Designer Postmortem: Porting Marble Blast from XBLA to iPhone

Six months ago I was subcontracted as an independent studio, Red Thumb Games, to design Marble Blast Mobile, an iPhone version of the XBLA hit game Marble Blast Ultra.  South African studio Luma Arcade was handling the port for GarageGames and they needed a designer.  As it turns out, having previously been a producer at GarageGames for two years, I was already well familiar with the Marble Blast franchise, so it was a perfect fit.  I tackled the project in earnest.

HISTORY

Above: The original Marble Blast

The Marble Blast franchise has a long and colorful history, having begun as a PC, Mac, and Linux downloadable try-and-buy game in 2002, created in part as a proof of concept that the Torque Game Engine was capable of more than just FPS games from which the engine originated.  The game was upgraded with more levels in 2003, bundled with every new iMac G5 in 2004, was one of the few Xbox Live games for the original Xbox in 2005, was majorly updated as a launch title for XBLA in 2006 for the XB360, and was a launch title for the browser based game platform InstantAction in 2008. 

Above: Marble Blast versions

Over the years the game has built up a huge fan base and passionate community, and hundreds of user generated levels exist for the game.  That's a lot of legacy to live up to.  As anyone who's worked with licensed IP will tell you, it's always harder to make a game within the constraints of fitting in with that IP than to simply make a brand new game with no license.  This wasn't a licensed game, but the pressure was the same.

MAKING THE JUMP

The first challenge then was to figure out what kind of game we could make that would please old fans, bring new fans in, and keep the IP holder (GarageGames) happy as well.  The first step towards this was making the game look as great as it possibly could.  Marble Blast Ultra for XBLA is the most graphically advanced version, and one of the goals for the project was to show off what Torque for the iPhone was capable of, so we had to use the XBLA game as the starting point.

Above: Marble Blast Ultra for XBLA

However, the XBLA version has some of the most complex controls as well.  As if the original game's marble movement, independent camera movement, jumping, and item use weren't complex enough, the XBLA version added a "blast" feature that can give you an extra boost in jumps and can push players away in multiplayer.  The challenge was to simplify the complex controls as much as possible while still retaining maximum gameplay in order to use the XBLA version as our basis.

SIMPLIFYING THE DESIGN

In order to simplify the controls and game design, the first thing I did was cut the blast feature.  Jumping, item use, and camera control were already a lot to manipulate on the small no-button device, and I wanted the game to be less hardcore than the Xbox version, as the iPod platform has a much broader demographic than the more hardcore leaning Xbox players.  As anecdote, my sister-in-law has an iPod and might play this game, whereas she doesn't have an Xbox and is too intimidated by a game with so many buttons to play the Xbox version.

However, cutting the blast feature presented a significant problem.  Levels designs for the XBLA game were based on the ability to blast, allowing you to "double jump" your way to areas that could not otherwise be accessed.  Moreover, as the game evolved from the PC to XBLA, the game became even more hardcore, requiring even greater control precision and puzzle solving to complete levels. 

Control ability on the highly analog iPod would be mostly accurate at best, but never as accurate as a dedicated game controller with digital and physical buttons, making the XBLA levels a poor fit for the device.  And levels requiring ever-greater depths of understanding of the underlying game mechanics in order to complete, puzzle-style, was a poor fit for the more broad and casual audience we were trying to capture with the device.

MERGING OLD AND NEW

The dilemma was solved by making the bold decision to base the game on the older PC version which excluded the blast feature and had levels compatible with simply rolling, jumping, and using items.  Since a delivery requirement was that it showed off the best graphics possible, we then upgraded the graphics using assets from the XBLA version.

Above: Marble Blast PC

Above: The same Marble Blast PC using the XBLA version's art assets

This was no simple drag-and-drop conversion and required meticulous work on the part of Luma Arcade's artists to down-res textures and polygon counts for models, and in some cases create brand-new models where the old ones, for reasons of transparency, animation, or shaders, simply didn't work.  Level geometry also had to be carefully converted, imported, and checked for errors.

Further, all particle effects had to be cataloged and reverse-engineered from the XBLA version and re-introduced into this upgraded PC version, including subtle effects such as a shockwave that appears on the ground in proportion to the intensity of the marble landing, and a dust poof that appears on walls when the marble strikes it.

Above: Two particle effects from the XBLA version to be recreated on the iPhone

GAME DESIGN CUTS

In addition to conversions, cuts had to be made.  The original PC game actually had more powerups than its XBLA sequel, but these powerups were cut from our version to retain the XBLA feel.  The cut items were easily replaced by other powerups that were stronger and gave the player a better sense of control and power.  For example, the "Super Bounce" allowed players to drop a great height then bounce back up, but this was a passive not active player control.  Instead, the "Gyrocopter" allows the player to fly and gives them an active role in crossing a gap, making them feel more powerful as a result.

The PC version was not the only version to have cuts as we cut things from the XBLA version as well.  For example, the XBLA version featured save points, which is fine for longer, more complex levels on a console system.  But a mobile device is played in shorter bursts of play, so removing save points lent itself more to fast-paced, quickly played levels.

The XBLA version also had leaderboards and easter eggs, which were cut to keep the product scope and development time down and simplify level design.  Since we were using the PC version for single player levels, we would have to find clever places to hide all the eggs, and completely code in the feature from scratch, which was nonessential for this first version.

GAME AUDIO

Game audio also had to be cataloged and mapped to upgrade the PC version's effects to the XBLA version.  A unique consideration for the iPhone device is that many players will be playing the game with no audio.  This is common for cell phone games, where the player is playing in a public place and wishes to play silently, but not always considered for the iPod, which is thought of as a music device natively.  Further, first generation iPod Touches have no built-in speaker, further underlining this consideration as players may not always carry headphones with their device.

To solve this, we added visual cues to duplicate crucial in-game audio events.  Only an audio cue is played when the player picks up a gem, so we added a visual flashing of the gem collection number on the HUD when a gem is picked up, to eliminate soundless players from asking "Did I get that gem or not?" after rolling near it.  We also added a flashing of the in-game clock to show when a "time freeze" item is picked up, which otherwise plays a tick-tock sound during the power-up's duration.  Lastly, we also flash a new item on the HUD when it is picked up in lieu of the XBLA's voice work to signal the same.

LEVEL DESIGN

Level selection was a highly considered process.  What levels would be used in the game would contribute enormously to the game's end feel, as much or moreso as the game's control scheme, graphics, and physics.  Our feature requirements were to have a minimum of 10 single player levels and 10 multiplayer levels.  Since the original Marble Blast had no multiplayer feature, we had to select the multiplayer levels from the XBLA version, which wasn't a problem as none of them require the blast feature to complete. 

Since the game would be played mostly by single players, it seemed inefficient to have to dedicate half the maps and half our time on multiplayer levels.  However, there was no reason the multiplayer maps could not be used in the single player game, effectively doubling the number of single player maps.  This is just what we did.  Moreover, we wove the multiplayer maps in seamlessly with the single player maps such that players would not even know which one was which.  This was a smart strategy for maximizing game assets for greatest in-game use, especially since the original PC version had over 100 levels, so maximizing level quantity while minimizing development time was crucial.

Above: The original Marble Blast prototype for PC

In selecting levels, I went back to the original Marble Blast to analyze its level design philosophy, playing through the entire game multiple times.  When the game was first developed, the designers experimented with the marble rolling on terrain, then found that rolling the marble on platforms suspended high in the sky created a sense of excitement as players race through levels with the constant risk of falling off.  So the concept of rolling on platforms was key to level selection.  However, with the relatively clumsy input of the iPod compared to using a controller, we couldn't have platforms that were too narrow, or users would constantly fall off and be frustrated.  Since we didn't have development time to create all new iPod exclusive levels, I had to find levels that would work well on the device.

Above: A cool yet "gimmicky" Marble Blast PC level based on the concept of a pinball machine

The single player levels were selected to be neither too short and easy, nor too complex and frustrating, with most levels taken from the "intermediate" difficulty category.  Since the level quantity was limited, no single player levels rely on "gimmicks" such as dropping a marble a huge height and avoiding obstacles as you drop, which would be fun once but not be conducive to replayability, key for a portable device.  The tutorial levels in the original game, which introduce game concepts to you one at a time, were greatly condensed into a single level that would still have replayability without feeling like a tutorial.

Overall, levels were selected to be representative of the gameplay and spirit of an average Marble Blast level.  The iPod device was always kept in mind, so that controls do not need to be highly sensitive in order to play well.  Levels were selected for maximum replayability with substantial content in mind.  Further, levels could not rely on deprecated power-ups such as the "Super Bounce" or use highly custom objects such as basketball hoops which would not have corresponding XBLA version art.  Lastly, levels were selected that have regular geometry and a visual appearance that would be not too difficult to skin the XBLA graphics on top of.

Above: An XBLA version multiplayer level with a nice classic platform feel to it for use in the mobile version

The multiplayer levels from the XBLA version were selected to be neither too complex nor requiring the blast function which we omitted for simplicity.  They were selected to easily translate into satisfying single player levels for dual-use and to retain that classic Marble Blast feel.  Finally, multiplayer levels were selected that would integrate seamlessly with the single player levels without a feeling that you are jumping back and forth significantly in both level and visual design.  I believe we succeeded on these goals on all fronts.

CONTROL SCHEME

Obviously on the iPod platform, like the Wii, controls play a major part of the game's design.  I've already discussed how we simplified the controls by removing the blast feature, which then in turn directed the course of which levels we selected and which version we based the game on, so now let's delve into designing for the iPod's unique tilt and touch capabilities.

A marble rolling game on an accelerometer device nearly necessitates that the device be tilted to roll the marble, and that's just what we did.  However, a horrible problem introduced by this control scheme, is that when you are tilting your iPod the only feedback you have on how far forward or to the side you are making the tilt is the marble rolling.  But by then the action has already been performed, and it is too late for correction.  That is far too late in the feedback loop to be informing the player of their input - by then they may have fallen off a platform. 

So we introduced horizontal and vertical markers that nonintrusively show how far you are tilting the device, essentially acting as a buffer between input and outcome to help the player error check that the action they are performing is the action they actually intended to perform.

Above: The tilt feedback HUD

This feedback is especially important for this game where on an Xbox 360 controller in order to set your marble at rest you simply let go of the stick and the stick re-centers itself and provides no input, whereas on an iPod you are constantly streaming input to the device and there is no "auto-center" capability due to the inherent hardware design. 

This feedback is further important for this particular game, as opposed to any other marble rolling game, because momentum plays a large part in the physics of the marble, so if you have built up enough forward speed, simply tilting the device backwards does not start rolling your marble backwards, but rather acts as a brake.  But the lack of the marble rolling backwards when tilted backwards may result in some players feeling the game input is "broken" because they are not getting the feedback they need that their input was successfully received by the game.

Tilt marble rolling aside, the next thing to tackle was the cam movement which is independent of the marble rolling.  No smartcam feature was ever developed for any Marble Blast game in the past, and we did not have the development time to introduce a brand new feature for this version of the game.

Above: The default camera options in the original Marble Blast

In the original game, your arrow keys moved the marble, and moving the mouse left and right moved the camera left and right.  You could hold down the right mouse button to activate up and down looking, or turn on the option to always look in all directions, but by default you could only look left and right with the mouse because essentially that was all you ever needed.  The vertical was automatically locked.

To duplicate this for the iPod version, we simply added left and right on-screen buttons to look left and right with.  You could still press the camera icon in the corner to look around in all directions, for example to try and see where a tricky gem is placed on a ceiling or high wall, but in most cases for basic navigation you need only ever look left and right.

Above: Marble Blast Mobile with the left and right camera look buttons as large on-screen arrows

This control scheme proved to work very well for all levels and, for additional navigation ease, the left and right buttons are automatically pressed when you tilt the device left and right so that theoretically you need not use them at all.  This is not so much a smart cam as it is an auto-cam.

To help court those Xbox players who helped make the game a best-seller on that platform, we provided an alternate control scheme with dual on-screen sticks similar to those on the Xbox controller.

This had the same problem as the tilt scheme in terms of user feedback of their inputted control in that the user never knows exactly which direction and intensity they are directing their marble.  So a visual indicator was added to show exactly that.

Above: Pac-Man for the iPhone

As an example, Pac-Man for the iPhone shows a virtual arcade stick to visually indicate which direction the player has selected.

Above: Marble Blast Mobile's dual stick controls

Marble Blast Mobile's on-screen dual stick controls do the same, with the center circle moving within the bounds of the outer circle as the player moves their finger.  This gives the player the potential to input very precise controls into the game, as well as duplicating their XBLA experience in a mobile format while still providing visual feedback as to the inputted controls in the absence of an actual stick to feel.

Above: iPhone game iDracula's dual stick controls

Another excellent example of providing player feedback of dual stick input is the iPhone chart topping game iDracula.  Unlike Marble Blast Mobile, you need only select a direction for each of the sticks, not a direction and intensity.  So iDracula concerns itself only with the direction you have inputted (for moving and firing respectively).  The game rather cleverly spins the dial of each circle as you move your thumbs in circles around them, giving you superb visual feedback as to what input you are providing the game.

COMPETITION

When designing Marble Blast Mobile we sought ways to stand out from the competition, as Marble Blast isn't the only marble rolling game on the iPod, but it's certainly in a league of its own.  The main game to compare it to is the highly publicized Super Monkey Ball.

Above: Super Monkey Ball for the iPhone

While Super Monkey Ball contains more levels, the levels are much more simple than the twists, turns, and complex geometry of Marble Blast Mobile.  The ability to jump and use items greatly adds to the depth of Marble Blast, and the ability to freely move the camera in the desired way adds to greater gameplay freedom.  Marble Blast has four times as many marbles to choose from, giving the player greater personalization and customization, and is one of the first action games to have realtime multiplayer matches, a huge feature that will be expanded on in future versions by adding internet play to the LAN play.

Above: Manic Marble for the iPhone

Another marble rolling game, Manic Marble, is certainly a fun diversion and has proved popularity as a chart-topping free version game, but doesn't even begin to compare to the depth of gameplay and graphical polish of Marble Blast Mobile.

There are sure to be more marble rolling games to roll in on the iPhone, but the Marble Blast franchise has proved itself as a unique offering on other platforms, and I expect it to continue to be a stand-out game on the iPhone as well.

CONCLUSION

Developing for the iPhone presents some interesting challenges unique to the platform.  Constraining complex art into the slower hardware processor is one, providing adequate visual feedback for player input is another, and porting from another platform altogether provides even further unique challenges still.  But it's an infectiously fun platform to design for, and the lack of specific controller or keyboard buttons open up game design possibilities further than any platform since.

I hope this designer's postmortem has been an enjoyable and informative read, and invite you to comment here or contact me through Red Thumb Games directly with any questions or feedback.  For the technically inclined, you may also wish to check out the technical director's postmortem posted as a blog.

Marble Blast Mobile was a blast to design.  Congrats to Luma Arcade and GarageGames on the release!

Joshua Dallman
Red Thumb Games
www.redthumbgames.com