[suggestion] Tracklayout based on a triangle grid

[Steve]

Mmmm, freedom.

Key:
Blue : tracks
Black : roads
Red : Buildings
Orange : Station

Notes:
The station follows the smaller grid on my picture, but it would actually just be around the track, where ever the track is.
The road would obviously look A LOT nicer than my model.

Edit: Attached the picture

[eis_os]

uzurpator:
well can you explain me how the gui should work on that?

I guess you would display the possible connection points of a next node?
To what angle a train can switch?

[Zuu]

uzurpator: didn't we agree to have 5 snappoints on each edge on last meeting? Not 10 or what you have on your example.

Steve: Nice too see that you like the snappoints variant.

[uzurpator]

We can have 5 - whatever

[PJayTycy]

Most of this is the same as in my trackprototype, so I completely agree with the decision on the meeting and the proposals by uzurpator/steve here.

The only difference is the question if we need snappoints in the cells too or only on the edges. I don't really see the benefit / reason for restricting them to cell-edges. Could you elaborate on that zuu / uzurpator ?


As for the triangles : You have the problem of an unnatural approach. In your first circle, you say the 12 segments are equal. That's true, but both ends of the segments are different. One side ends on a point, while the other side ends in the midle of a traingle. Does this give equal possibillities to the player?

[TossIB]

From your diagrams, it appears you are using a grid of parallelograms, which would form a basis, [...] so now I'm confused as to hy this is different from what we're already proposing.

The technics of calculating where the points are are naturally very similar. The differences are more visible for the tracklayout. Not every direction supported by the tracklayout needs to be represantable by some basic vectos of the coordinate-system. But every point important to the tracklayout (corners of the grid) is callable by integer multiplication of the vectorbasis.

You might claim that right-angles aren't prevalent in the natural layout of networks, but you haven't provided any evidence that might suggest that equilateral triangles are prevalent.

What I wanted to show is that both systems (tri and rect) are no natural systems. Rectangular and triangular grid are not perfect in representing the real world, but I think it's wrong to say, that rectangular grid is a better approximation than triangular grid, not because triangulare is better, but because both are unnatural systems, so approximation to reality can be no reason to prefer on system befor another.

The next step is to state, that triangular grid has some preferable advantages: It's more eye-appealing and it's able to provide a superior tracklayout (that's my personal opinion, for sure).

You also probably missed the obvious reason for right angles - visibility in close quarters. Traffidc approaching at a right angle is equally visible (in theory) from both directions along an axis.

No idea what you want to tell me with that. Please explain why this could be relevant to a TT-like game.

I also don't see how you "proved" the longer detours [...]

The in-game logic would count like this to not punish the player for the restrictions caused by the grid. The player himself will sit in front of the screen and trys to find the shortest way to connect Start and End. And I myself hate it when I'm not able to build a direct line, but have to build a rather long detour, just because of this stupid grid. It doesn't make any difference in the game, but it doesnt support wish to build a sophisticated and realistic looking tracklayout (real railways will never do a detour of 41,4% if the land is flat and they have no reason).

"almost any direction" is not supported, that's true.
My system is only able to provide 6 basic and 6 or 9 track-switching directions.

- note that it's not hard to find a pair of roads (whether or not they meet) that run at right angles to each other, so the angle needs to be suppported.

Truely right, but I don't see the need. There are also many connections and crossings which are not in the right angle (especially when viewed at the relevant scale). Any system will bring restrictions, but I'm not willing to accept the restriction in thinking, that everything has to fit a 90°angle!

@ uzurpator:
I'd like to see some curves, cause that's where all the problems begin. What will happen if you do some S-curves in you standart radius and want to meet the next parallel track? An if you want to meet the 4th parallel track? Do you still meet the standart distances then?

I think it's not the ideal way to smoth some edges afterwards, caus the ther will be no difference between a crossing, a single switch-crossing and a doubble switch-crossing (what are the correct terms?).

As for the triangles : You have the problem of an unnatural approach. In your first circle, you say the 12 segments are equal. That's true, but both ends of the segments are different. One side ends on a point, while the other side ends in the midle of a traingle. Does this give equal possibillities to the player?

That's right and no, not equal possibilities for all the 12 directions. There are really only 6 basic directions, the other 6 or 9 are "helper" directions that are needed and can be used in larger switch systems or anywhere else, whereever you like. The segments are larger than normal segments (in TT diagonals are shorter), but you are able to connect them to a curve (left or right) at every breaking point (in TT only left at one braking point, right at the next on). If you do your gameengine carfully, you should also be able to have tunnel-entrances, slopes bridges for those tracks (not like in TT Loco, which really suck at this point).

[ChrisCF]

[The differences are more visible for the tracklayout. Not every direction supported by the tracklayout needs to be represantable by some basic vectos of the coordinate-system. But every point important to the tracklayout (corners of the grid) is callable by integer multiplication of the vectorbasis.

You seem to be missing the point here - if you want people to change, you have to provide advantages to them changing - what you've said here is also true of a square grid.

No idea what you want to tell me with that. Please explain why this could be relevant to a TT-like game.

Because it affects the way that road intersections are built in real life, and since we should be looking to at least form a basic model of real life ... you do the maths

And I myself hate it when I'm not able to build a direct line, but have to build a rather long detour, just because of this stupid grid.

By my reckoning, a sensible system based on a square grid can build that line without detours, with three lengths of (1,1) track.

I'm not willing to accept the restriction in thinking, that everything has to fit a 90°angle!

Who said that? The way you seem to be coming across, it sounds like you're thinking that nothing must fit a 90-degree angle, which is not the same thing.

Ultimately, there's no real reason why on a square grid we can't support just about any direction, and why things have to snap to specific points in the grid - and in this respect, a square grid just makes it easier to measure. As I've already said, it should be possible to just overlay that onto a landscape formed from whatever shapes we end up cobbling together - we don't have to use the same shape in both cases.

[Steve]

Below is a city setting. It shows what a square grid can achieve. TTD couldn't and i don't think your triangle grid could either.

This game should be far better than TTD, Loco, whatever. Restricting ourselves to a triangle based world would be bad, and as Chris says, you need right angles. Right angles occur in road design more than any other type of angle, as it's easiest on average to turn each corner. Or near to right angles anyway. My city doesn't have many perfect right angles, but they are pretty close.

I think we need to see some more pictures of a triangle based grid, like my pictures, if we really want to consider it. At the moment, i don't think it'll be able to handle everything we need.

[TossIB]

I'm not good in making advertising images (and I don't have the time right now either)

I think, there is no need to build every wall every roof and every garage matching to the grid. And I also think there is no need to build every street in a way that yopu can use it and build or delete it. In fact there are many streets in a town which are much too small for a truck. These street should be build by the game engine and not be available for building and driving for the player. (These streets may follow a square grid)

The streets a player is able to build are the bigger ones (interurban). You can see them on the photos of Munich, these are the ones not following any grid.

Ultimately, there's no real reason why on a square grid we can't support just about any direction, and why things have to snap to specific points in the grid - and in this respect, a square grid just makes it easier to measure. As I've already said, it should be possible to just overlay that onto a landscape formed from whatever shapes we end up cobbling together - we don't have to use the same shape in both cases.

Maybe I misunderstood some details, let me point out:

Now it seams to me, that you want to have grid-based system that is gives the player the impression of full flexibility and freedom (like building gridless) at lest for the railroad design.

I myself think, that beeing restricted to a fairly easy grid, that doesn't allow every direction and every layout (like TT, Loco, Simu) isn't all that bad, cause you'll have less problems in adapting the track to terrain and building track (especially the interaction between player and game is rather complicated for pseudo-gridless or real gridless systems). Anyway, the system I thought of is based on the assumption of an restrictive grid. And my thesis is that an triangular restrictive grid is better than an rectangular restrictive grid.

But now it seams to me, that you are already a step further and going to implement a pseudo-gridless system. At least from what Cris is saying. Steves townscapes on the other hand look like an restrictive form of grid!?

Maybe you still want to take a look at this station setting i did some time ago:

[ChrisCF]

I myself think, that beeing restricted to a fairly easy grid, that doesn't allow every direction and every layout (like TT, Loco, Simu) isn't all that bad, cause you'll have less problems in adapting the track to terrain and building track

(especially the interaction between player and game is rather complicated for pseudo-gridless or real gridless systems)

Really? Point. Click. Click. Done. Does that sound like a complicated interaction? Perhaps we make the ends draggable so you can line up your tracks nicely before you build.

Anyway, the system I thought of is based on the assumption of an restrictive grid. And my thesis is that an triangular restrictive grid is better than an rectangular restrictive grid.

As we have seen (even if Steve's graphics aren't as smooth), it is yet to be proven

You haven't made it clear how restrictive your grid would be. In one sense, you're looking at keeping to gridlines, or joining points that require only one tile to be crossed, and in another you're proposing that you build anywhere as long as both ends are on the grid (as in the 9-axis model earlier). Ultimately, you can't have it both ways. A square grid which restricts you to 45-degree angles isn't brilliant, but it's still a little better than not being able to build perpendicular tracks - by mankind's natural settlement patterns, they become necessary. On the other hand, if you allow the freedom that you have proposed on your triangle grid also on the square one, you find no difference at all. You can construct any angle to the X axis (we shall call it A), by taking a part of the line (X,Y), where X/Y = tan A. For instance, a track running east-west might meet a track at approximately 60 degrees, where that second track is made from vectors parallel to (5,8) or thereabouts.

Ultimately, the best use we have for a grid with the kind of methods that people have proposed is one whose purpose is simply to measure a location on the map relative to the map as a whole. Having tracks running from (31,33) to (32.5, 28.9) shouldn't really pose too great a problem. If the transportation network is stored as a modification to a linked list (instead of a single pointer, or a back-pointer for double linked lists, we might have any number of pointers pointing to other track pieces you can get to), it should be fairly simple to establish the lengths of tracks, where they go, which routes are available, etc.

[TossIB]

(especially the interaction between player and game is rather complicated for pseudo-gridless or real gridless systems)

Really? Point. Click. Click. Done. Does that sound like a complicated interaction? Perhaps we make the ends draggable so you can line up your tracks nicely before you build.

You forgot about the hight. A mouse has only 2 Dimensions so you need to help yourself with a key. Landscape adjustment. Player's who want to build parallel and track at equal high but not directly together. Building a track through mountainous regions will need complicate examination of the landscape profile. After all it should be a game and not a track planning work.

I don't say it's impossible to do a good->perfect Transport Sim in gridless 3D, but the more freedom a player has the more work he has with building. You agree on this point?

Anyway, the system I thought of is based on the assumption of an restrictive grid. And my thesis is that an triangular restrictive grid is better than an rectangular restrictive grid.

As we have seen (even if Steve's graphics aren't as smooth), it is yet to be proven

There will be no objectiv proof ever. I've posted some good arguments before in several posts. After all it's still a question of taste and personal preferences.

You haven't made it clear how restrictive your grid would be. In one sense, you're looking at keeping to gridlines, or joining points that require only one tile to be crossed, and in another you're proposing that you build anywhere as long as both ends are on the grid (as in the 9-axis model earlier).

Didn't meant to say what you understood. The grid I thought of is a very restrictive grid, like Loco but based on triagles. The diagonals are longer to not brake the grid. In fact they will only be used in switches (see picture in my last post).

Ultimately, you can't have it both ways.

Agree.

A square grid which restricts you to 45-degree angles isn't brilliant, but it's still a little better than not being able to build perpendicular tracks - by mankind's natural settlement patterns, they become necessary.

90° might be a natural settlement pattern, but (on more repetition) not at the scale approximately represented by Transportation Sims like TT! Don't look at the town buildings, look at the distances between towns. Towns are represented in a symbolic way (should be kept like this). The maps you linked to disprove this argument are (except for the last) the best proof for it. The last one shows one of the cases, where an exact rebuilt is impossible as it would be impossible to exactly rebuilt the second map in rectangular grid. So both systems are unnatural and an extreme simplification in many cases. Conclusion: There is no natural settlement pattern (at this scale).

I agree with you that 90° angles are needed in urban areas. I want to suggest the following: (also posted before) Many streets in a town are two small for busses and lorrys, so they wont be available for vehicles in the game. These streets are also most common to meet at right angels, so the should in the game. Since the player cant use these streetsw, there is also no need for him to built them. Based on this a town will consist of several bigger streets usable and builtable by the player. The spaces inbetween these streets (larger than in TT or Loco and maybe a strange shape) will be filld by som sympolic town grafics, namely houses, skyscrapers and narrow streets, everything right angled. The gaps to match the bigger streets will be filld with parks, places and special formed buildings. If you look at maps of towns or some official plans for new settlements areas, then you'll see, that this is how settlement works (at least here in Germany).


After all I'd like to know if it was decided yet to choose a restrictive grid or an not viewable grid providing nearly full freedom.
Cause maybe I'm all wrong here with my suggestion and maybe you (the dev-team) are talking about most different concepts without noticing yet.

[Zuu]

(especially the interaction between player and game is rather complicated for pseudo-gridless or real gridless systems)

Really? Point. Click. Click. Done. Does that sound like a complicated interaction? Perhaps we make the ends draggable so you can line up your tracks nicely before you build.

You forgot about the hight. A mouse has only 2 Dimensions so you need to help yourself with a key. Landscape adjustment. Player's who want to build parallel and track at equal high but not directly together. Building a track through mountainous regions will need complicate examination of the landscape profile. After all it should be a game and not a track planning work.

I don't say it's impossible to do a good->perfect Transport Sim in gridless 3D, but the more freedom a player has the more work he has with building. You agree on this point?

I think I agree with you. And I am fighting Stevs asumtion that more freedome, will automaticly lead to a better game in the TE development forum.

A square grid which restricts you to 45-degree angles isn't brilliant, but it's still a little better than not being able to build perpendicular tracks - by mankind's natural settlement patterns, they become necessary.

90° might be a natural settlement pattern, but (on more repetition) not at the scale approximately represented by Transportation Sims like TT! Don't look at the town buildings, look at the distances between towns. Towns are represented in a symbolic way (should be kept like this). The maps you linked to disprove this argument are (except for the last) the best proof for it. The last one shows one of the cases, where an exact rebuilt is impossible as it would be impossible to exactly rebuilt the second map in rectangular grid. So both systems are unnatural and an extreme simplification in many cases. Conclusion: There is no natural settlement pattern (at this scale).

I agree with you that 90° angles are needed in urban areas. I want to suggest the following: (also posted before) Many streets in a town are two small for busses and lorrys, so they wont be available for vehicles in the game. These streets are also most common to meet at right angels, so the should in the game. Since the player cant use these streetsw, there is also no need for him to built them. Based on this a town will consist of several bigger streets usable and builtable by the player. The spaces inbetween these streets (larger than in TT or Loco and maybe a strange shape) will be filld by som sympolic town grafics, namely houses, skyscrapers and narrow streets, everything right angled. The gaps to match the bigger streets will be filld with parks, places and special formed buildings. If you look at maps of towns or some official plans for new settlements areas, then you'll see, that this is how settlement works (at least here in Germany).

In playerperspective I think what you are saying is OK. Personaly I have nothing realy aginst a restrictive grid as in TTD ir in your suggestion.

In a developer perspective I don't know enoght about the consequences of using a triagle grid instead of a square grid, to accept or reject your idea.

After all I'd like to know if it was decided yet to choose a restrictive grid or an not viewable grid providing nearly full freedom.
Cause maybe I'm all wrong here with my suggestion and maybe you (the dev-team) are talking about most different concepts without noticing yet.

The process in going against squared cells, which has several snap points either as a full grid or just at the borders where track segments will meet. However I think still there is room for a change, but then enoght people has to be convinced.

Actually untill the Design Document have been accepted by every developer things can be changed without to big diplomatic efforts. But the closer we get the more things will need to be changed if we change a made decission.

[TossIB]

Since you didn't contradict I don't have to reactivate my discussion instincts

For the consequences on the developper side, I also don't knkow anything, but I belive, that it won't get much more complicated (there is a solution for every problem, that's the beauty of math)

As far as I understood the Design Documents, they are summarys of basic intentions on how the game will be. The grid is IMHO a very basic thing of the game, so should the decision on it not be delayed.

It seems to mee, that you (the devs) are beating around the bush in some way *searching some cover*

[PJayTycy]

As far as I understood the Design Documents, they are summarys of basic intentions on how the game will be. The grid is IMHO a very basic thing of the game, so should the decision on it not be delayed.

It seems to mee, that you (the devs) are beating around the bush in some way *searching some cover*

Complaints about beating around the bush are not really apropriate at this time if you look at what we've done the last 4 weeks compared to the 2 years before.

[TossIB]

Yeah, you're right
Was just said to provoke some reactions, cause it seams this thred went dead without anybody of you expressing his/her approval or refusal.

Topic locked, quadrangular tiles are chosen (12112006).