Postmortem: Bohemia Interactive Studios' Operation Flashpoint

The story of Operation Flashpoint's development is quite unusual in the game industry these days. For one thing, the team didn't start out as professionals; originally only the lead programmer was allowed to work on the game full-time. Switching publishers three times, starting a new company, growing the team from one to 12 full-time members, and moving offices five times during the game's development were just some of the hurdles we had to clear. Only the team's vision and obsession for the game remained consistent from the very first playable version until the end. It's not possible to describe whole story in the space given for this article, so let's just jump directly to the final moments.

It was 8 p.m. on Friday, May 25, 2001. Our publisher's representative, who had been in Prague for the last few days to make sure everything was going O.K. as we were finalizing the gold master, left Prague feeling confident that things were going well — the disc was almost ready and could be sent to final testing and then to manufacturing after some weekend testing.

Meanwhile, our lead programmer (to make matters even more exciting, he was then working at his temporary home in France for couple of weeks) was trying to resolve some serious graphical anomalies with the hardware transformation and lighting (HW T&L) rendering. If he were to fail, HW T&L would not be included in the final release. If he solved it, some data organization changes would be necessary to suit the needs of the HW T&L. He spent nearly the whole day resolving some random crashes that appeared in the game during the last day, going back and forth over e-mail with an Nvidia support engineer. The crash was fixed by late afternoon, and by 10 p.m. it looked like the HW T&L problems were at an acceptable level. Around midnight, the tools that would perform the data format change were ready.

On the other front, the team had received the final localized strings for the game. However, the file containing the core strings of the game that had been delivered by our publisher appeared to be untested and unusable. After spending a couple of hours dealing with it, most of the team had to go home to have some sleep. Still, the team leader stayed behind at the office, trying to use the new HW T&L data format, going over each step by phone or
e-mail with the lead programmer (while also trying to implement new localized string tables and fix some problems in the campaign and missions). At 3 a.m. it looked like all the data had been converted — and both the lead programmer and the team leader could go have some sleep.

Saturday morning, our publisher realized that the gold master hadn't actually been delivered. Tensions rose even further, and nerves began to unravel. Only two days remained before mass production was scheduled to begin. Everyone on the team had been working since early Saturday morning, but at times a successful end to these last-minute crises seemed to be so far away. By around 5 p.m. on Saturday, most of the important issues in the code had been resolved, and the lead programmer decided to take another look at the HW T&L implementation. Luckily, within a few hours, he suddenly discovered the root of all of the HW T&L problems and fixed them. The plan was to deliver the gold master to our publisher via FTP by that evening. Nobody expected that it would actually take until Sunday morning. After a long, sleepless night of playing through the game and fixing any problems that appeared, everything looked fine, and most of the team could finally go to sleep again.

	Operation Flashpoint's mission was to deliver the tension of full-scale conventional military conflict.

With some relief, we finally started the game upload on Sunday around 9H But were we done? Not yet. Suddenly, a seagull stopped flying in some of the in-game cutscenes. The team leader called to wake up the lead programmer in France: "The seagull is not flying. What should I do?" We had to stop the upload until the lead programmer delivered necessary code fix. After the project leader received the updated files from the lead programmer, he started to rebuild the game in Visual Studio. It was Sunday around noon, and the game had finally gone to the publisher for final testing.

The publisher's test staff started playing the game Sunday afternoon. Everything went smoothly at first, but later they discovered one serious scripting bug in one of the campaign missions that made it unplayable. Late in the evening, they called the team leader about the bug, and he had to drive to the office after sleeping just a couple of hours over the past three days to fix the bug as quickly as possible and then upload the fixed version to the publisher's server in the U.K. Around midnight Sunday night, the disc was finally ready to go.

Three weeks later, hundreds of thousands of copies of the game were available in stores worldwide. In the meantime, the development team was playing the game, terrified of finding a disastrous bug. Fortunately, no such critical bug appeared. Considering the amount of work we'd done on the game in those last couple of days and hours, the risk of finding some major problems was pretty high. On Friday, June 22, the game was released, and it immediately became the top-selling PC game in many countries. The team knew that their mission was successfully completed. The passion and hard work of every single member of the development and publishing teams started to pay off.