The Rise of Middleware 2.0

The chronology of the loose collection of tools and technology labelled games middleware is remarkably straightforward.

Kickstarted by the complexity of PS2, the use of Criterion’s RenderWare in the development of the multi-million selling Grand Theft Auto III in 2001 finally ensured the topic switched from a pub discussion between junior programmers to an agenda point for company directors.

Middleware was suddenly everywhere, with Criterion CEO David Lau-Kee claiming in 2004 that between a fifth and a quarter of published games used its technology.

That was the high point, however. EA’s acquisition of the company eventually resulted in RenderWare’s withdrawal from the market, and a reassessment of the wider commercial risks of using thirdparty smarts.

“The acquisition of Criterion was a truly industry-shaking event,” reckons Felix Roeken, general manager of German company Trinigy, whose Vision engine competed with RenderWare.

“People’s trust in middleware was severely shaken. Some of them decided to fill the gap by developing in-house technologies but some, particularly publishers, took the chance, reviewed their internal processes and went on scouting for new technologies.”

Emergent sense of scale
So two years on from that extinction point, middleware has begun to flourish again, evolving into its new 2.0 guise.

The technology and the business models have changed, however, and few companies now claim to offer a complete solution to developers’ needs. Instead, like small mammals running through the bones of their giant ancestors, available technology has become flexible, lightweight, and highly componentised.

Two of the classic examples of this new approach are Emergent Game Technologies and Freescale.

Both veterans in terms of providing game tech, Emergent’s core product was Gamebryo, a renderer originally launched as NetImmerse in the late ‘90s. Since then it’s been continually overhauled, while provider NDL has become part of Emergent, itself a game tech start-up hit by the dotcom crash and since resurrected.

Freescale’s CodeWarrior programming environment has experienced similar convolution. Originally part of Motorola under the Metrowerks division, it was spun out with the Freescale IPO of 2003, and now is housed within Freescale’s tongue-twisting Game Technology Organization moniker.

What’s interesting in these cases is that longevity in the market, despite corporate chaos, has provided an opportunity to hone products which fulfil the market’s next-gen needs. Perhaps best demonstrated in the case of CodeWarrior, this integrated programming environment has recently been relaunched as CodeWarrior Radix, a lightweight cross-platform coding framework, which enables developers to link into any other tools thanks to hot-swappable plug-ins and an extendible SDK.

“I think this generation of consoles can be summed up by the phrase ‘tools matter’,” says Freescale’s Roger Edgar. “Being cross-console will be a key to winning the hearts of developers. If you make the developer’s life easier and you can reduce the cost of production you will go a long way.”

The situation with Emergent is even more characteristic. Its Emergent Elements package offers a wide range of components, from Gamebryo Element, and the various smaller packages it’s integrated with, to online gaming hosting infrastructure, build tools (Automation Element), and development and gameplay metrics services (Metrics Element).

“Emergent Elements is Middleware 2.0,” states vice president of engineering Larry Mellon. “If that sounds like marketing speak, consider this scenario: A development studio uses a game engine to create the game. That engine is tightly integrated with advanced tools for extracting, displaying and analysing in-game data, which feeds development decisions.

Every night, automation tools create the build and generate a detailed report on exactly what went right, what went wrong, and why, allowing for iteration times that are a fraction of what they are today. All of these tools work in sync with a next generation server engine for online games. They’re modular, so you only use what you need, and they work seamlessly with other commercial and homegrown tech. For us, that’s middleware 2.0.”

Acting on Instinct
Even taking into account the company-centric view expressed by Mellon, it’s clear middleware is now a much more slippery concept to pin down. Ironically, it’s more fragmented and yet for that reason also more connectable. One metaphor used by many is the idea of the industry sitting down to a smörgåsbord of technology.

One start-up that’s relying on this approach is Irish provider Instinct. It’s currently beta-testing its Instinct Studio technology. “Our ultimate goal is to provide a middleware platform combined with best of breed solutions for components such as physics, networking and AI,” explains CEO Frank Gallagher. “It’s like a Sky TV package where you can mix-and-match what you need, providing a tailor-made solution.”

One reason such a situation has arisen is the work carried out to provide the various types of glue which underpin the entire edifice. Fundamental building blocks such as programming languages like C++ aside, the two that have had the most impact are universal file exchange format FBX and the Collada. Both have experienced prolonged gestations.

Originally created by Canadian animation company Kaydara to act as a file format for linking motion capture data with hand animation, FBX has grown in importance as Kaydara was first acquired by Alias, which used FBX as a connector between Maya and Kaydara’s MotionBuilder. When, in turn, Alias was bought by Autodesk, FBX became the ideal channel with which to link Alias’ products to Autodesk’s 3ds Max.

And synergetically, as FBX has become more important to its various owners, so it’s become vital for third-party tools providers to link into. It’s now supported by over 30 companies including Softimage, NXN, Natural Motion and Maxon.

Plugging into the pipeline
Also contributing to the virtuous fragmentation and reconnection of the art production pipeline is the COLLAborative Design Activity or Collada. Initiated by Sony to ease the burden of PlayStation 3 and PSP production, but since taken over by the Khronos Group (the industry consortium also overseeing the OpenGL specification), it’s an open standard XML-based schema for the exchange of digital assets. These range from art assets, such as geometry and animation, to physics and shader data. Microsoft’s XNA technology performs a similar role, albeit it in a deeper, proprietary manner.

The result, of these and other technologies, is a more open playing field within the boundaries of which middleware providers can focus on their core area of expertise, relatively safe in the assumption clients will be able to fit the technology into their production processes.

The classic example is IDV’s SpeedTree. This tree and plant creation technology, available as a plug-in and a runtime, is being used by over 70 companies, from large publishers to tiny start-ups, because it only costs $10,000 per project per platform and it works extremely well out the box. For that price, if your game requires trees, why wouldn’t you use it? Its success has also generated a rash of possibly apocryphal startups attempting to do for water simulation, sky and cloud simulation or terrain simulation (SpeedRock, perhaps?), what SpeedTree has done for foliage.

Companies providing archetypal middleware such as networking, artificial intelligence and sound components are also using this philosophy to great effect as middleware gets its second wind.

“I think audio middleware technology is a niche as it’s something a lot of developers don’t really want to have to think about. To develop technology as powerful as what we can provide, they’re going to have to spend a lot of their own resources to do it,” says Brett Paterson, CEO and lead programmer of Australian tools company Firelight Technologies. In a similar manner to SpeedTree, its FMOD Ex audio engine is available for an affordable $6,000 per title for the first platform with an incremental payment of $3,000 for additional SKUs. Small wonder that it’s been used by over a hundred companies in titles ranging from World of Warcraft to mobile phone games.

Modular technology
Even in the more constrained case of networking technology, a small number of proprietary standards means technology providers, and developers, can be confident of a workable out-of-box solution.

“It’s a case of ‘drop it in and it works’, while still maintaining flexibility,” says Mike Drummelsmith, developer relations manager for Canadian provider Quazal. Its product line consists of two extremes of this attitude. On one level, its Spark! technology provides a closed-box solution with a defined featureset that can be integrated into a game in a couple of days, whereas its full Rendez-Vous product takes a sandbox approach to let developers build custom features.

“Now, we’re looking to build on the middle ground, with the flexibility Rendez-Vous affords, while maintaining the ease of use of Spark!,” Drummelsmith says.

But while, at this level, the middleware dilemma is becoming simpler for developers, providers, especially those who sell boxed product, find themselves locked into an increasingly commoditised business model. As their technology becomes more robust and simpler to use, it’s harder to increase your prices.

And it’s some companies’ reaction to this – seen in a movement to more complex pricing such as revenue share licensing – which is another marker of Middleware 2.0.

The classic example royalty-based licensing is Epic’s Unreal engine, which as a complete cross-platform game engine and production environment, has always been the something of the exception that proved the rule. For example, despite having a list price of $350,000, Epic’s previous generation technology, Unreal 2, also came with a three per cent royalty based on a title’s wholesale price.

Few middleware companies wield this type of commercial muscle, though. Even fewer clients want to sign such deals.

“In our experience, most serious developers and publishers prefer a flat-fee based licensing deal, which provides them with a secure basis for their financial planning and makes sure their success is truly theirs,” reckons Trinigy’s Felix Roeken.

Chris Doran, CEO of UK-intellectual property startup Geomerics also points out there are good reasons why such deals aren’t ideal for small middleware companies anyhow.

“If you’re Epic, striking royalty-based deals make complete sense. But if you are a new arrival, a deal that only leads to revenue in two years time, combined with uncertainty about the amount of revenue, is far less attractive,” he says. “Royalty-based deals are the ones that can net the most profit, but they are a gamble and you need to be a certain size before you can afford to play this game.”

Modular business
For certain type of middleware technology and services though, more flexible deals can be cut.

For example, facility houses such as motion capture studios have always worked on a day-rate, but as the services they offer become more sophisticated, so they become more like production consultants. Examples include the new Contour capture system from San Francisco’s Mova facility or Artem Digital’s 4D live action scanning.

Emergent’s Larry Mellon reckons properly constructed royalty models can help kickstart emerging development talent. “Royalty-based licensing deals make particular sense in the casual game space, where initial development budgets are lower. We’ve just introduced a royalty-based licence for Xbox Live Arcade, which will bring the capabilities of Gamebryo to teams that are building games on smaller development budgets.” Clearly the company is also looking to royalties or revenue splits when it launches its online hosting services.

But Torsten Reil of UK animation specialist NaturalMotion, which has a co-development deal with LucasArts, warns such deals are only viable if they are advantageous to both parties. “Obviously, publishers will only accept a revenue share model if it benefits their bottom line, in which case the technology needs to create a new game experience capable of shifting more copies.”

And, in turn, this highlights the final trend which makes up Middleware 2.0; widespread publisher acceptance of middleware in an attempt to standardise development and reduce risk.

RENDERWHERE?
Ironically, it was EA’s decision to buy RenderWare that was the genesis of the entire problem. But proving what EA did yesterday, everyone else will try tomorrow, the past 12 months have seen an avalanche of press releases which highlight multi-title, multi-site middleware deals.

Once again, Epic leads the way thanks both to the strength of its technology and a licensing model with a tapered backend the more titles encompassed, but players such as Engenuity, Trinigy, and Quazal have done similar deals.

“Sound business practice, as well as the increase in next-gen complexity, means that there’s a growing desire among publishers to leverage technical economies of scale across their studios, while still preserving creativity and innovation at individual studio level, says David Coghlan, vice president of development for Havok, which has deals with most top publishers. “Increasingly, developers and publishers are developing technical strategies that allow them to manage and plan for their whole pipeline, not just one game on one platform.”

Of course, in this respect, Middleware 2.0 is much the same as what went before it. However we labelled it, fundamentally middleware is just an opportunity for developers to get a powerup in their eternal battle within the time, quality, cost triangle. But for middleware providers, everything has changed. The technology itself is no longer enough. To be successful, they now have to be as innovative in terms of how they deploy their smarts and how they charge for them too. But maybe that will just be the next big thing – middleware for middleware companies anyone?

■ www.developmag.com