After finally completing Die By the Sword in 1998, the guys at Treyarch were pretty damn proud of themselves. Sure, it was late, and it wasn't a phenomenal seller, but the reviews were great and we learned a lot from our mistakes. We were dying to start over and write our dream game, a game that would combine the animation, sword-fighting, and creative level design of Die By the Sword with cooperative online play. We imagined a sort of 3D Diablo. Our three-day postmortem meeting for Die By the Sword turned into a design and planning meeting for Die By the Sword II, as we were calling it then. We signed a contract with Interplay and we were off. While the game designers were writing design documents, the programmers were busy with R&D into dynamic adaptive level of detail, asynchronous models of online play, and a new scripting language. All of the while we were simultaneously kicking out an add-on pack for Die By the Sword called Limb From Limb.

Screen shot from Draconus

The new game would have the same core management as Die By the Sword: Peter Akemann was lead programmer and Final Word on Everything, Chris Busse was producer, Chris Soares lead artist, and Mark Nau lead game designer. Three months later, we had finished the research, completed the design document, created an initial milestone schedule, and coded a Direct3d renderer and a nifty scripting language. But trouble was brewing; Interplay didn't want a game that wouldn't be done by next Christmas. They weren't excited about a PC game. The Dreamcast was the hot new topic and naturally they wanted us to do our game for that console. They were looking for someone to sell our contract to.

Still, the new accelerated schedule was approved. In my experience, publishers love overoptimistic schedules. They don't seem to care if a developer is smoking crack, even though it happens quite often that the publisher is the one who ends up funding the inevitable, costly slip. Interplay sold our Crave, but then something weird happened. They didn't sell the entire game to Crave, just the domestic publishing rights. They kept the international publishing rights for themselves. We ended up developing the game for two publishers: In America for Crave as Draconus (Interplay did not sell the Die By the Sword title to Crave) and internationally for Interplay as Dragon's Blood. We were living in interesting times.

All things considered, it's no surprise we didn't make it. We did, however, perform many small miracles. We managed to get our C++ code base ported to the Dreamcast Shinobi/Kamui libraries in time for E3, surprising Crave and creating a small sensation. The artists created all kinds of cool characters and levels and special effects. The programmers pushed various Dreamcast features further than they'd been pushed before. We built a new engine from scratch, using object-oriented techniques, design patterns, and generic programming: The programmers had all been exposed to C++ before they started work on Draconus.

In The Mythical Man Month Fred Brooks points out that you'll realized you're not going to make it three weeks before the scheduled end of a project. Appropriately, three weeks from when we were to have our "launch title" done, Akemann and Busse realized we weren't going to make it. We rescheduled for a Christmas launch. Three weeks from that second due date we realized we weren't going to make it. So we rescheduled for a launch that would come out the day before Christmas. (Not too useful if you want to sell titles for Christmas. We were telling ourselves that all those people who received Dreamcasts for Christmas would want to go out and buy our game.) Three weeks from that due date we realized we weren't going to make it. During this period of always almost making it we were fully death marching. The bug list went over three thousand and took five months and about $400,000 more than we has budgeted to tackle it, but we suceeded. To quote Patrick Hughes, one of the coders, "Hey, it shipped." It was nine months past the date we had originally scheduled, but still, it shipped.

Tools

On the PC side, we used Visual C++ 5.0 for development. After switching to Dreamcast, we ended up settling on the Metrowerks compiler and the Codescape debugger. Although one of the promises of porting to the Dreamcast was that you'd be able to use Windows CE with DirectX and just recompile, it didn't work out that way for us. We found that it was easier to rewrite our renderer to use Kamui calls and our file access routines to use Shinobi calls than to try and fix the mysterious problems brought on by WindowsCE. On the bright side, Kamui and Shinobi are nice, thin, and straightforward libraries that were really quite easy to use.

The characters were modeled in 3d Studio MAX 2.5. We didn't use Character Studio. The artists divided the characters into segments manually and the exporter would reassemble them, joint by joint, into a single mesh and then reduce the detail on them. Some of the animations, like ladder climbing and certain cut-scenes, were done with these segmented characters in MAX and then exported for the game. Other animations (the flying dragon) were hard-coded by Pete using his VSIM technology, or were done by a game designer creating sequences of moves. Textures were painted in Photoshop.

For the sounds, we used the Digidesign Pro Tools system: Pro Tools 5.0 on the software side and a d24 system on the hardware side. We also used Bias' Peak for down sampling and editing.
All the characters voice over effects and processing was done in the Pro Tools environment. Various plug-ins were used to get the desired effect on certain creatures and characters. Most notably, Marto the Troll and all his troll friends had to be pitched down in order to make them sound like hulking brutes.

The music was edited and mastered using Pro Tools as well. This had to be done in order to get the music to "fit" in the Dreamcast. We had a lot of cool music and a limited amount of space to put it in. The sound effects Sergio Bustamante and Keith Arem created all were made in the Pro Tools environment. Most, if not all, of the sound effects are originals and were created using Pro Tools to mix sound effect-elements together.