Postmortem: Sierra's SWAT3 Close Quarters Battle

SWAT3: Close Quarters Battle is a first-person tactical simulation where the player assumes the role of element leader in a five-man SWAT entry team. SWAT stands for “Special Weapons and Tactics”. In SWAT3, you are responsible for serving high-risk arrest warrants, rescuing hostages, neutralizing (killing or arresting) terrorists, and defusing bombs and other weapons.

Our goal was to recreate the SWAT experience. We didn’t want to simply create another first-person shooter. In fact, it seemed like everyone was coming out with a first-person shooter and we were concerned about getting lost in the crowd. Instead of shooting everything that moves, we wanted to create a simulation where winning a mission meant following proper SWAT tactics: using proper room clearing techniques, ordering suspects to drop their weapons, and, yes, using deadly force when necessary. But we also wanted it to be fun, and choosing between fun and realism turned out to be quite a challenge.

Officially, SWAT3 is the sequel to SWAT2, but SWAT2 is a 2D RTS and was developed by an entirely different team. Both our producer (Rod Fung) and designer (Tammy Dargan) worked on the original SWAT, which was based on full motion video. While SWAT and SWAT2 received only marginal reviews, they sold like crazy. SWAT alone has sold over a million units to date. The name recognition has been great for sales, and Rod and Tammy’s SWAT experience and enthusiasm helped keep the team focused. While working on the SWAT and Police Quest series, Rod and Tammy developed several contacts within the LAPD SWAT community, including Police Chief Daryl Gates and a SWAT element leader named Ken, our primary consultant.

Early in the development cycle Ken put on a one-day presentation on SWAT procedures and tactics. He brought some videos of his more memorable missions, including the 1996 Hollywood bank robbery. It was like a scene from The Matrix: two guys (Ken calls them “knuckleheads”) with body armor and automatic rifles walking down the street shooting at everyone in sight. At one point, they even start shooting at the helicopter cameraman.

Ken’s dynamic entry missions were especially amazing: SWAT officers would blow open a suspect’s front door, throw in some flashbangs (extremely loud and bright explosive devices), and charge in yelling for compliance -- not something you’d want to be on the receiving end of. We realized that if we could recreate that experience, along with the other things we wanted to do, then we’d have a winner.

Midway during development, several titles were released that contained elements of SWAT3, the foremost being Red Storm’s Rainbow Six. That fact that Rainbow Six was so well received really excited us, because it confirmed what we believed all along: a lot of people were ready for a tactical simulation that went way beyond run-and-gun. This helped energize the development team and gave us someone to compete against. In the end, though, it’s unfortunate that we’re so often compared to Rainbow Six, because at the core they’re very different games. People who like tactical simulations should enjoy them both.

Conceptual sketches for SWAT3

Most of the code for SWAT3 was written from scratch. The character animation system, art import tools, client/server architecture, AI, and most of the graphics engine were all developed over the eighteen month development cycle. A couple of people on our team developed a prototype using Quake, but our project had so many new features that we decided we’d be better off developing our own engine. That decision really paid off: between the graphics engine and some amazing level designers, SWAT3 has some of the most realistic environments ever seen in a game. That said, I’m not sure if writing your own engine would be a good decision today. Besides being a tremendous amount of work, it’s extremely difficult to develop gameplay elements until the engine and art process have stabilized. The programmers can only estimate the memory, texture, and polygon limitations, and artists can’t really lock down their levels until late in the project. Having a finished engine helps you focus on what’s really important: gameplay.

We started programming in May, ’97 and planned to ship for Christmas ’99. Management was, um, very insistent on meeting the Christmas deadline, so we had to build the best game we could in the available time. We originally planned to ship both single and multiplayer gameplay, but later decided to focus on making a great single player experience and shipping multiplayer at a later time. There just wasn’t enough time to build a killer single player game and multiplayer that would compete with Quake3, Unreal Tournament, and Rogue Spear, all of which were coming out at the same time. It turns out we made the right decision: SWAT3 stands out for its compelling single player gameplay. Not having multiplayer has hurt our reviews a bit, but we’re convinced they would have been much worse otherwise.