If you haven’t already saw, check out the summary of tech talk 13. Some pretty big news such as cross-platform support, a new graphic engine, a new hire, and scripting support.
If you haven’t already saw, check out the summary of tech talk 13. Some pretty big news such as cross-platform support, a new graphic engine, a new hire, and scripting support.
sux
That’s all pretty sweet. I wonder what rendering engine they’re considering?
Here’s hoping there’s some basis of physics
Thats awesome news. Finally I can catch up on Linux been a good six months since ive had it running.
I’m taking bets on Unreal.
Might want to wait a little on that. It may still be 6 to 9 months before we see an Alpha copy, and then it may even be inhouse only. Right now Chrispeg wants to work on the current browser and making it as stable as possible. Also, he has to wait on Management to make decisions on the “lite” edition of features that will be in the initial build as well as what graphics engine will be used. Chrispeg is a pretty fast programmer, so once those decisions are made I am confident he can get it done in a reasonable time. This new browser will also be a seperate product from the old browser. Both will be maintained side by side, and the initial new “lite” browser will use the same uniservers, world servers and object paths we currently have… initially. Chrispeg also said the new browser would likely have a different user interface than the one we are used to in order to be used with handhelds and other operating systems.
Industry standard engine? Oh boy. Unreal, idTech and Source would all suck for something like AW
Are you suggesting Active Worlds need to have its own custom graphics engine like Grand Theft Auto IV does with one created by its own developer (they’re part of Take Two, a rival of Electronic Arts, who acquired Renderware so naturally they didn’t want to put money in their rival’s hand and thus create their own engine)? I can’t imagine Activeworlds Inc. having time for that…
Pretty much yes. The industry standard engines are really not optimized at all for scenes containing lots of low complexity objects like you find in a virtual world; they’re designed for scenes with a few very high complexity objects. In fact, graphics cards aren’t designed for lots of small objects; the work on making something like AW perform decently is all in working around these problems.
Hopefully http://www.crytek.com/technology/cryengine-3/specifications/
I and another former citizen are placing bets that AWI will use OGRE.
http://www.ogre3d.org/
I’m particularly interested in its dedicated road & river tool. I’m a roadbuilder, you know.
If that’s the case, I’d like to see AW perform at least as well as SL, which completely runs off of the concept of numerous low complexity objects, but has far more advanced graphics and higher frame rates. I can run in the 10 fps range at 1920×1200 at 2048 meter visibility in extremely complex areas in SL. I barely manage 5 fps in AW at 200 in complex areas.
Something like the CryEngine would be a really cool step forward for something like AW. From fooling around in the editor I’ve seen that most of the items in the environment are indeed smaller, simple objects (albeit not to the same level as AW), so I could see an engine like this being potentially better than something like the Unreal engine. Although I can’t really even say what the new Unreal engine is like, my experience with it merely goes back to the first Unreal…
Whatever engine is chosen though, this is some actually almost exciting news about AW. Now only time will tell if this is actually going to go somewhere, and if it will rejuvenate the program on a whole.
The thing is, most games have a static map or terrain, in which most objects are occluded by walls, or hills, or some other static object.
AW doesn’t have any static objects, therefore, as far as the engine is concerned, nothing is occluded, and the whole scene is drawn. And they’re probably not going to be calculating visibility at runtime as that can be very, very slow.
What do you mean by a static object? SL has object occlusion. Does SL not have static objects?
There are herustics by which it can be done in realtime; perhaps SL implements them. They’re not perfect; they tend to overestimate whats visible and waste a few batches, but they get close.
But the thing is, to take advantage of this, you need your own 3D engine. No off the shelf 3D engine allows you to easily plug in visibility like this. Ogre3D comes close, but theres still a heck of a lot of code you need to write, and you’re getting close to the point where you’re better of writing your own engine.
Oh, and AW would need a way to designate a static object as well. Which is possible, but difficult, because AW objects don’t designate this.
I’ll simplify my question: what are static objects?
Static objects are ones which don’t do anything. Don’t change texture, don’t move, don’t rotate. Do nothing.
Would server-dictated object movements help with determining occlusion clientside? If an object is to move in second life, it must have a command which is executed by the server, and runs all the time as long as the server is running, even if nobody is around. I’ve noticed that SL is excellent with occlusion. I can walk outside of a room and I occasionally get a 1 or 2 frame delay where I see no exterior at all. This includes avatars and objects.
That can help quite a lot, though it’s also possible with AW style client side scripting. The main requirement is predictability; or guessing whether something will be static or dynamic correctly. If you guess something is dynamic when it’s not, then you waste a batch (Which is relatively minor); but if you guess something is static when it’s not, then you end up waiting around for a bit while the engine re-optimizes the scene, so any algorithm is best somewhat conservative.
New engine? This great news!