR that we could expand upon to create a better experience in a flavor similar to that of the first game. Almost without exception, every game design enhancement to KOTOR II was a natural extension from the original game, not a redesign.

This project was Obsidian's first, though most members of the 33-person development team had several years of game development experience. About half of the team worked together at Black Isle Studios, while other members came from companies such as Blizzard, Electronic Arts, and Activision. About one third of the team was hired within the last six months of the project, including 70 percent of the design staff.

KOTOR II had dedicated producers on both the developer and publisher side, and all the members reported to one of three leads in art, programming, or design. The leads guided the development process, but gave considerable ownership to team members. BioWare provided the engine and toolset from KOTOR, which was the used as the foundation for KOTOR II. LucasArts provided all quality assurance, audio (sound, music, and voice), rendered cinematics, and also assisted in art asset creation. Most of the production phase of the project involved considerable overtime; however, employees were driven to work any extra hours not due to management, but due to their dedication to making a great game.

What Went Right

1) ALIGNMENT AND INFLUENCE SYSTEM

The best example of expanding upon BioWare's strong game design is the influence system that Obsidian developed for KOTOR II. One of the most successful elements of the original KOTOR was its treatment of non-player character party members (companions). Their elaborate histories, personalities, and roles in the story were frequently praised as great features of the first game. Another powerful trait was the player character alignment system, the players' ability to earn lightor dark-side points for many of their actions.

Both of these elements were emphasized in KOTOR II. We expanded them by allowing players' relationships with their companions to change based upon the their decisions as the game progresses. When companions interject their own expertise or opinions during conversations, the player either gains or loses influence with that companion based on how they react to what was said.

The player's influence (either positive or negative) affects what types of information the companion will divulge and can even have more extreme effects, such as allowing the player to take on the companion as a padawan.

Additionally, the player's alignment on either the good or evil side alters most of the companion relationships based upon his or her influence with them. Two of the prestige classes, Sith Lord and Jedi Master, had an even greater effect on their companions' alignments. The influence feature has been enthusiastically received by KOTOR fans.

We also increased the importance of alignment in other ways. Some of the dialogue, particularly from companions, changes when the player noticeably leans toward the light or dark side. A special lightsaber crystal attunes itself to the player's alignment and level, with its powers and abilities changing as the player does. The player's light or dark side decisions even effect who can become a player's companion.

2) MEMORABLE MOMENTS

Another aspect of KOTOR that received high praise was the non-linear game play. After completing the initial sections of the game, players were allowed to choose their own course through the game world. We sought to expand upon this idea by combining the non-linear aspect of the game with a philosophy of implementing memorable game moments.

Often, what players remember from games are a few key moments that have great impact. Through cut scenes and innovative exploitation of the game engine, we created many of these moments throughout the game's story to continuously entertain and surprise.

Many of the events that occur in the game feel outside the scope of a traditional RPG, providing an experience that one normally doesn't find in the role-playing genre. For example, at several points in the game, the player takes the role of either a companion or a non-player character that's not in the party. These interludes allow you to experience different perspectives and to interact with the world in different ways. We felt this was also true to the Star Wars genre, shown m A New Hope and Empire Strikes Back, among others, as the "party" splits up to pursue different agendas.

We also emphasized the impact of the player's decisions. The influence system, as discussed above, is a primary example of how we actually did this. Another example occurs on Dantooine when the player learns of a impending conflict between the provisional government and some mercenary groups. You can take several actions to assist one side or the other, such as repairing the settlement's defenses or reprogramming them to fire upon friendly troops. Your decisions affect the battle that ultimately takes place in visible ways, and the conflict's outcome affects events later in the game. Many of the late-game elements vary considerably based upon the choices made throughout the entire game.

3) AUTOMATED DESIGN SYSTEMS

To improve balance throughout the game and to conserve design resources for content creation, we created semi-automated systems both for placing items and scaling non-player character difficulty.

In KOTOR, items were placed throughout the game, resulting in the same ones being found during every play-through. Because of the game's open-ended nature, the order in which items were found was unknown to the designer, which can reduce the entertainment value of finding the items. For example, if players find great armor fairly early in the game, they would be less interested in any armor found later.

In KOTOR II, we implemented a random item placement system. Every item has an effective character level for which the item was designed. Each area of the game, upon first entering it, is stocked with treasures appropriate for the player's level. We included a small random chance for a particularly powerful item and incremented this chance each time a powerful item was not found. When a powerful item was placed, this random chance was reset. This implementation guaranteed the frequency of powerful items that we desired. We also could override the random factors to ensure, for example, that a key bounty hunter finds a powerful blaster rifle before he'll definitely need it.

The random item system helps ensure that players continuously find interesting items throughout the game. It also aids replayability since each play-through can yield different items. (A single play-through of the game might reveal about half of the total possible items.) Finally, by being automated, it was easy to make sweeping changes to item distribution and to add new items without requiring placed items to be rearranged. Plot-related items were still placed throughout the game.

A second automated system, called auto-balancing, was used for adjusting non-player character difficulty. In KOTOR, non-player character difficulty was, for the majority of encounters, static. Because of the open-ended world, this design created severe balancing challenges. Any given area might be traversed by either a level 8 character or a level 18 one. For KOTOR II, we used auto-balancing to both address this problem and to save design time.

Upon entering an area for the first time, the difficulty levels for all placed creatures are scaled based on the player's level. We differentiated between five degrees of scaling to meet various needs and used controlled tests to set the levels of difficulty to appropriate starting points. Autobalancing was modest for common enemies so that you feel that they're consistently gaining power throughout the game. For bosses, however, auto-balancing was more extreme. We wanted to encourage players to try new tactics and use different items to defeat these opponents. Using it in this way, the auto-balancing feature ultimately saved considerable design time.

4) ITEM CREATION AND UPGRADING

KOTOR's item system allowed the player to upgrade some of the equipment they found. The upgradable items had between two and four slots for upgrades, depending upon the type of item. For example, upgradable armor could be reinforced with a mesh underlay. Except for lightsaber crystals, item upgrade options were very limited. You didn't have to make any choices involving the upgrades, and all upgradeable items were acquired fairly early in the game. In short, you only had to take a few extra steps in order to get your upgradable items to reach their full potential.

For KOTOR II, we added choices to the upgrade process. For example, you can now equip your armor with one of 30 underlays and overlays (though many of these are variations on each other). We increased the number of lightsaber upgrade slots from three to six. All told, KOTOR II has more than 200 upgradable items. We also made many more items upgradeable, so that finding them is the norm instead of a rarity.

Another improvement we incorporated was an item creation system. Almost every found item can be broken down into components. You can then use these components to create different items. Your skill level determines which items you can create. And while most normal items cannot be created, all upgrades (except for lightsaber crystals) can be. Additionally, most disposable items, like medpacs, mines, and grenades, can be broken down into chemicals, which can then be used to make other disposable items.

The item creation system served multiple goals. First, it increased the importance of the player's skill choices because we felt that skills were underutilized in KOTOR. The item system also improved inventory management and customization by allowing players to exchange items they weren't interested in for those that better suited their preferred style of play (demolitions experts could break down items and turn them into mines, stealth characters could break down mines and use them for stealth equipment, etc.). Upgrades were made fairly rare in the game, so the item creation system could fill its own niche without reducing the thrill of finding new items.

5) NUMEROUS MISCELLANEOUS ENHANCEMENTS

We expanded in small ways on many other elements of KOTOR. These improvements helped the game to feel more upto-date despite using the same basic graphics engine and hardware as the original.

The choreographed combat, which many KOTOR players enjoyed, was expanded to include new unarmed animations. To help showcase these animations, we implemented some side quests that required them, such as duels with the Handmaidens and the Mandalorian battle circle on Dxun. We also made unarmed combat more appealing by providing Jedi with unarmed combat bonuses and having two of the player's possible companions be especially powerful when fighting unarmed.

We further enhanced the combat system and the distinction between the Jedi classes by introducing the concept of lightsaber forms, which were taken from the Star Wars source material. Each provides some significant benefits, but also includes weaknesses. For example, the Soresu form was best at deflecting blaster fire and was also suitable for squaring off against a single opponent. But against a lightsaber-wielding foe or many enemies, it had significant drawbacks. Force forms had a similar effect on the use of Force powers. Players can eventually learn seven of the total 11 forms, depending on their class, expand ing their tactical options.

The user interface was another element we sought to improve. We added a second weapon configuration and allowed a quick shortcut to switch between the two so that players can easily change weapons from melee to ranged combat, for example. We also reorganized the inventory and equipment screens so you can sort items by type. We added additional information to many screens to reveal some modifiers and effects that were hidden in KOTOR.

WHAT WENT WRONG

1) QUANTITY OF CONTENT

Given the short production period and modest staff for the project, our goals for number and size of areas, quests, and non-player characters were unrealistic, resulting in some aspects of the final product feeling unfinished. Simply put, we were too ambitious in terms of total content, and this was realized much too late. Although our content ambition was the biggest problem we encountered, the problem seems to be fairly rampant among most development studios. In fact, this specific issue has been discussed in so many Game Developer postmortems that we don't have much to add to what's been said in the past.

2) CUT SCENES

KOTOR II has dozens of cut scenes, plenty more than the first game. Many of these cut scenes, as well as numerous others that were not in the released version of the game, asked more of the game engine than it was designed to handle. In particular, any cut scene involving movement-which was most of them-suffered from unpredictable results during the making.

Countless hours were spent by gameplay programmers to finalize cut scenes only to later find them broken for unknown reasons. Late in the project we observed that something apparently innocuous, such as adding an ambient sound object to a distant location on the map, could throw off a cut scene's timing. These problems added up to make an ambitious schedule even more difficult. Our programmers were often unfamiliar with the deeper nuances of the engine, so debugging and correcting these problems proved problematic.

3) TECHNICAL LIMITATIONS

One of the common criticisms of KOTOR II has been low frame rate. One assumption made by those who played the game was that we failed to address some of the technology limitations of KOTOR, which also suffered from choppy frame rates.

These performance issues were partially the fault of design, and partially due to the time we were able to devote to optimizing the engine. While the time we were able to spend optimizing the game engine helped, we used most of those savings on making more detailed models, more nonplayer characters, and larger areas (to reduce the frequency of load times]-often having more than twice the active content than a comparable region in KOTOR. The load times between KOTOR and KOTOR II are comparable, although the frequency of them is much less in KOTOR II.

This approach had mixed results. Few who play the game notice the liveliness of the areas or the fact that they contained a greater variety of non-player character appearances. Had we reduced area size to KOTOR-sized levels (and thus decreased the amount of content required to make an area feel sufficiently and diversely populated), the overall game would have been more enjoyable.

4) AI IMPLEMENTATION

One element we wanted to improve in KOTOR II was the Al for both enemies and, especially, nonplayer character companions. The Al in KOTOR is quite simplistic. Companions attack their target, closing in on it if they're using a melee weapon, and attack it until one or the other is dead. They would never switch weapons. The ability to control a characters' Al was limited to specifying if grenades or Force powers should be used. How these special abilities were used was based upon simplistic algorithms, without taking into account the tactical situation.

We did not focus on improving the NPC AI until late in development, and a lack of programming resources allowed us to implement only the simplest improvements. We added a couple of behavior varieties to allow the player to control how far his companions would stray and whether they would prefer ranged or melee weapons. We had already committed to removing the companion Al interface from an obscure start menu screen and we placed it as an option on the main interface. This would have been a good improvement except that the Al options were not interesting or useful enough to warrant such prime placement.

In the end, our efforts had essentially no meaningful net impact on the game. We should have either realized that improving the Al was beyond the scope of what we could accomplish, or we should have scheduled more resources to the task so that the Al would be significantly improved.

5) INSUFFICIENT POLISH TIME

Many of the new game systems did not receive significant polish time. Instead of being well-refined systems, the final implementations were essentially rough drafts with design flaws that we were able to identify but not address.

Companion dialogue was implemented fairly late in the development cycle, which limited our ability to fully experiment with the influence system. The number and quantity of influence changes were not well-mapped or balanced between various characters.

During late testing, we found that we simply did not have enough influence opportunities for some of the characters. With insufficient time to properly address this issue, we simply increased the magnitude of each influence shift. Fortunately, given the praise the influence system has received, this minor change adequately addressed the issue. But at its core, the influence system is not as well designed as we would have liked.

The auto-balancing system was less successful in achieving its goals. We achieved complete play-through of the game only shortly before our gold date and were unable to evaluate issues like game balance until it was too late to make many changes. As a result, we erred on the side of making the game too easy and, therefore, largely undermined the potential of auto-balancing as a tool to provide a consistently challenging experience.

Obviously, better initial design and planning would have been the best way to prevent these types of problems. But a more realistic approach might have been to schedule more polish time for properly addressing design flaws such as these.

MAY THE [ENGINE’S] FORCE BE WITH YOU

Without the high quality engine and toolset from BioWare and the extensive support from LucasArts, a game of this caliber would not have been remotely possible in little more than 14 months. We are applying the lessons learned from the goals we didn’t fully attain to our future projects, including Neverwinter Nights 2. Overall, we’re pleased with KOTOR II and the feedback we’ve received. Given the ambitious development cycle of the project, we feel we accomplished our goal of creating a worthy sequel to the award-winning KOTOR.