Transport Tycoon 3D

[robo]

If you think 3D can do it, look at the demo of Sid Meiers Railroadtycoon. Graphics are relativ pretty, but you probably need a highend computer to show 20-30 trains, the maps are small with limited zoom level. Gameplay isn't really interesting, because it reproduces the same mistakes as most of the other transport simulations, which are for example missing real passenger destinations (real= people go to work, supermarktets, school and entertainment buildings) or a missing system of supply and demand to direct the flow of goods, although there are some good ideas and features in the game.

[gix]

[...] With respect to graphics, I really like the hand drawn graphics (not in the last place because they are appropiate for a game from the 80's). Compared to generated 3D rendering, there really is no competition imho. Too bad it is so much work. [...]

Just look at sc4's approach. The detail does not lie in the models, but in the textures, just as it is now.

[Wezz6400]

Personally I really hate pure perspective 3D for such kind of games because it makes interacting with the world harder and it's impossible to make it look good and detailed without *severe* performance impacts.

I disagree. You just have to think it through and do it right with techniques such as LOD. It can be done, though your way might be easier.
As for interacting, a 3D game can use a grid system just like 2D TTD does now. I'd use a smaller grid, but still big enough for easy (rail)road building, I think that would work much, much better than the way Railroad Tycoon did it.

[gix]

Personally I really hate pure perspective 3D for such kind of games because it makes interacting with the world harder and it's impossible to make it look good and detailed without *severe* performance impacts.

I disagree. You just have to think it through and do it right with techniques such as LOD. It can be done, though your way might be easier.
As for interacting, a 3D game can use a grid system just like 2D TTD does now. I'd use a smaller grid, but still big enough for easy (rail)road building, I think that would work much, much better than the way Railroad Tycoon did it.

People like those working at Maxis think otherwise. Just compare sc4 with the new sc societes. And read the developer diaries for sc4 about the engine. (Though I give you that OpenTTD might not have such a high object level as a large-scale city in sc4).

[CobraA1]

People can't even get themselves together to make a complete set of non-3d graphics, so the 3d graphics will never get done.

I'll have to agree with this. 2D is apparently hard enough to make a complete set of graphics for - and 3D won't be any easier.

Sure, it's a great idea - but how are we going to get the artwork for it?

but could you please forget about doing this in C# (as it is interpreted bytecode (like java is)

Both Java and C# do in fact compile all the way to machine code. C# uses JIT, and Java does it on the fly with Hotspot. Neither one of them stay in bytecode very long. I use AOI, a Java raytracer, and it renders just as fast as any C++ raytracer.

The idea that everything is done in purely interpreted bytecode is a long dead concept. Bytecode is merely a stepping stone on the way to being compiled for all modern "interpreted" languages.

[DaleStan]

Bytecode is merely a stepping stone on the way to being compiled for all modern interpreted languages.

Are you sure about that? Or is Perl not modern?

[CobraA1]

Or is Perl not modern?

The last major revision of Perl (Perl 5) was thirteen years ago. That's longer than C# has existed (and C# is already 2.0, with 3.0 coming very soon), and Java 6 was released less than a year ago. Take a wild guess what my response will be.

When the next major version of Perl is released, it will getting JIT via Parrot. Then I'll consider it modern.

Anyways, back on topic . . .

I really don't think anybody around here has the time to implement and create the art for a 3D OpenTTD. I also think there are more important things to work on, and I also think that there's not really much of a benefit other than eye candy. I think OpenTTD works great in 2D, and IMHO it should stay that way for the moment.

Maybe in the distant future? But not now.

[Expresso]

Ok, apparently I was wrong about C#. However, I still can't help but wonder how you are going to keep the performance up to par, as 3D is going to require floating point calculations (which is slow), while you need that processor to be able to handle thousands of non-static game objects.

As for why I'm all for 3D, I have had multiple stations in big towns which I couldn't see clearly anymore (even with transparency on). I, for one, would find it pretty useful if I could change the camera position in situations like that.

[CobraA1]

I still can't help but wonder how you are going to keep the performance up to par, as 3D is going to require floating point calculations

This is very true, and plagued the very early days of 3D.

However, at one point graphics card manufacturers realized it was a problem. They then decided to accelerate it on their graphics cards. It was called "hardware transform and lighting" or "hardware T&L" - and the rest is history . A modern graphics card can perform these calculations on the order of billions per second.

[exe]

This is SharpTTD by me, not TTD3D.

At first I was trying a straight top-down port. But it was then very hard to integrate the game for even simple train movement. So the current approach is to go window-by-window, rewriting the game logic (using all algorithms and code fragments from openttd). Game now has three very distinct parts - game model, gui windows and switchable viewports.

[prissi]

Honestly, this and all the other attempts clearly show, that the missing graphics are really the main problem. I admire however you effort in rewriting all the routines; although I would have used one of the other codes. (Selfishly I would point out that simutrans already has a distinction between backend and display.) Nevertheless Ottd would also profit from a clear distinction between objects on th map and their graphics.

However, seeing all those projects, I wonder if cooperation is not essential. Especially for the models you need some artists. Teaming up is usually a clever idea to get things done.

And probably, if one start to reimplement the game in 3D, then the copyright problems are best solved using not the old graphics.

[gix]

[...] Especially for the models you need some artists. Teaming up is usually a clever idea to get things done. [...]

Well, models are created anyway for 32bpp-sprites. This also fits in well for textures for fixed-angle 3d.

[exe]

Perhaps the data packs (xml and graphics) could be done directly on some wiki pages, and then the game would download them on compile or even run time. http://sharpttd.top100.net.pl/wiki/inde ... imates.xml

Well, models are created anyway for 32bpp-sprites. This also fits in well for textures for fixed-angle 3d.

But models for realtime 3d should have tens of triangles, not tens of thousands.

I would definitely like 3d graphics to be looking almost like original transport tycoon. They could be very low poly (detail comes from textures) but instead showing massive parts of map.

[gix]

[...]

Well, models are created anyway for 32bpp-sprites. This also fits in well for textures for fixed-angle 3d.

But models for realtime 3d should have tens of triangles, not tens of thousands.
[...]

That's the whole point. I don't aim for "real-time 3d" you speak of. When you use fixed-angles in a sc4-like manner models don't need lots of triangles because most detail comes from the fixed-angle-textures (which also helps tremendously with performance).

[exe]

[...]

Well, models are created anyway for 32bpp-sprites. This also fits in well for textures for fixed-angle 3d.

But models for realtime 3d should have tens of triangles, not tens of thousands.
[...]

That's the whole point. I don't aim for "real-time 3d" you speak of. When you use fixed-angles in a sc4-like manner models don't need lots of triangles because most detail comes from the fixed-angle-textures (which also helps tremendously with performance).

Ah sorry, I have missed that post on page 2. But I don't think that it is worth making everything just for 4 angles.

[CobraA1]

That's the whole point. I don't aim for "real-time 3d" you speak of. When you use fixed-angles in a sc4-like manner models don't need lots of triangles because most detail comes from the fixed-angle-textures (which also helps tremendously with performance).

Well I haven't played SC4 - but wouldn't fixed angles pretty much nullify the need for 3D? You could just make a sprite for each angle, like the earlier Sim Cities.

[gix]

That's the whole point. I don't aim for "real-time 3d" you speak of. When you use fixed-angles in a sc4-like manner models don't need lots of triangles because most detail comes from the fixed-angle-textures (which also helps tremendously with performance).

Well I haven't played SC4 - but wouldn't fixed angles pretty much nullify the need for 3D? You could just make a sprite for each angle, like the earlier Sim Cities.

Well of course you can do like everything with sprites only, but many things need a tremendous effort to pull of. Just look at how many sprites are already in use, and again and again people run into the problem that they need even more sprites for special things. And of course you don't get hardware acceleration and other stuff with 2d sprites.

[CobraA1]

Well of course you can do like everything with sprites only, but many things need a tremendous effort to pull of. Just look at how many sprites are already in use, and again and again people run into the problem that they need even more sprites for special things.

That's going to be a problem no matter how you do your graphics; that's just a part of how game development works. Most major game producers have more people working on art than development.

And of course you don't get hardware acceleration and other stuff with 2d sprites.

Not true: A sprite is just a billboard with transparency. I've seen a few 2D games remain 2D and switch to hardware acceleration. It's entirely possible to port the game to DirectX and retain the original graphics.

[prissi]

Thats imho the way Transport Giant did it. Because the smooth zoom was surely done by Direct X.

[Zephyris]

Just thought you might like to see the mock-ups i made before... 3d graphics of buildings at this scale is simple - and can look very good with generic bump-mapped textures and shadowing.