IBM's Bob Sutor Questions Linux Gaming

[L33F3R]

its his opinion so as it may be full of shit to you its not to him. I do agree that you need good gameplay for obvious reasons: Good gameplay in itself means a good game. I do belive you need good programming, eg. nexuiz runs better for me (fps wise) then robombs. I dont agree that you need a good story, for example the original Mario Bros. If i diddnt know better, you are not even told of a princess until the end of the first castle.

V!NCENT not every game engine has what every game requires. I dont want to see anyone make a flight simulator out of tech 3 or a racing sim out of cube2. This is why people re-invent the wheel. The perfect game engine does not and can not exist.

[Setlec]

its his opinion so as it may be full of shit to you its not to him. I do agree that you need good gameplay for obvious reasons: Good gameplay in itself means a good game. I do belive you need good programming, eg. nexuiz runs better for me (fps wise) then robombs. I dont agree that you need a good story, for example the original Mario Bros. If i diddnt know better, you are not even told of a princess until the end of the first castle.

Thanks for respecting my opinion and depending of the genre of game you don't necessarily need a background story!

V!NCENT not every game engine has what every game requires. I dont want to see anyone make a flight simulator out of tech 3 or a racing sim out of cube2. This is why people re-invent the wheel. The perfect game engine does not and can not exist.

The perfect engine is when there is no evolution of the hardware, so that said no perfect engine will exists...

[curaga]

L33F3R, in the age of the nes games the story was supposed to be read from the manual :P

[L33F3R]

L33F3R, in the age of the nes games the story was supposed to be read from the manual :P

Who the hell reads the manual?

[V!NCENT]

its his opinion so as it may be full of shit to you its not to him. I do agree that you need good gameplay for obvious reasons: Good gameplay in itself means a good game.

Define good gameplay. Oops, you can't. Mario has a poor story, 2D graphics and gameplay of a sidescroller. Want 25 30min long cutscenes for a good story? No.

Want a total of 60mins of story in Metal Gear Solid? Hells yeah.

Want to play a game with the ultimate gameplay with 2,5D graphics in a flightsim? OFcourse you don't! If you make 1:1 24 hour flights with a flight sim then you might want to enjoy a beautiful vista, and not with 2D cloud sprites.

It differs for every game out there because every game requires different things. Yeah the gameplay of Duke3D is awesome, but it's a parody in 90's movies so you might require it to have awesomevolumetric rolling explosions.

You want to play a game like Unreal Tournament? Ray-tracing would ruing the sci-fi atmosphere and thus doesn't realy make you believe that you are kicking some alien ass.

I dont agree that you need a good story, for example the original Mario Bros. If i diddnt know better, you are not even told of a princess until the end of the first castle.

Like a said in this post; Mario would be ruined with an intence story line. Metal Gear Solid without one would be ten times worse.

V!NCENT not every game engine has what every game requires.

Because nobody wants to invest all the paid programmer hours into functionality that they aren't going to use and hobbyist programmers don't have enough time to spend on a can-do-everything engine.

One could easily get it done with standard base that can be extended with modules. For example:
Mario would load the standard camera moving API (x, y movement) and Crysis would also load a scripting camera module to make cutscenes.
Mario would load the standard movement API and the scripting module so Bowser would move in certain pre-defined ways, while Crysis would also load the extention module for physics.

I dont want to see anyone make a flight simulator out of tech 3 or a racing sim out of cube2.

Yeah... so? Those engine are not made for all games in general so they do not work... which is why a generic game engine should be made that has all kinds of extention modules to add upon basic functionality.

This is why people re-invent the wheel. The perfect game engine does not and can not exist.

Humans will never be able to fly. 64K ought to be enough for everyone. A bomb the size of a car is never going to be able to destroy an entire city unless you change the laws of chemestry, etc, etc, etc.

[mirv]

Making a game engine to cover all genres doesn't really work. Leaving graphics aside for a moment, there's A.I, networking, physics, and a slew of other issues to take into account that really make it unfeasible.

[V!NCENT]

Making a game engine to cover all genres doesn't really work. Leaving graphics aside for a moment, there's A.I, networking, physics, and a slew of other issues to take into account that really make it unfeasible.

That's why shared and non-shared functionality should be broken up. Much like with the Gallium3D architecture.

One game might require this instead of that, and so two different modules could be loaded. For example: A game of Nexuiz may require a fixed size map module for its arena's while a game like GTA 4 requires an infinit map size streaming module for dynamically loading segments on the cities a player drives through.

With A.I. one can do the same thing. Shared functionality like paths, fear, desicision, etc. while one could add formula extentions for different behavior like when to run, like when to attack and awareness (2D, 3D).

Then awareness can be plugged into the objects and map modules by means of interfaces defined in the specs.

Procedural programming, sharing, extension, specification and modules are key here.

Networking can be split up into a few different modules, like a standard multiplayer FPS module with extensions for certain game types, an standard multiplayer race sim module with extensions, etc. But who's on top (rank) can be shared, however time and kill/death ratio could be two different extensions.

[yotambien]

A good game should just be a good game. Period.

Well, it's hard to disagree, isn't it?

Jokes aside, I know what you are trying to say. If somebody had the magic recipe to produce good games, that somebody would be worth billions. Not even companies with a long history in the game business and employ teams of experts have that information. Simply because it doesn't exist. Having that knowledge would imply knowing what goes on in the mind of millions of potential customers, which is impossible. It's a bit like giving a definition of a masterpiece of art; if you had the perfect definition there would be nothing stopping you creating them in industrial quantities...however, what we usually see is art critics sticking to their commentary instead of getting rich :P

If you compare Space Invaders, Tetris, Super Mario, Street Fighter II, Zelda, Quake 3 or WoW there's not much they have in common, except being awesome in their time. Hell, it would even be difficult to say what makes Sonic and Super Mario different and cool in their own ways.

The only way you can define some features that a succesful game should have is to intentionally be very vague and general, making those definitions useless. Of course, I could be wrong and game companies make crap from time to time on purpose, yay.

[mirv]

That's why shared and non-shared functionality should be broken up. Much like with the Gallium3D architecture.

One game might require this instead of that, and so two different modules could be loaded. For example: A game of Nexuiz may require a fixed size map module for its arena's while a game like GTA 4 requires an infinit map size streaming module for dynamically loading segments on the cities a player drives through.

With A.I. one can do the same thing. Shared functionality like paths, fear, desicision, etc. while one could add formula extentions for different behavior like when to run, like when to attack and awareness (2D, 3D).

Then awareness can be plugged into the objects and map modules by means of interfaces defined in the specs.

Procedural programming, sharing, extension, specification and modules are key here.

Networking can be split up into a few different modules, like a standard multiplayer FPS module with extensions for certain game types, an standard multiplayer race sim module with extensions, etc. But who's on top (rank) can be shared, however time and kill/death ratio could be two different extensions.

A driver architecture analogy doesn't apply - drivers output to a screen, and often use opengl for the 3D stuff. All share a common goal. Games, and thereby game engines, don't.
But programming-wise, good luck with trying to create a game engine operating system. Some things just can't be modularised out - A.I being one of them. It's really quite glued to the underlying game, and it can't share some things (path detection for finite vs infinite maps, flight sims vs fps vs rts, plus more).
Nice idea in principle, but practically speaking it sadly really is a no-go.

[L33F3R]

Define good gameplay. Oops, you can't. Mario has a poor story, 2D graphics and gameplay of a sidescroller. Want 25 30min long cutscenes for a good story? No.

Want a total of 60mins of story in Metal Gear Solid? Hells yeah.

I for one just finished playing metal gear 1, it is a solid game. However i found it to be more of a movie then anything else. Its play-mechanics also suck royal ass. for a game of its age it however did a very good job. Just as mario did as a 2D scroller in the 80's.

btw:

Nvidia defines gameplay as

the way in which the gamer can interact with other elements in the game.

Good gameplay is the interaction with other elements, in a "good" way. Sinse the goal of any game is enjoyment, good gameplay then refers the the interaction with other elements that result in the end user having a good time. That wasn't so hard was it?

-----
I was going to quote a bunch of stuff here and apply reasoning, however its quick to realize you like to ramble on and that this post would take up 2 much page realestate for its own worth.
-----

Humans will never be able to fly. 64K ought to be enough for everyone. A bomb the size of a car is never going to be able to destroy an entire city unless you change the laws of chemestry, etc, etc, etc.

The perfect engine can be made, so start making it. because none of us are planning to do that. Dragonlord is sort-of trying to make a jack of all engines (might want his input on that) but to call it perfect is absolutely impossible because perfection requires the consumption of 0 resources while having absolute functionality. Humans cant comprehend what absolute functionality is and as a result will never be able to achieve it. Now i suppose you could argue that the basis of perfection surrounds the individual however at a basic level, nothing is/can be perfect. Unless of course you believe in God.