Newbie question...

[carbaher]

I think it is a good idea to use SDL with OpenGL. I've used it and they are quite easy to use. Anyway, if we decide not to use 3D it is also a good idea to use SDL

[Kurilka]

Blake - what do you say about it?

[Vultur]

SDL and SDL_image libraries have been used for last version of Map Builder (for now, the only program released).
Next version will use also SGE library for zooming/text writing routines.
NOTE: this library can be also used for textures (=3D graphic), that maybe can be implemented in following versions.

[Blake]

1. What about cities?

The center of the city should be shown as a sprite, which looks something like a city from most civ games.
For larger cities the surronding terrain texture would be changed into a farmland&suburbs type look.

2. Do we really need "3d realtime units"? Will they be so large on the map? Of course the quality will be good (if the models will be good), but what about efficiency?

I favour pre-rendered sprites because they look better and draw faster.
There is no real difficulty in using a combination of 3D and sprites.

3. Will it be possible to zoom/rotate map? (i mean to artitrary angle and magnification)

Zoom yes, rotate probably not.
Note that zooming generally works a fair bit better with OpenGL than software. I definitely prefer smooth zooming because changing zoom in large jumps tends to be disorientating.

SDL + OpenGL looks like a pretty good choice, altough AllegroGL could also be worth looking at. (Allegro is still being supported so is improving in quality, unlike SDL which is not being supported, altough is of a high quality...)

[Jeremy Buloch]

Well we also need to consider that people might not be able to play a game with an advanced 3D graphic engine. I faintly remember a discussion on this topic at the old FreeAC forums. OpenGL wouldn't work for 2/3 of my computers Video Cards so we need to be careful in choosing the engine. I like the Idea of having a nice engine and all. Yet we do have to consider this.
-J.B.-

[Rasbelin]

JB, get better graphics accelerators then. Overall it's still not yet that much required for systems as OpenGL drivers are available for most GPUs and it's one of the components updated quite frequently. OpenGL has been supported for many years, so there's no good reason to abstain from using OpenGL.

[targon]

BTW, 90% of amateur graphcs engines are borned dead due to low aim of their authors. Graphical accelerators now improve 4 times faster than processors. New nVidia NV30 will be (concerning their ad) as powerful as one of old mainframe Crays and have 125.000.000 transistors. This stuff is _cheap_: beautiful 3D engine may run even under low-end but _new_ cards with reasonable FPS. Human mind recieves 90% of its information via eyes, so why you so disregard you would-be gamers? If you think that eye-candy will harm gameplay, look at Homeworld for expample and think again. Again, IMHO, we may target high-to-medium range chips-of-the-line, as upon release of this game, this chips surely will be outdated... This is't dark predictions, it's realism. I'am fascinated by modern 3D tech, especialy vertex/fragment programs (a.k.a. shaders), they may be used for preforming mundane task as shading, shadows, motion blur, env. mapping, IK etc. _at_ _accelerator_ _side_. For example, nVidia says in one of its developer's presentaions: "if your ROAM algo runs to slow, forefit it. Graphics board is smart enough to draw unoptimized landscape faster than your ROAM may optimize it"...

[Kurilka]

Yes i agree with you 100%, but the point is:
do we need real 3D graphics in our game? As i see it evrything could be quite good imlemented as sprites (quasistatic landscape for example) maybe the units are the almost onle possible part to be realized as pure 3D model.

[targon]

No way! Why you need to do _independent_ drawing algorithm for landscape while engine itself still needs 3D acceleration? BTW, you forefit some nice ideas, like water reflections etc.
I'm artist here, not a coder, but I'm somethat competent in OpenGL (nothing like real StPDT coders, of course) and may help you via ideas, code snippets, patches, come coding etc if your would't abuse C++ usage (I hate C++, espacialy STL, while highly tolerant to C).

[Kurilka]

Mmm... maybe your're right. But what about Vultur's map builder and so on? And how do you see landscape modelling now?
I'm still still not sure if we need "real" 3D world.
BTW what's wrong with C++? Is it a new flame war?
For such a complex project we NEED C++, C sucks in big projects, it will be quite easy to get lost in the source code

[targon]

OK, you may use pretty flat world and still draw it using 3D engine becose it's simply faster. Nowdays, graphics boards are built around 3D rendering pipe and do texture mapping much faster than old good blits.

No, use, language you want, even Interkal if you desire, but without me. I still will be happy to be StPDT artist, in any case. Coding is't my work, BTW.

[Kurilka]

Heh , i wanted to write something about in my previous post. Yes this sound reasonable, but the question is : what parts of the game should be implemented as "real" 3D models? And not just simple textures on a plane or something like that...

And about languages: Russian is my favourite one

[targon]

OK, I speak Russian too, but do't use it in comments. BTW, every non-trival game needs some scripting language, and it's best place to use OO features, as C++ incomplete and clumsy object stuff is _really_ hard to export to any script.

[Kurilka]

"C++ incomplete and clumsy", why are you so sure?
The main reason (i think) when people get a bad experience with C++ i that it is a fairly complex language. But i don't want to raise a flame war, don't push me to it
As for scripting language i recommend Lua (though Python is more popular). And where do you see scripting to be used? (i know not so much about this stuff, i didn't use it)

[Rasbelin]

IMHO, we may target high-to-medium range chips-of-the-line, as upon release of this game, this chips surely will be outdated...

That's my opinion too. If we talk about OpenGL, I could add that even a nVidia Riva TNT GPU will be good enough for OpenGL support and that's nowadays a low end GPU or more like outdated, so it's not a concern in this case.

Yes this sound reasonable, but the question is : what parts of the game should be implemented as "real" 3D models?

I'm not an expert, but I would say the choice is between units only or units and terrain. Maybe something else too. Thoughts?