There are zillions of skilled engineers outthere and how many actually create something inovative as a car or plane ? I guess same goes with programmers and games … I guess it takes huge amnout of luck (besides) skills which are widely spread and can be acquired by traning to do something really new. Most brilliant game related ideas “I got” - came from my wife and little guy living across the street I live … ;D
After finding stuff like Black and White very compelling and seeing what everyone is saying about HL2 I reckon that anyone can do great graphics now and the next great leap forward will be great AI - not just enemies but underlying economic and storytelling engines that allow games to create their own, truly interactive, plotlines. The possibilities in this direction are so great that they make me want to flip out and wail on my electric guitar.
AI - that is a most likely next step - and just to reiterate what you’ve said - possibilities are enormous. I have started lobbying friend of mine who knows that stuff pretty well to start thinking about how to use it in games. But for him games are just …well…games… he is teaching AI grad courses at local Uni and dont wanna 'waste his time on childish untertaking" (as he put it)… But soon or later time will prove him wrong - AI usage in games will get big - at least that is what I think.
Doom 3 is supposed to have vastly improved monster AI. That should help the game play, and it will look nice too :).
I think it’s own child will eat it, though - by the time it comes out the point of reference will be Half-life 2 and I’m not sure how beatable that is going to be. The facial expression technology and everything like that on HL2 sound so impressive that unless Doom3 pulls more than some pretty images and monsters that don’t get stuck in scenery it is going to be overshadowed. Fair play, though - ID have ruled the FPS roost for the last 10 years, so it’s pretty much someone else’s turn by now.
That sounds good and looking forward to it. But I would hope to have AI applied not only to monsters in game - in general - but to all characters in game. Ultimately one would join and have to create allience with other users BUT there would be no way of knowing if there are real or AI driven characters. One could make meaninfull conversation with such an entity - and would have to figure if that is real person or drone… now that would make things interesting! …
BWT – I guess this thread somehow got morphed in the process as it belongs to AI or Game Design columns - hope we dont get punished for this :
This thread has gone through about 4 topic changes now, heheh. ;D
Ok back to topic. I think everyone agree that programing shaders is cool. Being able to use the new extensions of OpenGL and DirectX as soon as possible may be a very important factor in game development. Maybe thats why Java3d is not so successful. Being able to work unrestricted like this and work with an api like java3d would be perfect.
Just look at the api features you all know. The scene graph is the least of all. The Behaviors and behavior lod. Behaviors can take advantage of muti-threading if the right machine is available. Unlike others apis with scene graphs java3d api can be traversed in parallel.
If java3d has really failed it’s only because this api is too advanced for his time.
Now guys lets look at one of very aclaimed rpg like Morrowind. It hardly suports any level of paralelysm its AI is mostly script base and uses the most basic algorithms for path finding like those you guys see in gamasutra LOL. It lacks a proper scene graph to make complex lod schemes like those seen in Gothic. It lacks behavior lod like in java3d. Still the engine can draw high resolution textures like hell and that can make an average game look excellent.
Its true that high-res textures and polygons are critical, but if the player gets something he has never seen before like a new way to lod terrain or advanced AI interaction then the differences between java3d and a c++ api will be much smaller.
In short there is still great games that can be build with java3d but such games would have to take advantage of java3d and use other entertaining features other than high-quality textures. If I didn’t believe in java3d I would probably be programming in Visual C++ and DirectX, which bluntly speaking, sucks very much. ;D
And there I have been quoting Morrowind’s use of the Netimmerse engine as an example of the practical use of a Scenegraph API in a top quality game…
[quote]And there I have been quoting Morrowind’s use of the Netimmerse engine as an example of the practical use of a Scenegraph API in a top quality game…
[/quote]
Believe me, the reason of choice for NetImmerse in Morrowind is speed and rendering capabilites. Gothic 2 is more of a pratical use of a scene graph in my opinion.
Morrowind uses a plain array to represent terrain and a system of interior/exterior cells. 9 exterior cells for the player location and surrounding area and interior cells for the insides.
Thats simple and fast but … you can’t sneak peek in a window to look at an interior. Because they use an plain array for the terrain that (no compression at all) the landscape size you see is the absolute maximum without needing a 1Ghz file just for terrain data.
If you are thinking of the last version of NetImmerse now called Gamebrio.
Thats very impressive but game coders hardly every use all the features in a licensed engine.
[quote][…]The engines are very awesome in both cases and in the hands of people like Ion Storm (as they are now, let’s just forget the game that must not be named) some amazing things have been done with them.[…]
[/quote]
http://www.penny-arcade.com/view.php3?date=1999-11-22&res=l ;D
I’ve done a lot of stuff with the Q3 engine (and btw it’s written in C not C++). I’m really looking forward (ab)using the upcoming Doom3 engine (wich is - btw - written in C++).
And about AI… the bots in Q3 are incredible stupid… I had to use alot of strange tricks to “teach” 'em some simple trickjumps. Well… they can’t do basic timing… or predict rockets (well they did that for awhile but it got broken (and it’s now fixed again in the latest cpma release))… they can’t dodge and so many other things.
Well some things are kinda funny solved. They can’t “hear”… but if you try to shoot 'em in the back and miss they’ll turn around and shoot at you. Heh… that looks kinda “realistic” isn’t it? Well infact it’s done quite simple if you fire your weapon within a specific range the bot will immediatly know where you are (and not that “realistic” - he will also know if you are on his team).
Sorry for re-re-re-hijacking the thread :
The reason Morrowind used NDL was because it worked well, it has tools that integrated with 3DStudio MAX - Bethesdas artists package of choice, it had relatively good support, it allowed them to focus on game development and not engine development. In short it was a good workflow for them. On top of that it also allowed then to do an XBox port in parallel with the PC development. It also utilized some reasonably open hooks for plugging in shaders.
Beyond that I will not comment as I may still be under NDA.
In my opinion Java3D suckz!!!
JOGL is much easier to use and faster than J3D!
And the runtime is only about 0.5 MB!
WiESi
Wow, its always nice to see a carefully thought out opinion and so elegantly put. ;D
Jogl and Java3D are suited to different roles. Jogl is good for high performance, Java3D (generic scenegraphs generally) is good for rapid development.
KeV(z)
Apparently expertise is inversely proportional to the number of posts someone has made. We don’t have a chance…