Friday Facts #113 - Better rail building

[Ohlmann]

I love that !

My main concern is how to make a railway parralel to an existing one with that system. I am obsessional enough to *want* all my dual track to be parralel.

[Tev]

I'm concerned with not having curved rails that we can build ourselves it will make building tightly compacted depots much more complicated.

You just blueprint custom curves you'll use in the design.

And for large scale network FARL will probably remain the state of art.

[hitzu]

I love the automation feature, but if I'm right and you are dropping curved rails from being built I'm concerned it will create more issues then it solves.

Let's try it first.
For me it is a very long awaited feature.

[imajor]

I was expecting a smart solution, but this is even better, can't wait to use.

One question: are you also making something for the case when the user is connecting two really far points? FARL is very comfortable for that (but always felt like a cheat to me).

[kinnom]

other rail shapes (like the S for switching close lanes)

we meet again, the wait for a feature I want.

[3trip]

Cool! Do you plan on implementing this for conveyors as well?

[infogulch]

How does this iterative approach handle pathological edge cases?

E.g. two completely separate mazes where at one end tile the first maze is optimal, but only one tile over the other maze is optimal.

[EstebanLB]

@kovarex, if a player can understand how rail signaling works, then he can trully manage 3 different types if rails. PLEASE add the S rail to switch lanes, PLEASE THINK OF THE CHILDREN!!!
The S rail is also the only thing that will allow shifting a straight track one tile to the side without much hassle, So many times I had to redo an entire path just to avoid one single oil deposit

[Zeblote]

If players don't have to mess with curve parts anymore that means you can finally fix this thing right?

[SuperSandro2000]

But do not remove manual mode because it would be useful if you want to build pixelart or a very complicated rail network and the algorithm just don't get what you want him to do.

[Xterminator]

Yet another feature that looks really interesting and that I can't wait to try out! I feel like it will make laying rail a lot easier in most cases but could maybe pose some problems if you are trying to lay out something really complicated or unorthodox and the algorithms can't figure out what your actually trying to to accomplish. Either way I really look forward to 0.13 and all the new goodies.

[Rakshasa]

Time to start creating train theme-park rides?

[sillyfly]

Time to start creating train theme-park rides?

[roman566]

I just hope that we will be somehow able to lay down tracks one by one as what algorithm thinks is optimal and what I think is optimal are two entirely different things.

[Jonathan88]

The track pathfinder is AMAZING!!!
It makes so much sense to put in but I had never thought of it before! Now we need to mod that code with the train momentum code to have self driving cars! Somebody do that please!
Great work Devs - as always you keep the amazing work coming!

[cpy]

This is basically nerd porn!

[hitzu]

Just realized that we have to have bots in order to build rails. Is that mean that it will be impossible without them?

[kinnom]

most likely not. remember that this feature isn't done yet.

[Gemoron]

I do like the automatic pathfinding for trains, but I have a practival question:

An A* track will most probably a quite curvy road. in addition, most railsystems work with at least 2 lanes to avoid crashes. is there any planning for automatic doubletrack rail building?

Just take a brief look at rootnegatives 8 lane crusher rail system.
https://www.youtube.com/watch?v=L5a0CKaG1eE
I have my doubts that A* will handle any multilane rail system. I guess we'll mostly have to stick to blueprints after all.

[-root]

lol, that's the episode Xeteth gets squished . ~!choochoo

Two things. Firstly, its great you guys are looking at rail building. We've gotten used to it, but it is pretty horrendous currently. Curves namely.

Secondly, for those scared about what this means, I think if the algorithm is smart enough, it'll become easier. The way it looks is you click the start and you click the end of the rail. So long as you know where you're aiming, the algorithm works it out. For S track, I can't foresee a problem. Even for large scale junctions, click start point and end point, provided the algorithm is smart enough, will be much, much easier.

And unless I miss my guess, based on past experience this algorithm will be well tailored to the requirements of the player.