Warzone 2100 Forums

MetalWarrior95 wrote:final 3.0 comes in few years..

What makes you say that?

Jorzi wrote:I think the polycount right now (what is it btw?) is considered very low poly by any sort of half-modern standard. Even with the badly optimized game engine, it should work flawlessly on any midrange machine. For those with really old machines, the original models are still there, until someone comes up with a way of speeding up warzone's graphics performance...

Well if you want to have greater performance then ask people to make a new engine...Warzone 2100 engine has limitations...well if original devs tried to keep and update wz2100 engine...well it would kick Total Wars engine...Total Wars engines are amazing...

Well, the total war series indeed have very good and efficient rendering and I enjoy the games, but you propably realize that changing the game engine is not something you are normally able to do.

The only successful game engine switch I know of is total annihilation -> spring, which was, after all, the original goal of the spring project. They still had to remake the maps and skip the whole single player campaign, though.

Poly count: 174

Well this weekend a friend of mine told me about a game called wolfenstien ET: a FPS allies vs axis. So i tried it out - nice game only boring after a while BUT I would guess is has a LOT more polygons than wz2100 and i know it had things like normal maps too, however it ran perfectly fine on an old mac laptop i have with 64 mb graphics card 43FPS average! it did not even ever turn my fan on!

What puzzles me is: why does warzone, having less polys and less mapping features like normal maps, slow down on that laptop to about 15FPS and it does make my fan come on when wolfenstien ET does not?

Jorzi wrote:For those with really old machines, the original models are still there, until someone comes up with a way of speeding up warzone's graphics performance...

Working on it, but using modern techniques that the really old machines will not support. The problem is that model rendering in Warzone is extremely customized, and the only modern way to do it is by using shaders. The way that games after 1999 have been doing things substantially faster before shaders were commonly supported is by uploading static model definitions to the GPU, and we cannot do that unless we redesign a large part of the game to support it.

What comes out of current master (eg like the multiplayer test builds) will now require driver support for OpenGL 2.0 or higher.

Would it be possible to just use a 3d engine that is already made?

@macuser
Maybe because wolfenstien does not use a pathing algorithm, which is processing extensive, unless you use the gpu to help with the calculations, but then you'd need the latest high end gpu with hardware support. Just trying to guess though .

@per
Are you guys planning to make a custom graphics engine for warzone? That would be nice but the time and effort needed to would be great lol.

macuser wrote:Would it be possible to just use a 3d engine that is already made?

We are considering doing that, but it will not solve any of the problems I mentioned. There are no quick fixes here.

chobibo wrote:Are you guys planning to make a custom graphics engine for warzone? That would be nice but the time and effort needed to would be great lol.

Uhm, it already has a custom graphics engine. What do you think is drawing all those pixels?

No, what I meant was if you guys were adding your own custom graphics engine to replace the original.

How hard would it be to implement a really minimalistic solid-texture+tcmask-only renderer that works on machines not supporting opengl2.0 and maintain it in parallell with the standard renderer? Is that the most propable solution you will be going for?
How bad would it be to make opengl 2.0 a requirement? I guess there are quite a few people playing this game because it still works on old hardware...

It's all a question of time (and knowledge). If there is someone who writes (or integrates) a graphics engine that does fast good-looking graphics on old OpenGL versions, we'll probably add that. As long as there isn't, 3.0+ will need more hardware for new features, 2.3 will still be available for older systems.

Would it be possible to make the new renderer a game option? So people with the old renderer could play against people on the new one?

Not without a lot of work. Especially since we use models to determine game state (and differing game state is bad), either all people need to use the same models, or we need two sets, one for graphics, one for game state. Also all graphics features (e.g. higher terrain geometry resolution) etc. would only be for looks, while game mechanics need to use the lowest common denominator, determined by the most simple version.