EA on iPhone game development: the accelerometer's no good for precision

IMG_4036.JPG

Those who have played Super Monkey Ball on the iPhone basically all agree that its a cuspid-clamping affair. The merest tilt is enough to send your bubble-contained simian flying into an existentially terrifying purple vapor, Super Monkey Ball's metaphor for chimpanzee death.

So what's the deal? Why is the iPhone's accelerometer so crap at steering? An EA developer put it this way: "Think of it as a loose analog stick...you get lots of random data."

As developers have more time to figure out their various iPhone accelerometer smoothing algorithms, controls should get better (just as they did after the Wii's release), but using the accelerometer to steer or control an on-screen character is just never going to be as responsive or exact as a D-pad.

Which is fine. I think we all knew that: the iPhone is not magic. But I suspect we're going to see a gaming culture on the iPhone based more primarily around multi-touch. It's notable that the iPhone port of Star Wars: The Force Unleashed is using the accelerometer only as a means to move your character to set waypoints, where you then fight in a static position using touch gestures. Very on-rails.

It'll be interesting to see how developers approach the iPhone when they have more than a few months to come up with an app: on the surface of things, the iPhone has both the advantages of a Wii and a Nintendo DS as a controller, but the roux is very watery, and there's all sorts of crap floating in it.

The iPhone is more powerful than the DS but sucks as a controller [Gizmodo]


Discussion

Take a look at this

I've played a few accelerometer games on a jailbroken iPod Touch (1.1.4 firmware), and used an accelerometer graphing / logging program on the same device (in the pursuit of debugging an application I consulted on). The accelerometer seemed to generate very little noise while at rest or when moved by mechanical means to test drift on that particular device, and the response in the games tended to be accurate to what I intended, so much so in one game that it would pick up the subtlest movements of my hands, and I was unable to successfully play the game if I had recently consumed caffeine.

Perhaps it's not that there's noise in the accelerometers. Perhaps it's that there's noise in the hands of the person playing. D-pads de-bounce signals from their contacts, insulating players from the consequences of inconsistently applying pressure to a contact. Accelerometer game design on the iPhone is still in it's infancy, and it will take a while for people to adjust to games that do not de-bounce their cafffeine jitters, and for others to build games that debounce twitchy player inputs when appropriate.

Take a look at this

Yeah, in my anecdote-is-not-the-singular-of-data experience I've found that the accelerometers work just fine. In Super Monkey Ball in particular, any quirks with motion seem to be more of the result of Sega's decision to rotate the camera around the avatar at all points instead of leaving it locked behind, forcing you to continually adjust not just to steer, but to account for the camera. I get why they did it, but it'd be nice if they'd have given a 1:1 board-to-tilt option. And perhaps even a sensitivity slider like some other third-party games offer.

Take a look at this
#3 posted by Anonymous , July 18, 2008 1:19 AM

(Posted anonymously for NDA reasons...)

The raw accelerometer output is very noisy on iPhone. You can set the device flat on its back on a good solid table, and the vector coords still jitter like crazy. This makes me doubt that Bardfinn's graphing program was showing raw output: it was probably filtering pretty heavily.

The thing is, Super Monkey Ball will also be filtering heavily, as will any accelerometer-based game. There is a payoff between responsiveness and smoothness, at least with the dumb low-pass algorithm used in Apple's sample apps, but this is more a problem for programmers than for players. If raw data noise was actually showing up in-game, that would be a surprisingly poor piece of coding.

Why the plunging monkeys then? I've not tried the iPhone port of SMB but I have played the GC and Wii versions quite a bit, and Joel's comment sounds spot on. There's an inherent gimbal-lock problem with the camera physics when manoeuvring at low speed. Shame, because it's a damn fine game otherwise.

Take a look at this

"the iPhone is not magic"

you take that back!

Take a look at this
#5 posted by Anonymous , July 19, 2008 7:42 AM

I've developed games for EA Mobile's publishing arm and I found their attitude appalling. Rife with ego, stupidity, lack of respect and ignorance all hidden behind pathetic dominance games.

I'd always assumed that their school-bully attitude was reserved for those lower on the food chain.

It makes me feel both better and worse to see that they are happy to direct it upwards as well.

Maybe they're losing the plot as the death of the traditional game publishing model breathes down their neck.

Take a look at this
#6 posted by Anonymous , July 23, 2008 6:28 AM


Another developer point of view on iPhone game dev.
Can be interresting...
The iPhone a good platform for game developer ?

Post a comment

Anonymous