Develop 2008: Ideaworks3D on native mobile gaming
Tim Closs from Ideaworks3D had a session at today's Develop Mobile conference, covering the opportunity for native mobile games.
He started by outlining the advantages of native games, including performance 2-10 times J2ME games, less embedded software to work around, higher price points on some platforms, and commonalities with handheld and console platforms, allowing developers to reuse tools, art assets and code assets.
"This isn't about porting your game from console down to mobile, but you can definitely get cost efficiencies by reusing some of those assets," he said.
Closs also thinks native games have a longer shelf-life, particularly if you're adding incremental content or versions - "this is stuff people don't do with Java games".
He outlined the currently addressable native mobile platforms, including BREW, Symbian, Windows Mobile, N-Gage, iPhone and iPod Touch, WIPI in Korea, and Linux-based smartphones, and also possibly Android.
Can games automatically hit all these handsets? "Theoretically yes, but you need a usable distribution mechanism," said Closs, saying that BREW paved the way, and the launch of the App Store and N-Gage have been positive developments in this area.
Closs then looked at the West and the East separately. So the West has BREW with US operator Verizon and a few other operators. Then N-Gage - something Ideaworks3D has been doing plenty of work for - iPhone and iPod Touch, and then the handful of other platforms, including Symbian and Windows Mobile.
Closs then took each platform in turn, starting with BREW. Its strengths are it's a mature platform with a strong subscriber base, and there's no localisation required as it's one operator in the US.
However, Closs said challenges include Verizon wanting developers to hit nearly all handsets, meaning a limit of 2MB per install, and 5.2MB of heap memory.
There's a fair amount of handset variation and bugs, according to Closs, who also pointed out that the current gen BREW chipsets support limited hardware graphics acceleration, but at the expense of audio and other features.
What are reasonable sales for a good-selling BREW title? Closs says Guitar Hero III Mobile has sold more than a million downloads, although he speculates that many of those are incremental content rather than the game itself.
Ideaworks3D released two BREW games in 2005-06 - console franchises (i.e. The Sims and Need for Speed: Underground) - which sold more than 100,000 downloads each, at $12 per game.
"50,000 sales is still a good result on Verizon, and given that the publisher cut is around 60 per cent, you're talking about $360,000 of revenue from a good-selling BREW title," he said, while stressing that good deck placement and promotion is a necessity.
Is this a good return on investment though? Closs posed the question, but was careful not to answer it.
Next N-Gage. Strengths are its global reach, with billing in place with 83 operators in 23 countries, and credit card purchases being made in more than 200 countries.
He points out that the key metric is the number of handsets with the N-Gage client pre-installed. "Even though you can today download this client for an old N-Series handset, it's not the greatest experience in the world, so the key metric is how many handsets are they going to get with the client pre-installed," he said.
According to Closs's estimates, Nokia will have the N-Gage client pre-installed on 50 million handsets by the end of this year.
Meanwhile, he said there are 12 target handsets today - is that more than Nokia's officially announced roadmap? - and that there's a focus on quality, with a decent file size limit of 64MB per game.
"If anything, the game size restriction is crazily high - maybe 64MB will make sense in a couple of years, but today it feels unlimited," said Closs.
Criticisms? "The SDK is quite thin, it relies on the developers to do all their own rendering, animation and tools," he said. The N-Gage SDK roadmap will have OpenGL ES support by December, apparently - hardware accelerated games are coming our way in 2009, in other words.
Closs said that there's a heavy DRM and associated user interface requirements that take around two man-weeks to handle. And he said the same for Nokia's connected features requirement.
This isn't necessarily bad, but Closs's point is that you have to do this work specifically for N-Gage. "But it's shaping up to be a pretty fantastic platform we think," he said.
Next, Closs came up with a revenue model for N-Gage based on "some pretty informed numbers" - how much money can you make from an N-Gage game?
His assumptions - you're launching your game at the end of this year, and are getting a two-year revenue cycle from it. He's using real figures for current installed devices and Nokia's own estimates - but only counting users who'll have the client on their device.
He's considering titles as a percentage of the deck size - it's 20 now, but Closs says it'll be 30 by the end of the year. And in his calculations, he presented a worst case scenario and best-case scenario.
Worst case is two per cent of N-Gage users buying one game a month, while the best case is six per cent buying a game a month. And he's assuming that the price point is nine euros, and that the operator and Nokia will take 30 per cent of the revenues.
So, the worst case for an N-Gage game is 800,000 Euros, and a good case is five million Euros over two years. "Those look like good figures compared to what you're going to do on Verizon, but that's because Verizon is one operator, while Nokia is rolling this out globally.
Apologies for writing that in full, but it's fascinating and informed speculation, so I wanted to get it all in.
Next iPhone, which Closs skated over as it's been talked about throughout the day. Advantages obviously include no porting issues with just two devices, and "it's a nippy bit of hardware" with its ARM11 and GPU, plus the touchscreen and accelerometer.
He also praised the iPhone SDK as "definitely the best native SDK out there" - something admitted by even Nokia earlier today.
Challenges include how to stand out with 900 apps on the App Store already, 30 per cent of which are games. You have to develop games on Macs, and it's not C++ - it's a variant called Objective-C. And also the lack of any buttons on the device can prove challenging for some game types.
Speculative revenues? Closs didn't serve up a figure, but compared his metrics with N-Gage, saying that iPhone may have a longer revenue tail - Apple probably won't kick old stuff off the App Store.
Meanwhile, the average analyst forecast predicts 35-40 million iPhones sold by July next year, while iPod Touches are selling like the clappers too.
Closs pointed out that some free apps could cannibalise paid downloads, and put pressure on price points from their current norm of £6. But Apple is taking 30 per cent - similar to N-Gage.
"It's impossible to forecast how much an iPhone game could generate, but if you can get an app in the Top 10 or Top 20 on the App Store, you could potentially make a lot of money," he said.
Closs dealt with Windows Mobile, Symbian and Linux, pointing out that it's hard to distribute native games for these platforms - certainly not on operator decks anyway.
Operators aren't selling native games well either, said Closs. They either refuse to sell them, or they don't differentiate them enough from standard Java games - "We've seen the J2ME game and its native version sold alongside each other with the same name, but one's £4 and one's £7," he said. "Guess which one consumers bought..."
"If you can deploy to these platforms with little cost, it's worth it, but otherwise I don't think it's particularly worth pursuing at the moment," he concluded.
Next, Closs looked at how developers can target the three main Western platforms - BREW, N-Gage and iPhone. His first tip was to think globally from the start, citing the (bad) example of a game created in Japan as a Java game for DoCoMo, then the publisher's US office create the same game from scratch for BREW and Windows Mobile, and the European office then builds the game from scratch for N-Gage. And then the US and European offices wonder what to do about iPhone.
"The thinking isn't at all joined up," says Closs. "Today, there's a lot of inefficiencies going on there for sure."
Closs's advice is to think ahead about a cross-platform global release in advance, and plan the game design accordingly - "publishers could have saved several hundred thousand dollars by doing this," he said.
So how do you target these platforms design-wise? The game design has to be scalable across form factors, including touchscreen, portrait and landscape.
It has to be scalable across install sizes as well, so if your game on N-Gage and iPhone will be a 10MB game, you don't have to write BREW off, but just figure out how to make the initial install 2MB, and then bring other levels down over the air.
Closs also said developers need a scalable asset pipeline, across software and hardware graphics processing, and various GPU variants. And games have to work across different screen sizes.
Finally, games need a scalable and portable technology - across operating systems like BREW, N-Gage and iPhone, but also across CPUs (ARM9 150MHz up to ARM11 300MHz plus a GPU).
At this point, Closs held up his hands and plugged Ideaworks3D's own AirPlay SDK as a possible solution to this.
Closs then showed off an example 3D beat-'em-up game running on an iPhone, N-Gage and BREW handset - it's one of Ideaworks3D's demos for AirPlay, apparently.
Then Closs looked at the East, starting with BREW in Japan on KDDI, which has 30 million subscribers paying an average of £10 for data a month. Oh, and a whopping great deck with more than 4,000 applications, but lower price points - £0.50 up to about £10, but skewing below the £2 mark in general.
The handsets are quite advanced though, but the install size limit is 1.5MB, although you can download a total of 10MB for the whole game, which can be all on the phone at the same time.
iPhone is available on Softbank in Japan, but in small numbers today. While the third native platform highlighted by Closs was WIPI in Korea, but he admitted he hasn't really dealt with it much, since the Korean mobile games market is dominated by local developers and publishers.
Closs pointed out that Java handsets on NTT DoCoMo and Softbank support the JSR 239-style OpenGL ES 1.1 API, with hardware graphics acceleration, so developers making iPhone and N-Gage titles can target these phones too, even though they're Java.
Finally, Closs talked about the commonalities between mobile and console/handheld development. The traditional 'develop for J2ME and port to BREW' development path won't work, nor will just porting the BREW version to iPhone and N-Gage.
"At the minimum you need to do some work to up-res the game slightly," said Closs, who argued that developing for these native mobile platforms is more similar to developing for handheld and console - PS2, PSP and even Xbox Live Arcade and Wii.
"I'm not talking about porting the game, but getting some re-use of art assets," he stressed. But developers should still keep their polygon counts low - under 1,000 for software rendering.
Next, Closs said that Ideaworks3D thinks native platforms could be "really cool" for massively multiplayer online games, in a way that Java couldn't be. The company has put a demo together to show clients how this might work.
MMO publishers have shied away from this, because it might not be a level playing ground, "but if you're looking at more social world based opportunitie, like Second Life and Sony's Home, that doesn't apply."