Monday, June 8, 2009

Problems with Natal

In my opinion, this type of motion control wont work out as well as we might hope. The three main downfalls of this technology, as I see it:

1. The lack of a tactile element to the control. Without a physical object in you hand, you may be moving you whole body, but it doesn't seem real. The action has no weight to it. The driving of the car, for example, without a steering wheel strikes me as fairly difficult because there is no physical object linking me to the game. If I can use any physical object I have laying around, say the Wii Wheel from Mario Cart, that would go a long way to fix that problem, but I doubt that I can. It would probably mess up the recognition. And what about other games? Games that don't have any real object to interact with? When you are playing the fighting game, there isn't anything to punch, which could be a real deal breaker. It is fun to practice punches on a punching bag, but not so fun to practice punching in thin air. There is a certain satisfaction to interacting with the real world that will be missing from these games.

This would be alright (we already are not interacting with the real world when we play games) except for problem #2: A lack of precision. The great advantage of the physical controller as an input device is that it solves 3 problems: It establishes a standard to which all developers must conform, its use is simple enough that most people can train themselves to eliminate human error when inputting commands, and it is simple enough that technical errors can be practically eliminated.

With Project Natal, these three things are thrown out the window. There is no standard to which developers must or even can conform. This makes development much more difficult, and make it much more difficult to pick up and play (when the game is any type of complicated.) Every time we pick up a new game we will have to learn a new set of movements that will input commands. Anything that would normally require interaction with a real world object or an action that many normal people couldn't pull off must be made abstract, so real world experience is useless. If street fighter was ever given support for this, how does one use Chun Li's spinning kick attack? I can't do anything that even approximates to that action. For that matter, how does one pause a game? Or reload a gun? Or pull the trigger on your gun? Or aim that gun? Or pick up a gun? Virtually every action that we perform on a regular basis must be made abstract and therefore cannot draw from real world experience.

Because we must learn a new input system for every game, we never perfect any one system of input. Human error with be compounded a thousand times for this type of game. Can't remember how to reload? There is no controller there to look at so you can remember how. You will have to pause the game, go into the menu, etc. Even if you are able to remember all the commands, chances are you will make mistakes and have problems with the input from time to time.

Even if you are able to perform all the correct commands perfectly, there will always be technology problems. There was some notable lag and imprecision in those videos. In a serious real time game setting, and especially in a competitive game, the level of precision must be absolute. There is no room for error, no time to think about input. It must be automatic, quick and precise. That can never be with this type of control.They claim to have done it, but I have only seen evidence to the contrary.

Problem 3: The inherent difficulties of programming with this type of control will scare away most third party developers, at least for the games we would care about. Don't expect anything special for this from any third party developer. We will get first party Microsoft support with a few good games at best, followed by dump truck loads of shovel ware that we will hate.

The real gem of the technology is this Milo thing, but not as it was presented. What we can take from the Milo demo: Precise voice recognition, precise facial recognition, advanced recognition and interpretation of context sensitive and imprecise data, advanced integration of imprecise data. This is obviously an early build of the technology, but what this means is that these four things, and probably other advanced AI functions, will start popping up in games we play 2 to 5 years from now. Remember how in most early RPG's you could name the characters? That has disappeared recently because of the need for voice acting. With this technology, in Final Fantasy XIV I may be able to name my character Oswald and hear the name Oswald in the voice acting throughout the game. There are millions of potential applications for this type of advanced AI, but its future use in video games will not be a super Tamagatchi, but integrated into our games piece by piece. This has real potential, this is what will be use in the future of videogames. The motion controls are flashy and fun to watch, but this AI actually matters.

No comments:

Post a Comment