The following is a very insightful written history of the development of Aliens versus Predator (1999) from the developer and CEO of Rebellion, Jason Kingsley. It was published over a span of eight weeks on Fox Interactive's website in the weeks anticipating the game's release. I post it here for informative.
Original link here: http://web.archive.org/web/19990903204829/http://www.foxinteractive.com/products/products.html?/avp/index.html
Game developer Jason Kingsley tells the creation of Aliens vs. Predator. Check back each week for updates.
Alien Versus Predator Diary
by Jason Kingsley
Part 1
Over the next eight weeks, I will try to take you through some of the processes that we've gone through in the creation of Aliens Vs Predator. Hopefully, this should give you some insight into just how complex (and enjoyable) a process games development is.
For this first installment, however, I thought that I would take a short trip down digital memory lane, to let you know how a relatively small company from Oxford found itself working on a massive project such as Aliens Vs Predator.
The story of AvP goes right back to the beginnings of Rebellion. Back in the days of the Commodore Amiga and Atari ST, my brother Chris and I were just another pair of backroom 'coders for hire', working on game conversions. We were real enthusiasts, working as much for love as for money (which, if you could have seen our pay cheques, you would agree was just as well).
Like all enthusiasts, Chris and I tried to keep up with the latest developments in the business, a process which led us to hear of some new technology under development at Atari.
So, armed with some demos of our own 3D work and the sort of cheek that you can only have when you're still young and enthusiastic, we approached Atari in the hope that we might be among the first to develop software for this new dream machine (well, if the truth be known, we really wanted to see if we could get our hands on some free hardware).
We should, of course, have been shown the door, but for some reason we found ourselves presenting our wares directly to Atari's UK boss, who offered us the chance to become the first team outside of the US to write games for his company's new cartridge-based games machine... the Atari Alpine. (Later, thankfully, this name was changed to the more threatening 'Jaguar'.)
Rebellion was duly born in the basement of our parents' home and we set about building a team to get to work on the two titles which Atari had commissioned us to write. One of these games - a licence from Activision - featured two of the best-known characters from cinematic science fiction: the Alien and the Predator.
At around this time, id's ground-braking Wolfenstein appeared and we immediately knew that we had a format for our game: although we decided to take things forward a little by introducing innovations such as texture mapping and stop-frame animation (using hand-built models of the main characters).
The resulting game, Alien Versus Predator (note the singular 'Alien') was a major hit (it apparently sold through to 84 per cent of the Jaguar's user base) and went on to win five magazine awards.
More importantly for us, however, apart from bringing a good deal of enjoyment to the majority of the world's Jaguar owners (which was, admittedly, a rather exclusive club), Alien Vs Predator also started Rebellion's relationship with 20th Century Fox. So, when Fox Interactive was formed in 1995 they came to us to talk about the future of AVP... which is where we will pick things up next week.)
Part 2
Cast your minds back to 1996. When in 1996? What month/season? Give the reader some kind of tangible timeframe. Football was coming home, Lara Croft was stretching her first Lycra top and the Spice Girls were telling us, at great length, what they really, really wanted. Doesn't seem that long ago, does it?
So, I suppose it would come as a great surprise to you if I said that, when we started work on AvP in late 1996, we were covering three formats: Saturn, PlayStation and the first generation of Pentium PCs. What's more, there was nothing to choose between them.
I'd caution against much mentioning of the cancelled platforms, it may cause rancour amongst the readers. PlayStation esp. could be bad. People understand dropping Saturn but people still WANT this game on PSX.
At Rebellion we didn't have too much time to think about Lara or the Spice Girls. We had managed to convince one of the world's largest entertainment corporations to trust us with some of their most famous characters and now we had to come up with the goods. Things were getting serious.
Not that serious though, as a quick look through some of original design documents confirms. For example, how serious could we have been when we suggested motion capturing Sigourney Weaver in her knickers? (For some reason, this idea seemed to disappear quite quickly.)
Or how about this for an idea: if you chose to play as the Alien, you would have to start as an egg, become a face-hugger and then inseminate another character before you could join the game? Needless to say, that idea didn't make it too far from the drawing board either. No, these aren't things we want discussed. How about talking about all the model building, the involvement of the guy who wrote the Colonial Marines Technical Manual or the use of actual movie props, instead?
Slowly but surely, though, a proper game began to emerge - and as it did it became clear that something was going to have to give on the hardware front.
Our main problem concerned the crucial decision as to whether the game should be built around sprites or polygons. Nowadays, of course, this is a non-argument, but we were still trying to develop a game for three separate hardware types and had painted ourselves into a corner by promising to match the graphics across all three. This meant that we had to base everything on the lowest common denominator, the Saturn.
The problem we encountered with Sega's ill-fated 32-bit machine was that it couldn't 'cookie cut' textures. (Cookie cutting is a process whereby you take a section of a large texture and map it onto a smaller object.) For this reason we decided to go with sprites.
I'd definitely gloss over this shady period (esp. this whole "lowest common denominator" stuff). I'd compress time a little so that these platforms were considered (and rejected) a lot earlier than they actually were. This stuff paints neither Fox nor Rebellion in a good light. I'd also cite more relevant comparisons to place the reader in the correct timeline (e.g. "we were considering sprites which seemed okay because Duke Nukem was doing very well with them" or "*and polygonal figures were a dream still 12 months distant to everyone*" or "*and round about this time a company called 3DFX started something big"). That kind of thing, more of what was happening elsewhere in games at the time.
In today's fast and furious, hardware-accelerated world, this decision seems ludicrous. However you should remember that we were still living in a pre-3dFX world. To make a comparison, in those days it was considered state-of-the-art if game characters were built of 80 to 100 polygons - in our final version of AvP the Alien Queen will use more than 2,000.
By the time that we were beginning to realise the error of our ways (the early Summer of 1997), Fox had just appointed a new producer on the product (Dave Stalker, who is still overseeing production to this day) and we decided to bite the bullet and approach him about switching our game to polygons.
By the early summer of 1997, we were beginning to come to a final realisation: sprites were not the way forward. We decided to bite the bullet and approach Fox about changing all of characters to polygonal models.
To his eternal credit, Dave agreed without hesitation...even though this meant that we would have to ditch virtually eight months of work. It also meant that we would have to wave goodbye to Sega's Saturn, but as so few people had waved hello to it in the first place, this wasn't too heart-breaking.
It was also round about this time that Dave Stalker came onboard as the Producer at Fox and he agreed without hesitation (a lot of the folks at Fox had apparently been having concerns similar to our own). This decision also meant it would be impossible to produce the same game for the Sega Saturn and it was decided to not continue with the Saturn version of the game.
This did leave us with one major headache though. 1997's E3 exhibition was just over a week away and we had to have something to show...all of a sudden we were looking for an extra end of the candle to burn.
Part 3
If you've been following the story so far, you will still be hanging on the edge of the cliff where we left you in part two, armed only with the knowledge that we had just made a significant decision on restructuring the game into true 3D.
So, if you can drag yourselves back to the relative safety of the cliff-side, let's pick up the tale with seven days to go to the Atlanta E3 show in 1997.
In these seven days we had committed to more or less create a working game from scratch. This may seem like an impossible task, but luckily we had 'future-proofed' a good deal of our technology for just such an eventuality. This effectively meant that, while our characters had been two-dimensional, our interior levels were already 3D and would now only have to be populated by equally well-rounded characters.
This still left us with a next-to-impossible task (rather than a totally impossible one), so it was a case of 'all hands to the pumps' as we set to the task of creating a single level, which featured the Predator fighting a squad of marines.
In short, we had three tasks to fulfil: to create all of the characters (although, as the action was designed to be seen from 'within' the Predator the Predator's point of view, we only really had to create the marine squad), render and texture-map them and then drop them into a working and tested game level.
You often hear about people working through the night to meet a particularly tight deadline, but have you ever heard of anyone working through a week? Well, this is exactly what happened, as some key members of the team went almost four days without any sleep whatsoever in order to complete this demo.
Amazingly, this intense burst of activity actually brought in a working game level with two days to spare, which gave us exactly that time to play-test it—a luxury which we never dreamed that we would have.
In addition to helping the team on these trifling matters, Chris and I also went about the important tasks of finding flights and accommodation for the show (not as easy as you might imagine with a week to go) and devising what we believed would be a fool-proof fall-back plan just in case the code which we took to America didn't actually work. (This involved having one member of the team on standby in the UK for almost 48 hours until we eventually called in to say that everything was in order.)
On arrival at the show, we soon saw how crucial our efforts had been, as Fox had honoured us with pride of place, right in the middle of their most impressive booth. If we had failed to come up with the goods, this would have been one of the most high-profile failures in recorded history.
Luckily, the opposite was true and we received some very favourable notices from the press who now saw our efforts for the first time. We returned from the show, exhausted, but more sure than ever that we had something special on our hands.
A small, but perfectly formed, postscript to our Atlanta adventure is that one of the first things that we did on our return to the UK was to ditch all of the character models which we had created in such pressured conditions. This was not because we had been unhappy with their quality (quite the opposite in fact), but because we had by this time made a firm decision to introduce 'hierarchical modelling' to our game characters, a reasonably complex process which we will cover next week.
Part 4
Although the reaction to AvP's first public airing at E3 1997 had bordered on the ecstatic 'ver favourable' you said last time, nothing about 'ecstatic', both Chris, myself, and the whole team knew that we still had quite a way to go before we had the game that we wanted.
The problem was that up until now we had been driven along by the changes in hardware technology (which at that time were more significant than they ever had been, even in the turbulent history of gaming hardware). It was decided that we had to put the brakes on this and turn things around so that we were in charge and technology was working for us.
So, almost as soon as Chris and I returned from Los Angeles, we sat down with the entire team and decided on a hardware specification that would be current at the time of release in the first half of 1999. This, of course, was an almost impossible task but, looking back, the team did quite a good 'Nostradamus' job. During these major meetings (the biggest that we had on the entire project, it lasted about three straight days) the whole team were agreed on the use of 'hierarchical models' to give life to the game's characters.
Hierarchical modelling is a process where each in-game character is effectively built from a complete polygonal skeleton, which is then 'fleshed out' by the careful placement of texture-mapped 'skins. This also allows for very fluid and life-like animation, as the skeleton is moved by the computer's maths processing power, rather than the traditional 'stop-frame' method where an artist draws a number of poses for a character and then these are animated by the programmer causing them to be flashed past the player's eye in quick succession (rather like a highly expensive 'flick book').
In addition to more lifelike and smoother animation, one of the added benefits of using hierarchical modelling is that it allows game characters to be blown apart in quite beautifully 'realistic' ways. For example, with hierarchical modelling, it is a relatively simple mathematical process to remove a character's limb (in fact, with some of the more complex characters, you can even go so far as to amputate single fingers).
This not only adds an extra layer of realism to the visuals, it also adds a few neat little touches to the gameplay. For example, anyone playing as a Marine character has to be quite careful about removing an Alien's hand in combat. As any fan of the Alien series will know, the Alien has acid for blood, meaning that an open artery can be even more dangerous than the severed limb that it has replaced (I forgot to mention that hierarchical modelling also allows us to replace severed limbs and body parts with individual 'blood generators', for extra realism).
Now, at last, it seemed that things were finally moving towards a final product, At last we felt we had a common goal that everyone was happy about, although we were fully aware that there was still more than a year's work to be done. The technical guys were being challenged because the updated system was new and demanded excellent programming skills, while the artists were happy because the character's new-found realism gave them another excuse to delve into Fox's archive library for reference materials... but maybe we should leave that story for next week.
Part 5
One of the most interesting and enjoyable aspects of the creation of Aliens vs Predator has been the fact that we, as a team, have been allowed access to all manner of original development materials which went into the creation of two of the world's favourite science fiction movie series.
As you would expect in this industry, the whole of the team Rebellion are fans of both the Alien series and the Predator movies Aliens and Predator movies. However, while we all thought that we knew a great deal about those films and their characters, it was vital that we conducted research to make sure that our collective memories were accurate.
To this end, we analysed all versions of the relevant movies until we could more or less repeat the scripts verbatim (the things we do to make a living!). This, it turned out, was a necessary 'chore' and we discovered that our own conflicting memories of the films were correct a lot of the time and that, no doubt for artistic reasons, the films themselves contain a great many inconsistencies and contradictions. In the original Predator film alone there are several separate sequences where the Predator uses his 'invisibility' cloak, each one slightly different to the last (there were several different special effects companies used presumably). Also, from Alien to Alien3, the alien creature goes through a number of small, but important transitions in shape, proportion and colour.
Our task, therefore, was to work as closely as possible with Fox's archive department to develop a standard model for each of the three main characters, using the best and most 'game friendly' elements from each of their respective on-screen appearances.
At Fox, Dave Stalker and AvP Associate Producer Chris Miller, along with numerous other people in and around Fox Studios were up to their necks digging out original reference material including photos, drawings, sound effects, even press releases from 1979, for the premiere of Ridley Scott's 'Alien'. Our task was to work with all of this material to develop a standard model for each of the three main characters, using the best and most 'game friendly' elements from each of their respective on-screen appearances.
While this task was a sheer delight which any science fiction fan would happily immerse themselves in, it was also a more difficult task than you might imagine. Every piece of equipment had to be researched thoroughly and then a decision had to be reached as to how (or, indeed, if) it could be incorporated in the game.
Take, for example, the motion tracker device used by the USCMC characters . Take, for example, the motion tracker device used by the USCM characters (that's 'United States Colonial Marines', trivia fans). In the film Aliens, this is a rather 'lo-tech' device which only scans the area directly in front of its user... and even then doesn't do a great job. However, it is clear that the filmmakers limited this device for cinematic purposes (it's a great tension builder) and, after a massive internal debate, we decided to retain the limitations for the game too. After all, it would be a much more effective device if it worked in a full 360 degree range, but it would also remove the 'look out behind you' element that is an essential part of AvP's gameplay.
Also, what about the all-important game element of 'balance'. The cinematic Alien was designed as the ultimate killing machine, so should we introduce some weaker type of Alien life to give beginners a chance of survival? At the end of the day we couldn't bring ourselves to 'dilute' the death machine that is the Alien so, for balance, we gave the Marine a much better entry-level weapon than is standard in First-Person Shooters instead.
This entry-level weapon is the M41A Pulse Rifle which, is a 10mm pulse action automatic rifle, adopted by the Colonial Marines as standard personal issue weapon. The rifle has an over-and-under configuration with a 30mm grenade launcher.
This balancing act is especially relevant when it comes to the multi-player game, which allows players to choose between any of the three main characters... something that we will cover in next week's piece, I think.
Part 6
One of the major attractions of Aliens Versus Predator was the multi-player opportunities afforded to us by the variety of characters involved. However, as we touched upon last week, this was something of a double-edged sword.
Typically, any game which offers a choice of characters, balances the gameplay by making one type heavy, slow and powerful and pitting this against an opponent with less firepower but more speed and agility. This usually works, but it was too simple an equation to fit our needs. This was far too simple an equation to fit the needs of three very different, yet very capable characters.
Our dilemma stemmed from the fact that all three characters arrived at our game in more-or-less complete form. The Alien is nature's most potent animal, but doesn't possess any weaponry (and therefore has to get 'up close and personal' with its intended victim); the Colonial Marine is as vulnerable as any human being, but compensates for this by carrying a complex array of armaments, and the Predator is massive, powerful, 'loaded for bear' and can become instantly invisible. It's not too difficult to put a pecking order to that list, is it?
The traditional way to solve this problem would be to either strengthen the weaker characters (allow the Marines to become invisible or create a range of Alien weaponry) or handicap the strongest (make the Predator walk on his knees perhaps?), but there was no way that we could do this without compromising some of the credibility of the characters and it was crucial to us that they reflect their movie counterparts..
Our solution, therefore, was to institute a 'handicap' system. A simplistic, but useful, way to explain this is to use the age-old children's game of 'scissors, paper, stone' as an analogy. The Alien can easily beat the Marine if it gets to hand-to-hand combat and the Predator can usually beat all-comers by virtue of his superior strength. However, the Marine can defeat the Alien if he keeps his distance and uses his powerful weapons, while the Alien 'sees' right through the Predator's invisibility cloak. The Predator also has a limited amount of energy to operate his weapons and equipment and can only track one species at a time with his enhanced vision system. Also, there is a rewards system within the game that gives far greater points to a Marine for defeating a Predator than vice versa.
Best of all from a game design point-of-view, though, is the fact that this system also eliminates the race to get the most powerful gun, which is prevalent in just about every other first-person 3D game.
Speaking of other first-person games, I have often been asked why we didn't use one of the many 3D engines that are now available. The simple answer is that they were not available when we began work on AvP (when Doom was the benchmark), but the real answer is that we believe that using our own code allows us to include some features in AvP which are unique.
For a start, no other engine allows creatures the freedom to run on ceilings and walls. The lighting in our game world is 'real', so if you want to fight in a dark space, you only have to blow the lights out. We also decided that we didn't like the 'fixed enemies' offered by most of the competition. You know the type of thing - if you walk around a corner and are blown away by a lurking alien, you know that next time around he will be waiting for you, so you plan accordingly. We have created a system which, we believe, removes this drawback and allows for more impulsive, frightening, gaming.
What's more, the decision to create our own engine has given the Rebellion team a chance to develBODY TABLE --> [Note by Olde: the entry prematurely cuts out here]
Part 7
The past six episodes of this diary have been retrospective. However, in this week's piece we are going to come right back up to date by looking at the stage of development which Aliens vs Predator has now reached...testing.
Play testing - the process of balancing a game and making it as pleasurable an experience as possible - is immediately preceded by bug-testing, a slightly more complex process which usually begins at the first 'alpha' stage - i.e. when the game is almost complete, but just not very pretty to look at. This is the stage where major software 'bugs' are unearthed.
Software bugs, as anyone who has ever tried to string two lines of code together will know, are the most infuriating and annoying obstacles facing any game developer. Bugs can occur for the simplest of reasons, a comma or full stop in the wrong place for example, and, worse still, fixing an apparently simple problem can often lead to the development of a whole new bug (or bugs).
The bugs which are found earliest in the testing process are generally the most obvious and the most severe. They can range from something as slight as a character's head being a slightly incorrect shade to the game simply refusing to work from a certain point. As the bug-testing process continues, the bugs become more invidious and difficult to find.
We had a couple of beauties thrown up by the bug-testing of AvP. For example there was one graphical glitch which reversed all of the polygons, i.e. everything which should have been visible wasn't and vice versa. This produced some of the worst graphical atrocities ever to make it to a game screen.
Bug testing never stops! There's no clear defining line between play testing and bug testing. In fact, bug testing is likely to continue after play testing. It's much worse to ship a game with a crash bug than it is to ship one which needs some playability tweaks. Before Fox ships a game it needs to run for at least 100 hours with no (or acceptably trivial) bugs. Even then, this doesn't means it's bug-free, just that we haven't found them yet.
Once again, however, AvP's unique combination of characters has ce game during the testing process, there are probably half-a-dozen which will either be saved for a future product or simply discarded. Like everything else, the finished Aliens Vs Predator will be a compromise of the best ideas put together by a truly creative team, delivered within the final deadline. Next week, in the final diary part, we will pay proper tribute to the team.
Part 8
And so, here we are at the final part of our Aliens vs Predator diary. I hope that, over the past seven weeks, we have given you some small insight into how a game is built, from concept to completion.
We are now into the final weeks of what has turned out to be a three-year project. During this three years, Rebellion itself has grown beyond recognition. As a point of comparison, the entire company that produced the original Alien vs Predator for the Jaguar numbered seven people. When we started work on Aliens vs Predator we were up to 18 and still had a few gaps to fill. Now we have reached double that number and we are still making space for more desks.
Despite this growth, the core team responsible for AvP has remained intact. This, I believe, has been one reason why the resulting game has been so solid.
Although it is impossible to count the exact number of people who have worked on the game both here and at Fox in the US (bearing in mind that a good deal of the testing process was carried out at Fox, as detailed in last weeks diary), I would like to give special mention to the core development team who have stuck together to bring us to where we are today.
The engine, dynamics, lighting system, vision modes and other special effects were all created by our lead programmer, Kevin Lea. Richard Rice created a custom-built set of software tools to integrating the artists work within the game engine while Chris Fox strafed endlessly to implement his ideas for the AI.
The creation of the all-important graphics and visual effects was led by the games internal producer, Alastair Halsby, who was ably assisted by lead animator Ken Turner. Additional artwork was also provided by Adam Comiskey, Matthew Riordan and Dominic Jeffreys. The level design, sound, AI way-pointing and texture-mapping was handled by the three-man team of Julian Breddy, Ed Cookson and Tim Jones.
Last but not least, the entire project has been overseen with a canny eye by Fox Interactive through producer, Dave Stalker, associate producer Chris Miller and QA lead Aaron Blean.
It struck me when I was putting that list together that, when you add the contributions of myself and my brother Chris into the equation, the combined efforts of this team accumulated quite a hefty number of man hours. To put this in perspective, if one person spent their entire working life devoted to a single project, they might just about come up with a game of the complexity of Aliens vs Predator.
And would that be a life well spent? I definitely think so. And, if the 90% review which has just been published in PC Gamer is anything to go by, then the reviewers may just feel the same. But don't take my word for it, there are now three separate demos of the game available (one featuring each of the three main characters), so you have every opportunity to try it out for yourself.
As they say in all of the best TV programmes, I hope that you enjoy playing the game as much as we have enjoyed making it. It's been a learning experience for us all, but one that we hope will bring pleasure to millions of game players around the world.
As for Rebellion? We have quite a few projects up our sleeves that you will hopefully be hearing about over the next few months, but first... some sleep.