UPDATE: After reading your responses, it occurs to me that I failed to define "current generation frame rate." In general, PS3 and Xbox 360 games run in the neighbourhood of 30 fps. There are a handful of exceptions such as Rage, MW and many racing games. As a rule, games where the player is engaged in a high average speed are candidates for fluid 60 fps target. This article is in response to the notion that we can expect fluid frame rates across the board on the next generation of consoles.

With next-gen consoles on the horizon, expectations are starting to solidify. On more than one occasion the Giantbomb crew has suggested that these future consoles should guarantee 60 fps. Given that the theme of the last gen was the guarantee of HD, the guarantee of 60 fps would be a good marketing theme for the next-gen, at least to the core demographic. However, even without specs to go on, I sincerely doubt that this will become a priority.

It can be assumed that the next-generation hardware is several times more powerful than the current hardware on the market. Assuming also, for discussion purposes, that all game programming is completely parallelizable, we can claim that targeting twice as many frames would be an easy advantage of this speedup. We’ll even have cycles left over to do additional visual treatments and more complex physics simulations.

That’s not how its going to pan out for two reasons: Greed and budget. By greed, I mean that developers will fill that extra time with more expensive techniques and additional features which will have a larger impact on the wow-factor. By budget, I mean that code optimization doesn’t come cheap; it takes a significant man-hours to achieve. If the target machine is twice as fast as the previous target, I don’t need to budget the time to find an optimization which will half the execution time. Better hardware could actually lead to lower-budget games, since the time needed to achieve a real-time frame rate will have decreased significantly. The main issue is that a game’s performance is only as good as what the project dedicates, and overall high/consistent frame rates are not in high demand from the general public. As Mike Acton, Engine Director for Insomniac Games, pointed out in a blog post two years ago, 60 fps is not a major selling point [http://goo.gl/aeyFb]. I believe this fact will not change, regardless of hardware improvements. Certainly there will be exceptions, but I wouldn't bet on 60 fps becoming a standard anytime soon.

18 years ago my parents got me a copy of Myst for the Macintosh. As I wandered through that stack of prerendered images spinning off of our state-of-the-art 3x CD-ROM drive, I was wholly transported. After Myst I followed the developer Cyan (now Cyan Worlds) with great interest. After enjoying great success from the record sales of Myst, the company busied themselves with its sequel. Four years later, this sequel would come to be called Riven, it came on 5 CDs, and it was ten-fold the experience of the original game. If you’ve never played it and you’re looking for a quiet, meditative experience, pick up a copy off Good Old Games.

After Riven is when things really got interesting. Rand Miller of Cyan saw a future in building persistent virtual worlds that people would want to inhabit. Cyan started a project called Mudpie, and for years they labored on what they saw as the future. In interviews, he talked about building a place people would go just to hike together, talk, enjoy the virtual beauty. He talked about a world that would dynamically erode and change over time. In short he was pioneering the concept of an MMO. Trouble is, during the long development of the Mudpie project, the MMORPG would be born, first with Everquest in 1999 and World of Warcraft in 2001. By the time Mudpie was released in 2003 as Uru: Ages Beyond Myst, it was a shadow of Rand’s dream, and its multiplayer component was meet with only confusion by all but Cyan World’s devote fanbase. Rand had thought that players would want to solve puzzle collaboratively with their friends online. They would pay a monthly fee for access to new Ages, or environments to explore. The subscription never caught on, and Cyan Worlds was not able to keep the game going alone. They attempted to partner with the game-subscription service GameTap to reboot the concept, but this did not last long. Uru was later provided as open source for fans to run on their own servers, and in this form it remains. Today Cyan Worlds develops iOS games, a sad end for developer once on top of the world.

Looking back on it now, I wonder. Was the concept of Uru’s cooperative puzzling-solving flawed or simply before its time? Let’s take a look at what has come afterwards in the long, but sparse history of the sub-genre.

There are few games that truly revolve around cooperative puzzle-solving. Before Uru, to my knowledge, there are no notable examples. Afterward, the next big player that I’m aware of is Media Molecule’s LittleBigPlanet. LBP was the first 4-player platformer of its kind, and while a large amount of its levels can be played alone, cooperative puzzle-solving is required to obtain all of its collectibles, useful for building one's own level. After LBP, there are suddenly a large number of multiple-player platformers on the market, including New Super Mario Brothers Wii.

To clarify, there’s likely a number of cooperative games and game modes I’m forgetting which feature a few simple puzzles as a part of their experiences. For example, requiring two players to hit switches at the same time to progress in Resident Evil 5. I’m specifically talking about games that focus on the solution of puzzles through-out the experience and require two players. Yet, by this definition, LBP is a weak entry, since contraptions which require two or more players are almost always optional.

The most recent game I’ve seen in this arena, and the reason I thought to write this post, is the cooperative mode in Portal 2. In this mode, two players cooperate to solve test chambers in the same way they would in single player, only with the advantage of two bodies and two sets of portals.

The most interesting thing about this mode is that it tells the epilogue of the main storyline. Mechanically, it is also an extension of the main game, as it presents a few new combinations of the game’s components which were not explored in the single-player. In general the polish of these test chambers is in line with the single player. Its clear that Valve was very interested in creating a successful multiplayer experience. Gabe Newell has gone on record saying that Valve is likely to focus on multiplayer games going forward, and even went so far as to say they would not create wholly single-player experiences in the future. The business reasoning behind this statement is clear; multiplayer provides replay value, which keeps games from being resold and keeps players buying add-ons. However it is not clear that its best for a story-driven game like Portal.

As much as I like Rand Miller’s beautiful dream of hiking through a storied virtual environment with a close friend, I think the concept is fundamentally flawed. Although it seems like an incredible concept to experience a story in real-time with a friend, in practice the following occurs:

1. Two players, Orange and Blue, enter an environment and begin looking around

2. Player Blue figures out the significance of the surroundings first (the a-ha moment)

3. Player Blue explains everything to Orange and they move on.

The problem is not simply that one player catches on faster than the other. Its that people can operate at significantly different and valid tempos. I suspect that variance between play times for the multiplayer is much less pronounced than it is for the single player. Cooperative play is often about compromise, and neither player wants to feel like they are holding the other up. So the player that enjoys taking things at a slower pace will often miss out on the a-ha moment (2) and end up being dragged along by the hummingbird-tempoed player. The best case scenario is to be paired with someone who shares a similar tempo to yourself, but ultimately there's still pressure related to using someone else's time that goes against the grain of the puzzle-solving genre.

Even as I make this case against cooperative puzzle-solving, I realize that the concept still fascinates me. The games listed above are easily among the most compelling video games I have ever played, and where they don't succeed I see a humble attempt at something new. Still, questions about the future of purely single-player puzzle-solving games are unsettling to me. I know my apprehensions might never be allayed... and so I close, realizing that perhaps the ending has not yet been written.

The Xbox Live Arcade game Snoopy Flying Ace strikes me as a clear example of an underlying aspect of stylized games and movies. From left to right: An original drawing of Snoopy, a Vinyl figurine of Snoopy, and finally a screenshot from the game. Notice the character in-game is almost identical in appearance to the toy figure. It does not have much in common with the original art, nor would it make much sense to attempt to bridge the gap; the drawing of Snoopy, while pleasing and iconic, has several view-dependent features that don’t map into a 3D virtual space.

On the other hand, we certainly do not want to see a completely photorealistic, bipetal beagle in aviator googles in the game, even though the character will be rendered with realistic lighting and textures. The solution is as old as Pixar’s 1989 animated short Knick-Knack, a concept they would later expand on in 1995, with the first 3D animated feature film Toy Story. The lighting techniques that existed then made for convincing matte material, but little else. Thus, creating characters who are made of plastic allowed them to wow audiences with expressive characters while avoiding complex materials.

Fast forward to present day. While existing rendering clusters can create near photorealistic scenes, interest remains for more abstract designs. You may have heard this has to do with the uncanny valley, but that is a bit of an oversimplification of a complex problem. The uncanny valley has to do with unrealistic motion (or, as more often is the case, unrealistic stillness), but that seemly long understood phenomena will simply disappear once animating technology can catch up to graphical rendering, and it nearly has. No, while stylized worlds will always follow a historic thread of technological restriction and small-budget production, the truth is that style, artistic designs which embody expression and exhibit a playfulness of form, are just as meaningful to people as even the most polished photorealistism.

Thus, even today abstract models exist in otherwise realistically rendered world, giving the impression that these characters are toys, whether intentional or not. With Snoopy Flying Ace, I believe the vinyl figure aesthetic was intentional. With several recent games like LittleBigPlanet, Modnation Racers or Toy Soldiers, the toy concept has again been made a literal one, same as it was 30 years ago with Knick-Knack.

I’ve always felt that artistic form has taken a backseat to photorealism in computer graphics. I get excited whenever I see a game developer use a more stylized approach. This realization that these designs look like toys when rendered within a realistic settings demonstrates that artists have found purchase within this virtual space, and that maybe stylized games will be seen as less risky endeavors in the future. I wonder if this is the best scheme for artists, or if the physicality places hard limits on their expression.

A few game developers, and many more computer graphics researchers, have tried mimicking artwork directly, but for reasons I mentioned previously, this is no easy task, and while a few good techniques exist, they each have their restrictions and graphical artifacts. Furthermore, matching the appearance of a specific artistic style or medium is no better than mimicking reality in terms of enabling artistic expression; an artist working an engine that can produce convincing watercolor drawing is limited to that style and aesthetic.

So this remains an exciting, open problem. Although I just knocked it, I’m going to work on a game with a watercolor look. I’m going to see if I can approach the entire design of that engine from a watercolor perspective instead of a physical one. But as it stands, the best method is just to work with the well-defined approach of rendering abstracted characters and objects as if they were toy figures.