Page 2 of 2

Re: something I am working on.

Posted: 14 Jul 2008, 09:48
by DevUrandom
I needed the attached patch to make it work Blender 2.46, but it disables teamcolours, so I did not yet commit it.

Re: something I am working on.

Posted: 14 Jul 2008, 18:23
by themousemaster
The thing with newer, higher-poly graphics is...

Ok, you've taken the HC barrer, and "upped" it.

I assume you'd want to do the same thing to other weapons.

And bodies.

And propultions.

And perhaps terrain features and buildings.

And then, once those are all done, you probably would want to play a game with these new models; while playing said game, I doubt you will be attacking your enemy with 1 tank. Not that I think your models are bad or anything, but after making all of these cool models, I don't want to see anyone disapointed if the following stress-test turns out to kill even a high machine.

1GHz processors are easy to come by, agreed. However, if you have a game where your 60 tanks are crashing into an opponents base, and probably tanks as well, wouldn't all this start taking a hefty toll on the video card, rather the the processor? Right now WZ has the "diablo 2" thing going for it, in that it is SO processor un-intensive, that anyone who wants to can just grab it and go, old computers or not.




Basically, everything pyschopompous said. I'm just reiterating it.

Re: something I am working on.

Posted: 14 Jul 2008, 22:52
by rush2049
Well, we shall see, I want to get the cannon turrents finished, then I will stick them into the game and I will build a massive army of them and see how good it runs, right now I kinda ran into a snag with the converter, otherwise I would have tested it already.

I have at my disposal 3 machines with pentium 1 300mhz-450mhz processors, and if I can get warzone running on them with linux and play a full game, surely a new computer can handle 10x the load. Or at least I would like to assume.
I also have a some what advanced computer, an amd athlon X2 6400+ running at 3.2ghz, which is what I have to use to do most of my blender work and rendering, so we shall see what this engine can do. But I want to be sure that it isn't a built in limitation we are going to hit or if it is an artificial limitation that we are all creating because of it being originally an old game.

I fully intend to push the limits of the engine, and when I get there I want it to look stunning.

Re: something I am working on.

Posted: 15 Jul 2008, 01:20
by kage
i don't disagree with rush2049. it doesn't cause problems for anyone if he makes models that may turn out to obliterate performance on modern machines, since we'd certainly do testing for that sort of thing before we'd integrate any models into the trunk where it'd affect everyone. plus we could use some models like that to benchmark the engine (and help find the slower parts of the main loop). on the other hand, with some tweaking, the models may do fine, or we might switch over to a system/model format that supports multiple level-of-detail models, in which case rush2049's models wouldn't brutalize even a pentium 450 if it's used as the maximum-zoom model (in which case only one or two tanks could be seen on screen at once).
rush2049 wrote:Well, we shall see, I want to get the cannon turrents finished, then I will stick them into the game and I will build a massive army of them and see how good it runs, right now I kinda ran into a snag with the converter, otherwise I would have tested it already.

I have at my disposal 3 machines with pentium 1 300mhz-450mhz processors, and if I can get warzone running on them with linux and play a full game, surely a new computer can handle 10x the load. Or at least I would like to assume.
things don't always scale linearly. so a computer that has 10x the speed may run the same code 15x faster, or 3x faster, depending on processor optimizations and assumptions made by the programmers (usually for computers available at the time).

Re: something I am working on.

Posted: 16 Jul 2008, 03:33
by rush2049
thank you kage, that is exactly what I meant.

I find it similar to back when we experimented with 1.11 and 1.12, we had to find the limit before we could build right to it.

Re: something I am working on.

Posted: 30 Jul 2008, 19:24
by kipman725
also the algorithms used probobly won't scale lienarly, they may be exponential or cubic so doubling the model detail may result in 1/8 of the performance.