Matrix Games Forums

Left: Pathfinding
Right: Direct jump as near as possible without triggering the system, which would result in Left
Green: Much more direct routes pathfinding doesn't find...

Here's a collection of threads that cover pathfinding. Please follow the links for more details.

BTAxis in "Nebulas, pathfinding and order queueing" wrote: Fri Mar 11, 2022 2:00 pm Yes, but you're able to jump to empty space, then jump again to empty space from there, etc. manually. If you let the system path for you, it always goes through star systems, which is really the intent. It's just possible to circumvent it if you want to put in the micro.

I don't know if it's worth trying to do anything about that.

Yes, it's worth, at least from a player perspective.

zgrssd in "[1.0.2.8> Totally busted pathfinding" wrote: Tue Mar 29, 2022 11:40 am I have utterly busted Pathfinding in one half of my fleet.

See the prime fleet. Current order is to attack the Sun of RM379 form Trandusha.
The Flagship and all self-built ships go the direct route.

The other restored ships meanwhile go the most insane route possible, across the entire Galaxy. To then expose themself to the nebula for longer.

These galactic circumventing routes seem to have apparently been eliminated. At least it's been a while I've one happening myself, while these

Cyclopsslayerr in "pls fix pathfinding" wrote: Mon Apr 04, 2022 2:25 am I still see this too often...

are still there.
This one is from the last official release

Dudok22 in "[1.0.3.7> Bad Fleet Pathfinding when trying to go to the nebula" wrote: Sat Apr 23, 2022 6:30 am My fleet was instructed by the advisor to go kill some space monsters in a nebula. After the first ship arrived all the others jumped out of hyperspace at the same time even before properly arriving which means they are now in the interstellar space just outside the system. When I click in the nebula to try to regroup them the fleet wants to jump to the other system first and then go back instead of just doing the short jump directly into the nebula.
Here are 2 pictures with the situation:





Also the jumping out of hyperspace before arriving is sometimes a problem even in normal systems without nebulas, and it means that the first ship that arrives is outnumbered unnecessarily because if takes up to 30 seconds for the rest of the fleet to reengage the hyperdrive and arrive.

and while it's not exactly the same setup (fleet vs colony ship), it still suffers from similar problem: Nebula and Pathfinding.

So please, finish the upcoming release of this week, and than tackle this issue, because it's worth, at least from a player perspective.

Edit: [1.0.4.9]

Explorer had dropped out of hyperspeed and was approaching the near-the-star-target, which made the direct approach, which was nearly identical with the new long path, possible in the first place.

Please check the algorithm as these route make no practical sense.

Edit: [1.0.4.9]

When you thought you had seen it all... note the lighter circle segment in the lower left, that's an unknown nebula.

All ships have Ion Shields.

Green is the direct path using Prepare...
Red is the jump around the nebula path using any other order.

First question: why two algorithms?
Second question: why two algorithms?
Seriously, why two algorithms?

Edit: [1.0.4.9]

Adding my stuff

Thineboot wrote: Wed Jun 15, 2022 1:24 pm Seriously, why two algorithms?

I just wanted to underscore that question in particular.

Seems like pathfinding could be tons better.

It tends to want to only use a way-point that is very close to maximum jump-range instead of a more moderate way-point that wouldn't add any extra jumps - but would be vastly, vastly more direct to target than the longest one it can find and wants to use.

So if you have a staring point, a bunch of stars in the middle right on course, and then a target just beyond the end of your current jump-range - rather than jump to those middle worlds, and continue to destination, the algo will instead go WAY out of it's way to jump to some way-off-course star that is at near maximum-jump range, and then DOUBLE BACK to your target.

This is obviously dumb / inefficient / obvious to any human eye looking at it.

Not sure what the solution is, but surely there are many that would work.

Better - often times those "on the way worlds" have fuel and are friendly. Why NOT prioritize stops at friendlies as way-points that are also more efficient / more enroute to your target? Just smarter.

[1.0.5.1] Pathfinding without Nebula

What's going on here? No Nebula, just a short jump but it can't be done directly

Ah, you have to zoom in very closely to spot this micro nebula

At least it's not getting worse

In the case of pathfinding issues, what's far more useful than screenshots are save files made at the point you notice the pathfinding issue. If you have those for the shots above, please also add them to this thread.

Erik Rutins wrote: Thu Jul 07, 2022 3:45 pm In the case of pathfinding issues, what's far more useful than screenshots are save files made at the point you notice the pathfinding issue. If you have those for the shots above, please also add them to this thread.

Seriously? Just start a game and send a ship into a nebula. [1.0.5.3], fresh game, first order to an Explorer:

Thineboot wrote: Thu Jul 07, 2022 6:32 pm Seriously? Just start a game and send a ship into a nebula. [1.0.5.3], fresh game, first order to an Explorer:

Thanks. That path does not necessarily look like a bug to me. Passing through most of the nebula could indeed take longer than taking a path around it and coming in through the other side especially as the fringes of a nebula are often less costly to travel through than the core. I'll pass this on to Elliot so that he can check it out.

Erik Rutins wrote: Thu Jul 07, 2022 8:16 pm Thanks. That path does not necessarily look like a bug to me. Passing through most of the nebula could indeed take longer than taking a path around it and coming in through the other side especially as the fringes of a nebula are often less costly to travel through than the core. I'll pass this on to Elliot so that he can check it out.

Hey - just a suggestion: Give a way to ALT choose the more direct path, plus show the ETA for each (which must include the time taken to cooldown & reinitiate the hyperdrive for each way point)...

Then we can see -- oh, direct would be X longer than this crazy path!
And, if we have reasons to force the direct path, we can do so (using say an alt right click or whatever, or a context menu option that is "direct" - but I think that would be a horrid UI - since you'd double the number of options-- +1 for every option "direct")...

It would end the complaints, give players a sense of understanding (and ergo control), and, actual control for those rare moments where time is less important than pathing....

Bonus: ETA is seriously useful info anyway...

Erik Rutins wrote: Thu Jul 07, 2022 8:16 pm

Thineboot wrote: Thu Jul 07, 2022 6:32 pm Seriously? Just start a game and send a ship into a nebula. [1.0.5.3], fresh game, first order to an Explorer:

Thanks. That path does not necessarily look like a bug to me. Passing through most of the nebula could indeed take longer than taking a path around it and coming in through the other side especially as the fringes of a nebula are often less costly to travel through than the core. I'll pass this on to Elliot so that he can check it out.

I'm really concerned about this. First you want a save, because screenshots are far less useful. Now you're saying the screenshot, which is a top view, so no angle, looks not like a bug to you, and the longer RED part is less costly to travel than the GREEN part. It's nearly a right angled triangle and I'm pretty sure everybody here would agree, that a² + b² = c² means c is bigger aka longer than a.

Since screenshots are far less useful use the save file made at the point I've noticed the pathfinding issue and take some data:

RED path:
arrival at nebula ~ 2754.05.21
arrival at sun ~ 2754.08.27
~ 98 days through the nebula

GREEN path:
arrival at nebula ~ 2754.05.05
arrival at sun ~ 2754.08.01
~ 88 days through the nebula

So the path, that does not necessarily look like a bug to you, costs ~ 26 days longer to arrive and ~ 10 days longer to travel through the nebula. By what means does this not look like a bug to you?

And this is just a quick start that anyone can do with a few clicks. And since the save file is provided as requested, anyone can check the numbers themselves.

Speed outside the nebula is 1,240,013.
Speed inside the nebula is 132,301.
That's roughly 10% (9.37%) of the speed. Obviously less time spent in the nebula is achieved by a shorter route through the nebula. I hope we can agree, that the GREEN route is shorter than the RED one.
Since this is a tech level 7 start, the jump range is half the galaxy. So there is no need for another turning point.

DIRECT path:
arrival at nebula ~ 2754.01.15
arrival at sun ~ 2754.08.01
~ 198 days through the nebula

Well, I didn't expect this result, but even the DIRECT route was faster than the RED path, that does not necessarily look like a bug to you.

Different tech, different speeds, it all will depend on the actual ships.

I may be wrong, but to me it looks like the pathfinding starts at the end, looks for the fastest route to a star outside the nebula - it takes ~137 days for GREEN and ~127 days for RED - and than continues to find a path outside any nebula. While that's an easy approach and deals well and foremost fast for thousands and thousands of ships, it's frustrating for humans beings treated like a computer opponent, when it comes to their (manual) flight paths.

With the amount of ships DW2 has to handle, the fastest way to calculate a route is essential. But there is a single human player manually dealing with some ships, and denying the obvious isn't the way you should deal with this matter.


Also take mordachais points into consideration. ETA doesn't have to be Exact, it's Estimated, and all those mentioned delays are known values. No need to dry run the whole route, just offer an Estimated Time of Arrival based of the used hyperdrive. When the ship has to be refueled and the human won't see the yellow circle, their bad.

[1.0.5.6]

While trying to find a way around the nebula this happened.
It's top view, so there is no perspective involved, there is no nebula where the fleet is heading to. And still, it finds a path to two off systems before heading into the nebula from the opposite side and leaving it again.

The fleets war ships have Ion Shields while the Fuel Tankers have not.

If you need a save drop me a line. Last time you've asked I provided one and it's not even touched to this date...

you're just not accounting for the (invisible) space-time rift

Figured this belongs here.

Just stumbled on a fleet doing extreme yoga to get from A..B: Link to save game:
https://www.mediafire.com/file/3wy545lk ... 02.7z/file

Just noticed that none of the mediafire links I've posted have ever been downloaded. Not useful?