No-Go-Zones and overriding the Pathfinding Algorithm

Yes, it IS about specific movement decisions - that's exactly what you said:

a 'good reason', and 'control the units to move exactly like you want them sounds pretty specific. If you need to get specific, you need to add detail to you decisions, ie, extra waypoints.


Also, don't forget that there will be formations - if you want a group that sticks together in a particular pattern as it moves, you'll be able to set up a formation. we don't have any information on how unit formations are going to work yet.

That you can put waypoints to get around the issue is 100% irrellevant as that requires knowledge of the issue beforehand, which you are not going to have when it first bites you in the *** unexpected as some important unit in your group decided to go off and do its own thing.

You are requiring the user to work around something that should just work.

Disagree. It's not an issue, just a smarter way of path-finding. Like static path finding (working out a decent path before you leave then following it) is smarter than going directly from a to b. The direct route might be simpler, but if there's an impassable-brick-wall/passable-but-mega-steep-hill in between it's not better (you'll be micro-managing troop movements past everything yourself). The expectation is that in telling a unit to go from point a to point b, it gets there asap. You can't get a more powerful combo than path-finding that works well, with waypoints to make that path more complex than a single point. How can a unit know you want it to go a certain way without telling it so.

Path-finding is such a fundamental feature of RTS games, I can't see how improving it (imo) makes it worse (iyo).

Rubbish.

You're asking that the system makes decisions FOR you, and trying to paint it as some user friendliness issue. How is the system supposed to know that 'this unit is important' or that 'THIS particular blockage should not be traversed as quickly as possible, but THAT one should be', and make those decisions on the fly accurately without making mistakes? If you can't rely on the system to interpret your orders consistently, you will be forced to babysit everything.

It's REALLY EASY to see a unit stop, turn and start heading in the 'wrong' (to your eyes) direction. If it's critical that it doesn't do that, you'll already be watching those units - you grab it and tell it not to. If it's not critical, you watch it do its thing and notice that it comes right up with the other units on the other side of the blockage.


And you're still ignoring the formations feature, which would alleviate all the boohooing about units getting split up.

What we are saying is, units might follow your waypoints in an unexpected order, not that they might skip some. That unexpected order is a direct result of a non-context-sensitive step by step pathfining mechanism, as you describe, it combined with the flowfield.

no

and no...

We don't want our units to do things (in certain situations) which we did not tell them! That is a huge difference!
Also that formation thing is rather a workaround than an actual solution, because it is not solely a formation's propose to keep up the cohesion. You need ways to do that without the need of messing with formations.


Reconsider this post and you might see the problem:

I agree that if I select a bunch of units, I expect them to stay together while moving.

If I've wanted to see them going in separate way, I would have separate them in the first place. There is no arguments about it, that's how it should be in any RTS.

Real example, supcom map :



In the bottom right corner, there is a very small passage :



It's usual to put walls there to block it. Now, maybe you kill the engy making the wall and it's not completed.

If I ask my army to go through the small gap in the wall to raid the back of my enemy base, I want them to ALL go through, even it's in a small queue that will take time.

Because if half of them decide to go through the middle passage (in the middle of the mountain, bigger gap), it defeat the whole purpose of my order (raiding the back).
Not even talking of the fact that my enemy army is probably there to kill them, and seeing where they go in the back will tell him my plan and intercept my sneaky attempt.

But all this discussion is highly hypothetical, we don't know if "splitting" will be a thing with unit movements.

Right, that's why I would like them to tell us something more or at least this specific thing. Because from what they said in the stream it is a justified concern.


btw.: Just to make this clear: I am not arguing against the flowfield itself. On the contrary, everyone hates bad pathfinding and this is a great piece of digital technology to finally bring this scourge of the RTS genre to it's end. BUT I am afraid of some of the consequences that come with it - that is in particular, that my units might (in certain situations) choose routes that I did not expect and thus cause a loss of predictability where predictability is needed, which is a bad thing for a game that wants to be competitive. Of course not in all situations it is important that I can predict the exact path my units will take. But in some it is vital. Thus a way to force units to take this very exact path I want, might be needed.

On the topic of Theta Passage...

Just put a move order directly in the middle of the passage, that tells your army that at some point in time you want every darn unit to be in the passage. Then queue up a second order on the other side that you want them to mass at.

There's no cohesion, there's no sliders, there's no units going the long way around.

How much extra micro or APM did it cost the player?

• One.
  
  Single.
  
  Click.

Extra micro.
Clicking the middle of a hole slightly larger than one unit sounds really intensive for a game where you have to fight on multiple planets at the same time.

I actually don't see any case where I want a GROUP of selected units being split.
I want them split, I make 2 groups, not one.

@ BulletMagnet: Furthermore PA is not SupCom and SupCom did not have the flowfield thing. So your argument does not apply, because the flowfield is what is causing all the (hypothetical) trouble.

It does apply, because flow field will still tell units to go to points in the order that you've given them.

Even more, I would say that if a path is completely blocked (wall, wrecks, whatever), I surely don't want them to find an alternate route that can potentially lead them into the enemy defenses or superior army.

And of course because I don't want to play hide & seek with these units if I do the order, focus on an another planet, and find them 10 km from where I was hoping to see them just because they find it smart to go around an obstacle.

As a general rule, the computed path shouldn't deviate more than X degrees from the most direct path between my units and the destination.

No, you just don't see the problem... and I need to repeat myself... meh:

Of course they will follow your waypoints in the right order, but not necessarily the path you had in mind. Why? Because the flowfield might cause them to take a less obvious route because in the end it is more time efficient. But this causes loss of predictability and in the end may lead to critical tactical failures. So no, it does not apply.

Wait, you realise that it would be a nightmare? Units would be blocked in U shape except. They could not avoid lake, trying to swim.
There is no rts that actually have that behavior.

A good pathfinder should find his way out of a maze, with the fastest way it could find.
It should not run into the first wall because you couldn't not figure the way out.

SHIFT + right click to set middle waypoint FTW

How it's handled in FA is perfect : They go around logical things (like lakes) but don't try to be too smart -and annoying- either (the theta example).

Of course the lack of flow field make the units bumping into each other, but that's not the point.

I'm just saying that unit splitting or going around half the planet to reach a way point 10 meter in front of them is not wanted.
Not that they should be stuck in every obstacle

I think that fall in the "imagine the worst implementation possible" category

Of course. That's an extreme example, but the key point is : Group of units should never split by themselves.

Hey guess what the old system used to split up units as well so nothing lost.

Would formations fall under that category cohesion cause there is no formation done yet.
Let him finish it first they did say it is not finished yet and they did find a bug during the live feed

We all know that everything in this thread is speculations.
All the threads in all the forums about PA are speculations