Game Development 101

yup, i've had that myslef for awhile, i believe its the same interface as 3dsmax, am i correct in that?, thankfully making the switch from blender and gamemaker to 3dsmax based stuff has actually paid off already, aside from the numerous online forums, i dont have to worry much about incompatibility with files and such

also, fabrizo, you seem to know 3dsmax fairly well, and i look forward to having some more inciteful help, though i have a question. now i know i wont be making models like this anytime soon, buts its been bugging me latley, say when creating a car model for a game, lets say i want to have a cockpit view, 3d of course, would i create this cockpit as part of the original model, or something seperate>, reason i ask is after looking at the models from racing games, the car model interiors seem to have fewer poly's then the in game view. alas, thanks for the help

Ryan

 

There are two ways of doing the cockpit room. Arguably the easier way is to just completely model the car, then affix the camera to an interior position. Alternately, you can make two models: one for viewing from the exterior, and one that is just the interior view.

One major disadvantage to doing single piece car model is the level of detail. The closer your camera is to an object of interest, the higher the LOD for that object should be. At a distance (i.e. overhead view), the car would be constructed with fewer polygons than if you placed the camera inside the cockpit. If the detail is too low, the object starts to look worse and "edged". If the level of detail is too high, the system will be processing extra graphics detail that would go unnoticed. With a single piece car, you will be attempting to render the entire thing regardless of the camera placement.

You could do a localized higher LOD for the interior in a one piece design, but the system would still need to pass the extra detail through the rendering pipeline or explicitly cull it before rendering (if you don't know what I am talking about, it would do you well to look it up). Either way, it would require the system to do more work than is necessary.

Using two separate models allows you to customize the LOD depending on the camera position while minimizing the number of polygons that need to be rendered. The disadvantage of this method is that you cannot simply just zoom out the camera; it requires an actual swapping of the visible model during the transition.

BTW, does anyone know of any commercial games using implicit surfaces instead of polys?

 

Well if your concerned about performance, yes you could go and make the cars interior and exterior seperatly. What i'd suggest doing is making one car with the highest ammount of detail that your aiming for and once your done make two replicas of said car. One of them chop out most all detail from the areas the player wont be seeing when in the cockpit (depending on how much the camera will be turning inside their you can judge this), and the other do the opposit and chop away most all of the interior detail. The reason for doing it this way is that you get much more accurate general sizes shapes and placement of detail then you would making them seperate (like having a massive steering wheel when inside, but a mini one when outside). Then when switching camera angles you can have it swap car models durring the transition.

As for what mairsil was saying, to summerize, game engines render everything the camera is pointing at, despite if its behind another object and chant be seen by the player. for single objects such as a car this is how things are done (levels themselfs though get optimized differently depending on the engine). Because of this having one model for your car is a bad idea that will force you to either make it low poly throughout, or have a good ammount of detail and probably have it result in frame rate issues.

Also, two terms he mentioned you might not be familiar with:
cull = where the engine ignores certain polys when rendering the scene and leaves them out to save on performance.
implicit surfaces = Nurbs and Splines. Both are essentialy just differnt ways of getting generalized geometry (they ignore hard edges), as opposed to using polys.

 

i see, i would use a high poly model for close-ups, a low poly model farrer away, and maybe a sprite(i seen someone do that before) for the furthest away , for the cockpit again, i was wondering, nascar heat, NR2003 and nascar thunder(some games that i play ) all feature a "tv cam" on which these cameras located at various points of the track would zoom in and out on the cars, im assuming that these games only use one model?

Ryan

 

Not necessarily. The way they could be doing it is to render a particular camera view, swap the car models and render another camera view all before swapping the display buffers (i.e. showing the rendered frame on the screen).

 

that can be done in real time?, if so, damm im impressed. Hopefully the person who gauranteed 3dsmax can follow through, if not back to blender, also, does gmax use the same interface as 3dsmax?, if so, i think ill get crackin on using it

Thx

Ryan

 

gmax's interface is *very* similar to 3DSmax's interface. If you learn one you know the other.

-hl718

 

ahhh great to here, i shall get started on that soon, first i must get my website done, finish the railroad layout, ill be well prepared by xmas, then i shall code, create or do whatever gamers do!!

Ryan

 

darn it, again my "supplier" failed to come through today, ill ask tommorow......i've been mulling over designs lately, and came across 2 hand written designs dated 06-11-01 and 01-21-02, alas i had forgot about the game that actually inspired me to start game development, i called it at the time "A World Within" basically animal crossing, but bigger, more choices, more everything, im also trying to remember the design for a puzzle game i had awhile back....

I have a question though, though these were 2 games, I've also designed a console (actually many, but this one the farthest) I've tried to document everything from the design, to the name, to the layout of the motherboard, I've built a mock up, tried aqquiring parts for it, now i know this is crazy but where would one go after here?........ahhh it's just a bunch of crazy talk:-(



Ryan

 

again, ive yet to get 3dsmax, is there any other good modeller, that can support 3dsmax models, or is there a converter?

 

I think gmax can use them, and it seems to be geared towards games from what i've seen (and used).

 

unfortunatly they dont, it uses it own file system, seems to be simliar, but wont open 3dsmax stuff, or atleast the stuff i've tried

 

Write your own 3ds file translator. There is plenty of information on the file format and it is an extremely beneficial skill knowing how to work with different file formats.

 

wowzers, since gmax is based off 3dsmax, and looks to have ll the same features, that actually looks like a feasable option, though my programming skills are limited, where would one start with something like that?

Ryan

 

Well, first make sure you know how to open, read (and parse for necessary info) and write a file programmatically. Start with text files, then work your way up to the binary stuff, things like BMP's or JPG's. Once you are comfortable with those, then find the 3ds format descriptions and work with that.

 

wowzers, i might actually look into it, seems like a good project, though i musn't start it now, Ive got wayyyy to many projects on the table right now..

 

I uploaded the manual to Fab's ftp

 

thanks hogan

Ryan