It's hard to believe that Super Stardust HD recently celebrated its second birthday. Play the game today and it's still one of the most technically adept, brilliantly conceived and ultra-addictive shooting games available on the current generation of consoles. Where Xbox 360 has its Geometry Wars, PlayStation 3 has Stardust. Both superb, both essential.
The story of the game's genesis starts way back in 2004, when developer Housemarque had reached a corporate crossroads, having created state-of-the-art PS2 technology and several original PSP concepts just as the industry began its transition towards the next-generation platforms.
"We thought that PSP would be a great device to be working on and put in quite a lot of effort to make new game concepts to suit the device. We ported our PlayStation 2 engine and other technology infrastructure to support PSP," explains Housemarque CEO and co-founder Ilari Kuittinen. "Unfortunately, the game publishers were not interested in supporting PSP with original games based on new IPs, which were especially designed with the handheld in mind. On the PlayStation 2 front, the publishers seemed to believe in the summer of 2004 that it wasn't worthwhile to do any new original titles for PS2 either, because the launch of PlayStation 3 would happen in 2005 and the software sales would plummet immediately for the platform."
In short, Housemarque had made a series of decisions that market trends and industry happenings proved to be the right ones, but that the game publishers themselves couldn't foresee. PSP's early progress was stalled by a mentality of porting existing franchises and PS2 titles across to the handheld, while PS3 itself was delayed by a year, eventually launching in the US and Japan at the tail-end of 2006.
Kuittinen continues: "During the autumn of 2005 it seemed impossible to get any projects signed on either the PSP or the PlayStation 2, so it was even more unlikely we'd get into the next-generation console development business as it would have meant massive investments in new technology and concepts and we had depleted all our financial resources to make PS2 and PSP game concept pitches, game demos and prototypes. All this really changed when we saw Geometry Wars on XBLA in late 2005 and we were starting to see new opportunities that seemed like a perfect fit for a small developer like us."
The first of two gameplay tips videos put together by Housemarque, with hardcore gameplay from one of the world's greatest Super Stardust HD players, whose identity is revealed later on in this feature.
Initially that meant development of the game that would eventually become Golf: Tee It Up on Xbox 360, published by Activision, but the team's previous work on PSP concepts, along with its optimisation work on Guerrilla Games' high-end PSP title Killzone: Liberation meant that the team had good contacts at SCEE. At the 2006 E3, while taking a break from playing the God of War 2 demo, Housemarque co-founder and creative director Harri Tikkanen had his 'Eureka!' moment and from that point on momentum built towards a Stardust revival on Sony's new flagship platform.
"A few things needed to happen before he came up with the idea, so it certainly wasn't something that came naturally to him," says Ilari Kuittinen. "First, Harri had been going through the PlayStation 3 specifications available to us and started to realise the potential of the platform and how amazing games could be created for it. Secondly, Sony had started to contact developers like us and ask of our interest in developing downloadable PS3 games. These two things together made Harri start thinking of Super Stardust... putting the game on a sphere, having a game with hundreds of real-time objects on screen, massive visual effects and so on."
The second of Housemarque's gameplay vids, focusing on the Solo expansion pack. Higher resolution versions of these vids can be viewed through the EGTV link, with 720p60 versions available on PSN.
What is quite remarkable is just how quickly Super Stardust HD was developed - the game was complete in less than 10 months. However, in common with many Nordic developers, Housemarque has a strong technical focus and has built up a huge amount of experience in the many years it has been in business that held it in good stead for the oncoming trials.
"Obviously, lots of things have changed during the 16-17 years we have been creating games," explains Ilari Kuittinen. "We have seen many hardware platforms come and go and the games business grow tremendously. Having a £50,000 game-development deal during the Amiga days was huge, but now this barely covers costs for one developer for a year. This accumulated experience brings a lot of sophistication and professionalism into the way we make games. Instead of starting each game from scratch, we have a solid technology infrastructure, pipelines and libraries to start with. This also means that we can spend proportionally more time per game on content and gameplay than ever before, because we don't need to work on the underlying building blocks or core technology of the game."
Focused development began in August 2006, but the team didn't receive the necessary hardware for a couple more months after that. Amazingly, many of the key technical decisions made on how the game would work were based on studying the specs and paperwork provided by Sony.
"We received the kits a few months after the development had begun," explains Tikkanen. "We had already tried to guess the best possible implementation approach and were pretty much locked to it at that point. So basically, we made the most important technical decisions just by reading the documentation. A couple of weeks after we got the kits we had a version of the game running on PS3 hardware and the results were not spectacular. The game ran at maybe 15FPS with 25 per cent of the stuff on screen that we needed to make the game fun. This was as expected, but still pretty daunting. After that we began to transfer as many of the game systems as we could to the SPUs and learned how to effectively use the RSX. Finally, with some great support from Sony, we pretty much got the kind of game out we wanted without any major compromises."
With little experience using the actual hardware, there were many challenges in handling the PS3 architecture but, as Housemarque says on its own website, many things just seemed to click into place as development progressed.
"Multi-core programming had been coming for some time at that point, but we had not been doing it extensively. That was a major change," says Harri Tikkanen. "Cell is a variation of a multi-core CPU with its own strengths and peculiarities. At that point our multi-core support was limited only to places where we absolutely needed it... Our engine was in pretty good shape and we had a very talented and motivated team. Besides that, we took many lucky guesses and basically we didn't need to rework any of the major systems as we got them right the first time."
Tikkanen was also impressed with the low-level access they had to the RSX graphics core within the PlayStation 3.
"The PS3 libraries allow us to control RSX almost directly with a very low overhead," he says. "This is something we really appreciate. Basically we learned how to control RSX efficiently and made sure we could feed in enough stuff to the PPUs and SPUs to keep them busy. We made sure that we used every available RSX cycle and used them efficiently."
Optimisation of the game code came - not surprisingly - through intelligent usage of the SPU satellite processors unique to the PS3. Tantalisingly, there is the suggestion that there is far more raw horsepower available than Housemarque were able to tap during the development of Super Stardust HD.
"We had specific separate tasks such as collisions, physics and visibility using the SPUs," Tikkanen explains. "We could have taken many other tasks to SPUs as we had plenty of processing reserves still available for us, but we simply didn't have the time to do it."
Few arcade-style games get 1080p support on the PlayStation 3, but Super Stardust HD was one of them, even if a proportion of its 'Full HD' magic is achieved via the RSX's horizontal hardware scaler. Bearing in mind the combination of effects and the amount of objects on-screen, combined with the compressed development schedule, it was quite remarkable to see the game maintaining its solid 60 frames per second.
"We had kind of promised that we would have 1080p support to our producer... so we had to do it," Tikkanen says. "I also had to make sure that the game was fun, so I prioritised that aspect. Basically we completed the game mostly testing it with 720p only, and when it was really coming together and we felt that the game was starting to be fun and enjoyable we tried the 1920x1080 mode. It did not run that well. At that point, late in development, we didn't have that many options left. We optimised the last loose cycles out from pixel shaders and wrote a system that draws the objects in optimal order and we got some gains. 1920x1080 was still out of reach, but we got 1280x1080 mode working. As the image is only scaled one way it looks surprisingly good and it's still 1.5 times the amount of resolution compared to 720p."
By April 2005, very slick-looking release-quality preview code had made its way to selected European journos, including myself. The final game, improved in many ways over the playable sampler, was released in June, where I awarded it an overwhelmingly positive 9/10. In the lifecycle of a typical game that would most likely be the end of the story, but in a sense the Super Stardust HD phenomenon had only just begun.
Housemarque's debut PS3 project is a game that offered plenty of 'firsts' during its history. It was the first PSN game to be launched simultaneously across the globe, while the core codebase itself was regularly updated - a process that made it the first PS3 game to support Trophies. More than that, similar to Burnout Paradise, the game evolved through the use of additional downloadable content.
"I'm pretty sure that the DLC idea came from our producer Phil Gaskell," says Harri Tikkanen. "I liked it, as we could add new game modes quite rapidly. Making a sequel would have taken much longer. As we were developing the DLC it was pretty straightforward to add the Trophy support as we had the team in place."
"The theory was that adding more content to a successful game would give an expanded experience to the fans of the game, having more exposure, and extending the sales curve of the game," adds Kuittinen. "We also had a stroke of luck when our game happened to be the first and only game to feature Trophies when the system update launched, so this helped a lot. All this has worked out really well and is still proving to us that the long tail of sales is possible in our case. Over 300,000 unique gamers have posted a high score in our global high score list. On top of that, over one third of those players have also played one of the expansion packs as well."
New solo modes and also split-screen were added to Super Stardust HD as it evolved, and while the codebase was improved, a lot of the changes were cunning re-implementations of the existing tech.
"We did some optimisations but mostly the performance was already there," says Harri Tikkanen. "With the new game modes we used more small, single-piece asteroids. In the original Arcade mode we had huge asteroids which actually consisted of hundreds of pieces. We wanted the player to be able to shoot and split them in any way they like. So when we used those hundreds of pieces as separate objects it really looked pretty impressive. Actually the biggest change was the level editor which allowed us to spawn large numbers of enemies in cool patterns easily."
Split-screen was also added in a separate expansion pack, which allowed players to move to completely different areas of the sphere to independently continue their blasting. If there's a reason we don't see such multiplayer modes that much during this console generation, it's because the processing overheads involved multiply dramatically.
"Retrofitting a split-screen mode wasn't easy and maybe we shouldn't have tried it," muses Tikkanen. "To keep the development cycle as short as possible we could not fit the time in to do it originally - yet I always felt that it would be the right way to do the co-op mode. When we got it running we realised that we were more vertex-bound that we had expected and basically we had to process twice the amount of vertices in the split-screen mode. So with the current content we could not make the game run 60FPS without major engine changes. Also at that time we had already moved to develop the next major revision of our engine so we felt that it was not efficient to go back to older revisions and make major changes."
Housemarque also kept in touch with the community playing its game, and singled out ace Finnish hyper-gamer Tlo-Mek to visit their studio and show them how he played as they finalised the expansion packs.
"It was absolutely unbelievable to watch this guy play the game," remembers Kuittinen. "He aimed to maximise the score in every possible occasion and he had figured out all the possible ways to do so... we invited him over to test the new game modes for the DLC and we learned many things, like the modes that were meant to be short needed to get insanely hard as he was playing them without dying even at difficulty levels we had presumed no one could survive in!"
The team still keep a close eye on the Super Stardust high-score tables, and are staggered at the gameplay skill of a hardcore element of the userbase.
"We originally thought that breaking 500 million points in the main arcade mode would be nearly impossible," says Kuittinen. "However, after only a couple of weeks we had a leading score of over one billion and we were totally astonished, but today the top score is even more unbelievable as it is close to 1.75 billion played by a gamer with an alias 'Bridy'. Even today, breaking the 500-million score barrier is so tough that only 110 players out of over 322,000 have done that. This told us that there are huge differences in gaming abilities between players, and how hard it is to develop a game that offers truly hardcore challenges to the elite gamers without forgetting about the people that have reflexes and the hand-eye coordination of a normal person, rather than a super-human."
Moving forward, the Housemarque team is eyeing up the resurgent portable gaming sector and has a lot of heritage in that department, having previously produced several games for the likes of N-Gage, the Palm OS-based Zodiac and even the Gizmondo (!). However, the team is perhaps best known for its well-received port of Super Stardust on PSP.
"Scheduling-wise it seemed that we would not have time to reinvent the whole experience for PSP, so we went for a modified port and added a new 'Impact' game mode," says Tikkanen. "We kept all the levels, enemies, bosses and tuned them to fit the small screen. I feel that the experience is as close it can be and it's even running at 60FPS."
With the PSP seemingly on the cusp of a comeback with new hardcore and triple-A titles such as Gran Turismo and Metal Gear Solid: Peace Walker, Housemarque is also keen on returning to the fray and has plenty of praise for the hardware.
"We believed that the PSP has great potential even before the platform was launched," says Ilari Kuittinen. "It is highly likely that we are going to be developing something for this platform, but I really don't have anything to announce at this point."
"PSP is a solid piece of hardware, and it's easy to develop for," adds Tikkanen. "I'm still amazed how fast the rasteriser is for a portable device - it can fill the screen tens of times over at 60 frames-per-second."
In terms of the mobile phone market, the team has an interesting outlook on game-making for handsets and is intrigued the potential of the iPhone and iPod Touch.
"We have a history with mobile games as we founded a spin-off company specialising in mobile game development with the help of investment money back in 2000," says Kuittinen. "We found out that mobile games markets were too fragmented and under-developed in order to make interesting games, or to sustain a viable business for a developer like us. The value of the development side of things was in making the game work with hundreds of different phone configurations rather than concentrating on making a great game that would look and play well. The iPhone is a solution to many of the problems of the industry as it is a single platform combined with efficient distribution and a viable business model for developers as well, although the App Store's open-door policy has its own problems. It is certainly something we are taking a closer look at and hopefully we'll have creative opportunities that make sense for us in the future."
The emphasis in the here and now however is on something that PS3 owners should be getting very excited about...
"We are currently concentrating on new game projects and I am happy to say that we will be announcing something at gamescom in Cologne," revealed Kuittinen. "All I can say right now is that it is a PS3 PSN exclusive title we have been working on quite a while now, and yes... shooting is an integral part of the game..."
