JF make 189
JF Nav
JF Nav
Creation Date: 2002-08-06
As I promised, a perfect Making Of JF. Actually, a bit less than perfect, but better than yesterday's disgrace. It's a side project that I dreamed up the other day. It'll be a dating sim game possibly in 3D and possibly involving a cellular phone. Actually, it'll definately involve a cellular phone, possibly much like this one. This is a seriously high-poly phone. Well, professionals would call this a low-poly phone since it would run in realtime by itself on a 4 MB video card (Riva TNT or Voodoo 1). That's another story*. What I want to say is that this phone has more polygons and texture size than a full human model. Isn't that sad? Well, perhaps I will take a 128x512 picture of the phone and make it a gui item. That is one of the few times that number of polygons and size of texture doesn't matter. But still I want to be frugal because frugal makes good models, no matter the usage. That is the lesson for today. Tomorrow you'll see another good picture which is a continuation of this picture.

*Back in the day, low poly was an art. You had to reserve all the polygons for the frigging buildings. Everything else was a sprite (a forward-facing textured square (a sprite) takes less processing than a single polygon). They had characters as animated sprites, powerups as sprites, etc. It worked well. Back in the day, my 486/66DX with 32 MB of RAM ran Duke Nukem 3D with sprites and wonderful polygon buildings. But as the video cards got better, people started using those beautiful triangles. They made characters out of polygons. They made powerups as polygons. They made furniture out of polygons. Then we all thought of polygons as the primary geometric primitive. Some people like me still think that it is. But the truth is the next set of games will only use n-patches. GeForce4 and ATi's newest video cards both support n-patches as primitives. N-patches simply are curves that can be told how many polygons to split into. A character up close will be given a large number and a character far away will be given a smaller number. That means that the person in the background can be displayed with very low detail and the person in the foreground can be displayed with very high detail and the graphics render at high speed. It also means that developers must make software version that doesn't use so much processor for people with lesser video cards. The next few video games will do that. What will the developer do? Instead of having a progressively better model, they'll just have one or two models and fog to blur the further ones out. Then you'll get a bit of popping as usual. You'll get poorer performance which should be expected on newer games. But it wasn't expected back in the day. Back in the day, like System Shock 2, they used their polygons like they cost $250 dollars each. That attitude will make good models. If I say, "That model looks like a million dollars," I'm insulting it saying that it looks like it'll cost 4000 triangles. A $125,000 model would be good. ^_^ Anyway, System Shock models were so good that they used them in cutscenes. Cutscenes are another place where you can put high-poly models, but they decided not to. Why? Because they're better than any high-poly model that they could make. Also, System Shock used very good skeleton animation. In the cutscene, we see a couple people walking around. Throughout the game, all the monsters use flawless animation. But I noticed that people are only animated while laying down. Why is that? Well, perhaps friendly AI was not in their plans. People were all either dead, almost dead, or on the other side of the ship (on their way to be killed). System Shock is so sweet, though. The whole system is so beautiful. I'd be happy making something as good as System Shock 2 even in the day of GeForce4.

I haven't made it a secret that I believe that this graphics explosion is unwise at best. At worst, it will end in catastrophe. Here's my idea. If video cards do what people say they will, we're going to reach a limit to computing power in six years of doubling every nine months. That means that graphics processors will max out at 256*136M=35 billion theoretical vertices per second. The actual speed is around 50M max right now. That brings it to 12.8 billion actual vertices per second. To get 60 fps, they have to limit their scenes to 213 M vertices. If they are wasteful with polygons, they'll spend 100,000 vertices per character. They'll spend a cool million per building. They have a terrain with 50 M vertices. Then, they have to budget as usual. You'd be limited to a game with a few characters and a few breathtaking scenes, but nothing special. People who make games that budget their polygons correctly would be turned down by publishers. But at this rate of expanse, I predict a terrible catastrophe. That is a generation of young people will go their entire lives without seeing a single quality video game. They will see eyecandy until their brains will pop. Not literally. They'll just simply decide that games that are well-built take too much time and don't have any eyecandy or 'fun'. Games that actually are built to do something, challenge the user with an interesting idea and give them a good story with actual plot (intro, hook, climax, and finish). It's so hard these days to find an actual good game. The most recent one that I've played is GTA3, but it has made several people ill playing it. The colors and the motion are all wrong. The gameplay was not very good. The plot was less than nominal. However, it did have a bit of plot and for that it wins "Best Game since System Shock 2 Award." But it wins that only because of a lack of any competition. While eyecandy will keep the money rolling, it won't keep the minds rolling.
JF Nav
Home Characters Making Of Technical Mail News Links |< First < Prev Next > Latest >|  bandwidth version Goto Scene