Wednesday, January 16, 2008

DVD Controls for Video Games

Games should come with features allowing them to be controlled like DVDs. The player should be able to rewind, fast-forward, skip ahead and back, and jump in at any chapter point. Every cutscene, whether video or game-engine-rendered, should have full DVD controls. And all cutscenes should be viewable at any time.

Games are entertainment, and as such they should support the same kinds of user control we enjoy with our other media.

There are a number of problems involved in making this happen, and it's not right for every kind of game. There are technical, social, and design problems, and games with neither a linear nor a branching storyline are unsuited for some of these features. But for games which do tell a story these controls should be present. And even for the other kinds the underlying philosophy can be applied.

Games without any plot to speak of cannot have any kind of skipping ahead, as there's no plot event to skip ahead to. This includes some sports and simulation games (such as SSX or Gran Turismo), competitive multiplayer modes of games with a plot, and games light on plot (such as Crackdown). However, the principle is still valid.

Even if they can't fast-forward, they could provide ways to rewind, something similar to the undo feature found in many word processing programs. It would consume quitte a bit of storage, but a game could log all the player inputs and game events, and give the player the option to rewind to any point since the game began and play from that point.

This is nothing new and has been done before in various forms; games such as Doom and Quake and most recently Halo 3 have offered the option of recording gameplay and replaying it later (though not the option of jumping back into the action).

Skipping back is easier to implement, and really only requires keeping a save for each checkpoint passed. Something like this has long been available for PC games in the form of unlimited saves and save-anywhere, a typical feature of first-person shooters. It's more difficult for consoles, where storage space for game saves has until recently been quite limited. But the recent growth in available storage (in the form of hard drives, as well as higher-capacity flash) makes this less of an issue than ever.

Going back to a previous chapter presents exactly the same challenge as going to a previous checkpoint, but with an additional possible solution. Some games already offer this. Those that do, such as Halo and Doom, reset the player's status and give him a fixed set of weapons at the beginning of each chapter (or level, if you prefer). This makes it easy for games like Halo and Killzone to give the player the option of replaying any completed chapter.

Fast-forwarding is a tougher nut to crack. It would require the game to essentially take the player's place and play itself, and at increased speed. With modern games already using all of the processing power available to them, this would seem impossible. But there are tricks and shortcuts that can reduce the processing load, and some games that already do pieces of the solution:

  • Don't render every frame. In many games a large portion of the time is spent rendering each frame. When fast-forwarding at 2X, you would only need to render every other frame to achieve the same framerate.
  • Allow a lower framerate. This reduces the amount of time spent rendering even more. A framerate as low as 3 fps would be enough for most purposes, and is on a par with some DVD players.
  • Simplify the display. Gran Turismo 4 has a fast-forward when run in B-spec mode; it's just a 2D map of the track and a leaderboard. Something similar could be done for many games. RPGs come to mind. This would reduce the redering load even further.
  • Simplify the world model. There's no need to have the player's character aim his weapon, then compute bounding box intersections to see if there's a hit. Rather, combat can be resolved at least in part by using RPG-style dice rolls to figure hits and damage. Similar things can be done with physics-based events and other processor-intensive interactions.
  • Simplify computations during empty time. The game doesn't have to process visibility calculations on every frame; it only has to predict when the player will become visible to other characters and only calculate position until that time is reached.
The other part is that the game has to have AI good enough to take the player's place...which we can already see in a few games, such as the drivatar in Forza Motorsport, and the companion characters in Lego Star Wars.

Skipping ahead presents much the same problem as fast-forwarding, with some additional solutions. Skipping rather than fast-forwarding allows further simplification of the game world and its interactions; even visibility and proximity calculations do not need to be done. Skipping ahead a checkpoint's worth can also be handled exactly like jumping to a chapter.

Jumping to an arbitrary chapter or level allows the game to completely ignore game events, and simply set the player's state as well as the game world's to a reasonable value. Halo and Killzone already do this; the player begins each chapter with an appropriate set of weapons and everything else in the level in their predefined places. Games with more complex state such as RPGs would need to have a developer play through the story first to determine a reasonable state for the game at each chapter point.

As for cutscenes, some games already allow the player to go back and view ones they've already seen through gameplay (e.g. Star Wars X-Wing, Tomb Raider Anniversary). But I think that's not quite enough; they should be treated like chapter points or checkpoints and viewable at any time during the game.

And while I've seen a few games that allow pausing of cutscenes, far too many make them unskippable or only skippable. I've yet to see one that permits fast-forwarding and rewinding of cutscenes. When they can be upwards of twenty minutes long (Xenosaga), not being able to repeat or skip around is highly annoying. Life does intrude, after all.

I expect objections to these features to fall into two categories: the can'ts and the shouldn'ts. The can'ts in turn come in two varieties: technical and economic. I'll begin my discussion of the can'ts with a negotiated surrender. You're right, but.

It's not technically feasible to implement all of these features for every game. Some would require AI we don't yet know how to write, or more system resources (whether storage or processor cycles) than are available. Some games have a structure that just isn't suited for fast-forwarding or skipping ahead. But some do.

It's not economically feasible to implement all these features for every game, either. Some, though technically possible, would cost more than they'd be likely to bring in in terms of additional sales. Others might lengthen a game's development cycle beyond practicality.

However, this doesn't mean that these features shouldn't be considered, just because they might not be feasible. Rather, they should be part of any game's planning cycle, to be evaluated rather than dismissed out-of-hand.

Regarding the shouldn'ts, however, quarter is neither expected nor given. Here too the objections fall into two major categories. There is the competitive objection, and the directorial objection.

The competitive objection is that, particularly with the features that involve skipping forward, the player hasn't yet earned the privilege of seeing that content and should be prevented from doing so until he beats the current challenge. Skipping ahead is cheating, and also takes away the challenge that makes playing the game worthwhile.


There is more than one reason to play a game. One is for the challenge, to have one's prowess measured by the game (or by others through the game in multiplayer). But that's only one. There are others. Another is to have a interactive dramatic experience. A third is to have fun playing with nifty toys.

Only when playing for a challenge, or a combination of a challenge and the other reaons, is it appropriate for the game to treat the player as an opponent and withhold dramatic content as a reward for passing the game's test, for winning, for beating the game. In the interactive drama case, the player is not opponent but audience; he paid the price for viewing the content when he bought the game. Similarly for the toy case, the player paid the price for using the uber-weapon (or driving the uber-car) when he bought the game.

One way to handle these apparently contradictory requirements is to allow the player to declare at the beginning of play how he'd like to enjoy the experience - as a dramatic participant, player with toys, or as a competitor. In the first case, the player would be given access to the sorts of DVD-like controls I outlined above. The second is like the first, except that vehicles, weapons and other things not available at the beginning of the narrative would now be available at any point. And the third would turn off all the skip-forward controls and some of the skip-back ones, while adding rewards like medals and achievements for succeeding at various tasks in the game.

A player in challenge mode would be able to switch to either of the other two modes, and players in the other two modes would be able to switch between them, but they wouldn't be able to switch to the challenge mode without starting the chapter or game over.

Some games already implement these modes to various degrees, but without the DVD controls. A notable example is Lego Star Wars, which has a story mode and a challenge mode (free play). Dying in story mode has little consequence, except to cost the player in-game money required to buy trinkets that competitive gamers will appreciate but which don't affect progress through the story. However, like Halo, the ability to skip ahead is not provided. The developers could have offered such an option, since each chapter starts from a fixed state, but chose to make the chapters serially unlockable instead.

So it's possible to satisfy the competitive player while also giving the story-focussed and toybox players what they want - including DVD controls. This completely negates the competitive objection.

The second of the shouldn'ts is the directorial objection. Holders of this view would have it that indeed there are entertainment players as well as competitive ones, and that the entertainment player must progress through the game along the path(s) chosen by the developers, or communication of the developer's artistic vision will suffer.


Imagine the absurdity of this when applied to other media. How would you like a book where you couldn't skip to the end or to a favorite passage, where you had to finish chapter 9 before being allowed to read chapter 10?

How about a videotape that won't fast-forward? Or a DVD that disables all the navigation controls as soon as the movie starts playing in order to recreate the theater experience? Or a music CD that can only be played in order?

The only reason we put up with abominations like the unskippable cutscenes in Final Fantasy X is that we've been conditioned to accept it in videogames where we wouldn't in any other media. Partly this is because it's far harder to implement navigational controls for games than for linear noninteractive media, and partly this is because we're used to thinking of games as competitions rather than interactive storytelling.

There's one specific variant of the directorial objection I'd like to deal with before moving on: that DVD-like controls would ruin the feelings of suspense and fear in horror games. This was the argument used by the developers of Call of Cthulhu: Dark Corners of the Earth to justify their use of a save point system rather than save-anywhere, and the painfully large distance between save points. This argument has it that if you can just restore from a quicksave when you die, you won't feel any fear.

This is patently false. Were this the case, horror movies would create no sense of suspense or fear in their audiences; after all, there's no equivalent to repeating a tough part when you're in the movie theater. Repeating a long tough part in a game isn't scary. It's boring and frustrating. And if fear of being bored and frustrated is all they have to offer, then that's pretty weak sauce.

Horror movies create fear in the audience through the situations, sights, and sounds. Horror games should do the same.

So... the directorial objection doesn't hold water either.

And with the can'ts and the shouldn'ts out of the way, I'll summarize. Games should be enjoyable in the same way our other media are, with DVD-like navigation controls. These controls should be implemented whenever it is technically and economically feasible to do so, and let the objectors go pound sand.

No comments: