Transport Tycoon Forums

I don't think it is a good idea to start using 3D accelrated hardware. It makes it harder to make it cross-platform. It's also a much different way to make graphics. There are some people, who have spent A LOT of time to make new graphics. I think it would be a waste not to reuse that.

3D would be great, but this is not a game, which is about nice graphics and sound, but it is the gameplay

I don't know if this is on any wishlist, but anyway here is my idea (and the reason for itt):

1) TTD AI sucks
2) Main devellopers should focus on creating the game engine
3) Hardcore players often have a better playing strategy than the devellopers
4) Let those hardcore players (= all forum-members) design their own AI and play against it (or even AI vs AI wars to test your own AI against some others).


=> So, make it possible to insert our own AI's in some way. (C-evo, which is a similar remake of Civilization, does that).

I think those hardcore TTD players you describe can't program AI.
Otherwise they would be the developers and therefore just as good TTD players as the hardcore players.

PJayTycy wrote:I don't know if this is on any wishlist, but anyway here is my idea (and the reason for itt):

1) TTD AI sucks
2) Main devellopers should focus on creating the game engine
3) Hardcore players often have a better playing strategy than the devellopers
4) Let those hardcore players (= all forum-members) design their own AI and play against it (or even AI vs AI wars to test your own AI against some others).


=> So, make it possible to insert our own AI's in some way. (C-evo, which is a similar remake of Civilization, does that).

Do you have any idea of how hard it is to program a good AI?
It doesn't look like it. I have programming experience and I have no idea of how to program an AI (good or bad).

In the case of OpenTTD, we can probably divide this into several subproblems:

-Placement of stations for the most profit
--Can probably be done using statistical methods

-Placement of rails
--The current algorithm will probably be replaced with A*.

-Creation of networks
--I'll have to dive into my data stuctures and algorithms book to create an minimum spanning tree algorithm to determine which stations to build between when creating a network. Hopefully, we'll be able to make the AI smart enough to create networks instead of individual station pairs.

-Creation of intersections when rails intersect
--Will probably be the most difficult part of the AI, it has to determine which intersection to use depending on how many rails intersect and how much space it has available to build the intersection.

-Using the tactics of hardcore players is the best way to create a good AI.
--Hardcore players and developers should be working with each other, not against .

-Consider pluggable AI
--

Damn, Just had to shoot down my voxel idea, But I'll rant on anyways.

Hmm, RA2 never ate hideous amounts of memory on my comp, And TS only lagged when I built thousands of jumpjet troopers and set them on a patrol loop...

I could easily make voxels for this project though, I already have prior experiance with the voxel format, They're drawn almost the same way as people on these forums have been drawing their trains/carriages (Pixel by pixel)
And they dont have to be super-dooper big voxels, You could make them the same dimensions as a normal TTD sprite.

And voxels are very small if done right.

But then again, You guys are the programmers with all the knowledge, I'm just a modder. *bows down to the mighty programmers*

[/rant]

I vote for leaving it alone. 3D is really another game, I think that's what one of the other replace-TTD projects on the forums is supposed to be.

About AI. It would be nice if the AI could handle refit.

krtaylor wrote:I vote for leaving it alone. 3D is really another game, I think that's what one of the other replace-TTD projects on the forums is supposed to be.

No no no, You leave everything sprites, (Buildings, trees, bridges etc...) but then make the vehicles voxels,

Bah, I wont press it any furthur if it looks like people are against the idea.

At least give it some thought though, Each voxel for a train would only need to be about 20x10x8, Current 2d sprites could easily coverted, of course you ould have to extend the sprites along the Z axis, Ah well.

I can always dream.

BTW, Whats the minimum specs expected to be?

Making voxel engines isn't done too often, except in specialized fields such as medicine. I'm not sure any of us know how to create a voxel engine .

Ah well, Somone might know,

Or it can just be left for later, Or a for an off-shoot of Open TTD.

Now that might just be a good idea...

Anyone know any good websites to learn C?

Bjarni wrote:Do you have any idea of how hard it is to program a good AI?

Yep, that's why I made that wish. Programming a good AI is very difficult, takes a lot of time ...
It was only a proposal to remove that burden from the main programmers, but if they don't consider that as a burden and enjoy writing an AI, please do so.

Could it be possible to implement railway track with different speeds (with different prices). E.g. up to 100km/h, 200 km/h and for high speed trains?

i like that idea, but actually the solidness of the track also determines how heavy the engine can be. maybe more realistic if the train's speed is restricted by the value (engine's weight)*(maximal speed) on each class of track.

I think that might be too complex to keep track up, unless you used different colors of ballast or something.

What about wooden sleepers/ties [Schwellen] without bed, wooden ones with bed and concrete ones with bed? The first only for up to 75 km/h, the second for up to 160 or 200 km/h, and the last for more ...

But I also want narrow-gauge railways, and tramways

At least for high speed tracks i should'nt be problem. Track mostly looks different to normal track.

premade selectable junctions and a way to save youre junctions so that you don't have to rebuild the same type of juncion every time

hovering teacup wrote:i like that idea, but actually the solidness of the track also determines how heavy the engine can be. maybe more realistic if the train's speed is restricted by the value (engine's weight)*(maximal speed) on each class of track.

(engine's weight)*(maximal speed)
It's far more complex than that.
The tracks are built to be able to take a max weigth on each axle and max weight/meter (the last one is mainly a bridges issue).
Speed factors are:
-A max for any given track
-reduced if the train is near max on weight (like 17,9 ton/axle on a max 18 ton/axle track)
-reduced if the train have a hard time in the curves. Think of a long steam engine compared to a DMU.

So it is possible to built a fast track for aluminuim trains with a max of 15-20 tons/axle, while a slow track can handle heavy freight trains (Big Boy weights 31 tons/axle on drivers)

It's too complex to just add or multiply these factors, so the question remains: How do we want it

If we want the 3 factors (speed, axleweight, weight/meter), I would recomment a hotkey. When that hotkey is pressed, the display would show numbers on the track. This could be a hotkey for each number and the numbers could be in colorcodes. (40 km/h = red, 100 km/h = yellow ...). The colorcodes could be changed with time, like the best of the age is always green and there can only be one green.

If you're going to do this, then you would also need to add maintenance costs, and decay of track - that is, you could build it to the top standards, but if you didn't use it for that, or maintain it properly, it would still exist and be useable, but not for high-speed running. I can envision the graphics showing the rail rusting, grass growing in the ballast, even small trees eventually, but a very slow freight train could still carefully make its way along to pick up a load at a low-production industry.