|
This will be my first post here on TL. Thanks for existing!
About me, let's just say I've been a gamer since before NES, and my understanding of SC2 theory and strategy is much better than my actual micro and macro.
Anyway, there's this issue I've been following, and I'm curious what you guys think, and if you even care about it. I think most of us would agree the new and improved unit pathing system in SC2 is awesome. But perhaps it is too good?
In short: the pathfinding algorithm for all units has access to and makes decisions based on global map-state information which not availible to the player.
In other words, when giving units move commands, the path isn't calculated based on what you as a player see on the map and in the fog of war, but instead it is based on what the map actually looks like - in terms of buildings, (un)blocked ramps, (destroyed) destructible rocks..
For example, say I'm playing a match on Blistering Sands. Say I don't scout. Say my opponent walls off his ramp tightly and destroys the back door rocks. Then I finally give a scout a move command into his base. What will happen is that this unit will head directly for the back door. Believe me, I tested it. I will still see the rocks in the fog of war until my unit actually reaches them, at which point they vanish.
A more common but arguably more trivial side effect of this occurs if you send a scout just as the opponent walls off his ramp. Have you ever sent a scout which refused to go up the ramp and instead just sat at the wall on the side of the ramp, often ending up promptly getting shot by a marine without giving you any scouting info whatsoever? This is because the pathing algorithm already knows that the ramp is walled off even though you haven't scouted it, so instead of trying to go up the ramp (and showing you what building are actually blocking it) it goes to the closest reachable point, which is often next to the wall at the side of the ramp. Personally I avoid this problem by giving scouts waypoints on and above the ramp before continuing into the base..
I imagine that this "psychic pathing" could be used on certain maps for stealthily scouting for blocked-off ramps or destroyed back doors from a distance, but I'm not sure how relevant this could actually be for high-level matches. According to a blue post, the development team apparently had a discussion about this and decided that it was not a bug.
Like I said, I'm interested to see what you guys think of this, or if you even think it is noteworthy. So why not add a poll? ^_^
Poll: Do you care about Psychic Pathing?This is significant and should definitely be fixed. (274) 50% Not a huge deal but it should probably be fixed. (165) 30% It's an issue, but minor enough to not affect the outcome of any actual games. (68) 12% This is completely insignificant, I don't care. (41) 7% 548 total votes Your vote: Do you care about Psychic Pathing? (Vote): This is significant and should definitely be fixed.
(Vote): Not a huge deal but it should probably be fixed.
(Vote): It's an issue, but minor enough to not affect the outcome of any actual games.
(Vote): This is completely insignificant, I don't care.
|
I think fixing it would be great. Only potential problem I can see is that they would need to change the system so that a unit's planned path is updated frequently in order to deal with previously unseen obstacles (Side effect of this: Force field would bug out unit AIs much less). It's possible that such a change would increase lag in large battles because all the units would have to constantly be re-figuring out how to move as the units around them are changing position and if things like forcefields are popping up.
|
I agree with above, it might not be a simple fix. I would like it to be changed, but at what cost... don't know.
|
first off welcome to TL, i think this is a great post something i also have noticed when playing and didn't think much of it except that it was annoying but now that you have thought it all out i have to agree with you the system needs to be changed so that the units move as if they only know what us the player knows, how ever i do agree that there are some problems with this and that FF might make this tricky, but im gonna say its an issue but its not a huge one if they fix it great if not the only advantage i see it giving is that you know if rocks are down probably 10 seconds before your units see it just by seeing what route they are taking.
|
Nice first post, but unfortunately I don't agree with you, I think it is a minor issue and won't really have a significant affect on games.
|
I get this a lot on Desert Oasis when my opponent walls in and my scouting probe stops at some random point that's not my opponent's base. I'd be fine if it's fixed, but I don't think it's a problem as-is. Every player operates under the same rules. Moreover, the only information it can give you that you don't already have is that a choke is blocked or that a rock is destroyed, and it will only give you that information once you're reasonably close to the terrain in question. By the time you 'scout' this information with the pathing bug, you've already sent a unit to the area to scout, and there's not much you can do with that information until you complete a real scout anyway.
|
Maybe not top priority, but they should definitely fix it so that a unit will walk all the way to a blockage that you haven't scouted.
|
not a huge deal but it should probably be fixed - except that it's a risky fix and could make things worse
|
You'd level the whole playing field
To play with the pathing concealed.
But it don't mean jack
Just download a hack
And play with the whole map revealed.
|
like in scrap station when someone just a-moves their army to opponents main and doesnt even care about watching them because of macro or because its a noob, and the opp destroyed the rocks for a sneaky atack and their army just goes throught the rocks :S
|
As someone who's been programming for a while, I can tell you this likely isn't an easy fix, pathfinding is a computationally expensive thing, so you make it as efficient as you can. That is likely why the system is the way it is now. My guess is that they trade off pathfinding from player maps in exchange for the fact that the pathing in general is pretty solid, and consistent across the game.
|
Where's the option for "This is significant, but a cool exploit that deserves to stay in the game"
|
You can look at it one of two ways.
Either A) a tool that you need to regularly use to keep tabs on the map state, without having to send units into danger. Players that use it regularly, and well, will gain the advantage.
B) A bug that makes things unfair for the guy that doesn't use it.
Aren't we looking for things in the former category to make the game more involved and interesting?
|
Calgary25986 Posts
On July 09 2010 02:05 Tamlin wrote:
Believe me, I tested it.
Thank you for testing it  I'm really appreciative of people who look to bring information to us, as opposed to expecting information to be brought to them. I would have never known about this had you not made this thread.
I think this is one of those issues that I would like to believe is a problem, but I can only see this causing issues in a very specific and rare circumstances on the existing map pool. With new maps, however, it could reveal a lot of information pretty readily. I would like to see it changed, but I don't feel it's a priority.
|
On July 09 2010 02:30 neurolite wrote:
As someone who's been programming for a while, I can tell you this likely isn't an easy fix, pathfinding is a computationally expensive thing, so you make it as efficient as you can. That is likely why the system is the way it is now. My guess is that they trade off pathfinding from player maps in exchange for the fact that the pathing in general is pretty solid, and consistent across the game.
Actually it is not a computationally expensive thing but a memory consuming fix. To fix it each player must store a second navigational graph over the map that holds the current path data that is known to the player.
|
I agree with Chill. I think it's an issue, but mostly something to watch, maybe. If it becomes exploitable, then it should be changed ASAP. Sounds like something that shouldn't be possible, but unknown if it's destroying things.
|
Katowice25012 Posts
I agree with Chill, but this is the kind of thing that will suddenly become a huge issue when it effects a game 5 in some tournament finals. Some player will notice his units are pathing differently, and decide rocks are gone and change his gameplan and the other guy and his fans will be pissed as all hell.
|
Definitely needs fixing. You listed very applicable exploits of it, but there are more subtle ones too - just your army composition behaves way too intelligently (magically), foreseeing obstacles which otherwise would surprise it.
Having this internal maphack in the game reduces the fun with scouting and hiding information between the players. I'm all for Blizzard making the game easier when something doesn't need choices and revolves around repeating the same routines over and over, in all games, but this particular "improvement" is illogical, and only makes the game less interesting.
|
On July 09 2010 02:26 Frozz wrote:
You'd level the whole playing field
To play with the pathing concealed.
But it don't mean jack
Just download a hack
And play with the whole map revealed.
Please temp ban Shel Silverstein troll.
|
Holy smoke, what a great first post. Thank you for having a positive/neutral attitude and not filling your post with anti-Blizzard tripe.
I do think the AI functionality should be changed specifically because it has given away that my opponent killed the rocks on Blistering Sands a few times.
|
United States12237 Posts
On July 09 2010 02:36 Chill wrote:Thank you for testing it I'm really appreciative of people who look to bring information to us, as opposed to expecting information to be brought to them. I would have never known about this had you not made this thread. I think this is one of those issues that I would like to believe is a problem, but I can only see this causing issues in a very specific and rare circumstances on the existing map pool. With new maps, however, it could reveal a lot of information pretty readily. I would like to see it changed, but I don't feel it's a priority.
This is actually pretty old information, but it's good to bring it up again. The buzzword "psychic pathing" also adds to the urgency. I think a lot of people are downplaying the potential impact that the current pathfinding system may have in some games. The way that units behave when scouting now is just not logical. You should only be able to see and act upon the information presented to you through active vision and the fog of war, and your units need to follow suit. You can unwittingly pre-empt a clever move as a result of this.
|
On July 09 2010 02:40 theqat wrote:
Holy smoke, what a great first post. Thank you for having a positive/neutral attitude and not filling your post with anti-Blizzard tripe.
Is this a new TL trend, disguising bitching with a OP compliment?
"Great post. Wow, it's refreshing you didn't have your way with me like that other guy."
"Really cogent and non-biased first post. I wish people would be non-biased and stop gang raping me in 8-FFA."
...
Sometimes the estrogen in a SC forum blows my mind.
User was temp banned for this post.
|
Katowice25012 Posts
Both your posts in this thread are really bad, knock it off.
|
yeah i noticed this very early on in the beta and got used to it. would definitely like to see it fixed though
|
On July 09 2010 02:26 Frozz wrote:
You'd level the whole playing field
To play with the pathing concealed.
But it don't mean jack
Just download a hack
And play with the whole map revealed.
Great one!
Also, I think this is a minor problem. It is loosely related to the fact that the players host the matches. I think it should be done HoN style, that would mean that no map hack would be possible and it could also enable rejoin on disconnect in team games if people wanted.
|
Thanks for bringing this issue up. I think it will change the balance on a lot of maps. This definitely needs to be fixed.
|
On July 09 2010 02:51 hello-navi wrote:Show nested quote +On July 09 2010 02:40 theqat wrote:
Holy smoke, what a great first post. Thank you for having a positive/neutral attitude and not filling your post with anti-Blizzard tripe. Is this a new TL trend, disguising bitching with a OP compliment? "Great post. Wow, it's refreshing you didn't have your way with me like that other guy." "Really cogent and non-biased first post. I wish people would be non-biased and stop gang raping me in 8-FFA." ... Sometimes the estrogen in a SC forum blows my mind.
No trend; it's as simple as the fact that the OP demonstrated how to make a constructive post without unnecessarily demonizing anyone.
|
has anyone also tested this for unitblocked ramps? i noticed this bug too but never thought about it much, and i think its a minor issue if its just about buildings, but if units would do the same, it would be a big thing imo. I cannot test it my self, since EU is still down 
|
I have encountered this pathing problem before. I actually have an example on tape. What happened was I tried to attack a front door on blistering sands, decided against it, and move to attack the back door. I moved towards the back door - told my units to attack inside the base - and they turned around and ran head-first into my opponent's army because he has lowered one of his supply depot walls.
http://www.youtube.com/user/MakhStarcraft#p/u/28/1VPWdrIsKW4
At about 8:35
Just watch how stupid my army behaves when I am busy macroing in my base unaware that they have re-evaluated the 'best' route to their destination.
|
Calgary25986 Posts
On July 09 2010 02:54 heyoka wrote:
Both your posts in this thread are really bad, knock it off.
Whoops. Sorry I didn't read that you had already taken care of it before I banned him.
|
Calgary25986 Posts
This isn't entirely related, but I have an issue with units not updating their pathfinding. Usually I will send an SCV, then lower the Supply Depot. Which causes my SCV to go the wrong way and shuffle around. I hope that gets fixed because it's really frustrating.
|
On July 09 2010 02:26 Frozz wrote:
You'd level the whole playing field
To play with the pathing concealed.
But it don't mean jack
Just download a hack
And play with the whole map revealed.
You are the man Frozz!
But as for the pathing, you've already sent the scout, so it's really not that big of a deal.
On July 09 2010 03:32 Chill wrote:
This isn't entirely related, but I have an issue with units not updating their pathfinding. Usually I will send an SCV, then lower the Supply Depot. Which causes my SCV to go the wrong way and shuffle around. I hope that gets fixed because it's really frustrating.
Agreed wholeheartedly. This is an annoyance. A very big one.
Edit: responded to Chill's comment that I didn't see.
|
I've noticed this when sending probes to scout and they stop at the sides of ramps and dont go up them.
|
|
|
Wow, this is more feedback than I expected so soon, thanks guys. ^_^
Frozz: lol, gj
misirlou, great example of how this could affect games on Scrap Station, exacly the kind of "unwitting pre-empting of a clever move" that Excalibur_Z talks about, I would think.
Makh, another great example, which speaks to the conceptually idealistic argument for fixing this issue (which, again, Excalibur_Z also mentions): a player's units should execute their orders according to the expectations of the player. Based on the information availible to you, you expected your units to use the only route you knew to be open, but since your units acted on other information not availible to you, they chose another unexpected route. Unintended, unexpected behaviour like this leads to diminished player immersion.
Chill & co, what you're bringing up is a completely separate pathfinding issue: the fact that the path is calculated when the order is given and then never updated unless another order is given. This is the part of the pathfinding that Flameberger and others are talking about, which if "fixed" could be quite computationally expensive. Obvsiously all paths for all moving units couldn't realistically be updated in real time, so finding an efficient balance between only-when-orders-are-given and real time would be a tricky problem for the developers. But, OT. perhaps this would deserve its own thread?
Viruuus: If both ramps on Blistering Sands are unblocked, units will go for whichever path is shorter which will likely mean the back door, depending on the starting position. But, importantly, not depending on scouting information...
eNyoron: Unintentional maphax, another great buzzword for this "feature"! Also, the situations on Desert Oasis etc are yet more great examples of how this can affect gameplay. Though, it seems the same argument came up in your thread as it has here, pretty much: "not gamebreaking", "when you know it exists you can adjust your actions." I guess this is kinda valid, and if nothing else a reason for not putting this on the top of the Critical Issues list.. But even if you can learn how to live with and exploit a bug, it is still a bug, and exploitable bugs are at the very least aesthetically displeasing, especially considering the amount of polish that goes into all aspects this game (well, almost all.. chatchannels etc *cough*) .
heyoka wrote:
Some player will notice his units are pathing differently, and decide rocks are gone and change his gameplan and the other guy and his fans will be pissed as all hell.
Yes, exactly! I'm just waiting for this to happen so it is taken seriously. 
|
On July 09 2010 03:39 JHancho wrote:Show nested quote +On July 09 2010 03:32 Chill wrote:
This isn't entirely related, but I have an issue with units not updating their pathfinding. Usually I will send an SCV, then lower the Supply Depot. Which causes my SCV to go the wrong way and shuffle around. I hope that gets fixed because it's really frustrating. Agreed wholeheartedly. This is an annoyance. A very big one. I agree, but that could turn out harder to fix than OP. Because pathfinding cannot be updated all the time. It gets recalculated in certain situations only. That could increase the CPU load too much. On the other hand, if done properly the OP problem can be fixed with just some slight memory usage increase, but the same CPU (if not even lower CPU usage, because only the objects within vision would be calculated).
|
United States5162 Posts
On July 09 2010 04:57 figq wrote:Show nested quote +On July 09 2010 03:39 JHancho wrote:On July 09 2010 03:32 Chill wrote:
This isn't entirely related, but I have an issue with units not updating their pathfinding. Usually I will send an SCV, then lower the Supply Depot. Which causes my SCV to go the wrong way and shuffle around. I hope that gets fixed because it's really frustrating. Agreed wholeheartedly. This is an annoyance. A very big one. I agree, but that could turn out harder to fix than OP. Because pathfinding cannot be updated all the time. It gets recalculated in certain situations only. That could increase the CPU load too much. On the other hand, if done properly the OP problem can be fixed with just some slight memory usage increase, but the same CPU (if not even lower CPU usage, because only the objects within vision would be calculated).
Well, you may be right, but it wasn't like this more than a few patches ago, iirc.
|
I don't know if this needs to be fixed, although I'd like it if it was. I have lost a lot of b-lings before because I thought they would go up a ramp and take out a wall naturally with an a-click passed the wall in, but they just sat next to the ramp and got rocked.
P.S.- I know how important it is to micro b-lings, but sometimes that isn't always viable.
|
It should be fixed but won't because of money.
|
I bet you could find 'sweet spots' on most maps, where if you place a unit there, and issues a move command to some place behind a choke you want to check, the immediate direction of movement of the unit will tell you whether the choke is blocked or not. This would make scouting the choke both safe and instant.
|
On July 09 2010 02:05 Tamlin wrote:
A more common but arguably more trivial side effect of this occurs if you send a scout just as the opponent walls off his ramp. Have you ever sent a scout which refused to go up the ramp and instead just sat at the wall on the side of the ramp, often ending up promptly getting shot by a marine without giving you any scouting info whatsoever? This is because the pathing algorithm already knows that the ramp is walled off even though you haven't scouted it, so instead of trying to go up the ramp (and showing you what building are actually blocking it) it goes to the closest reachable point, which is often next to the wall at the side of the ramp. Personally I avoid this problem by giving scouts waypoints on and above the ramp before continuing into the base..
I also find this annoying. Not huge, but should be changed
P.S. I find it funny that we are complaining that the pathing is too good
|
It's pretty annoying to send my scouting worker or zergling to the opponent's base and it just get stuck at the ramp because they walled off. I guess we just gotta be more active with checking the scout.
|
I have used this before.
When my scout goes to a base and ends up going the other way before it hits the ramp, i know the player is there without ever revealing my scout.
It's definitely a bug, at least IMO. Voted yes+fix.
|
since it COULD be abused providing an advantage - no matter how small - it should be fixed
thanks for sharing this with us
|
Pathing is a very complex thing to compute and requires alot of processor speed. I am sure whatever they have done to make the game run smoothly with hundreds of zerglings all pathing at the same time means they have made HEAVY optimizations to their pathing code.
Trying to fix an issue like this is likely at the root of the engine itself. I may be wrong, it might be easy, but I would guess it is. Fixing this would likely take more than just designers but require skilled programmers redoing a fundamental algorithm of the game. Blizzard isn't (and in my opinion shouldn't) risk changing something like that at this stage as the number of bugs that may get introduced from changing something like this would be huge.
I don't think you want to risk the bugs that may come with making a change that only minority influence balance.
|
when I go back to bw from sc2 I'm always taken aback at the pathing, and when I go back to sc2 from bw I'm equally surprised by the pathing. Anyways, it's pretty unrealistic how it currently is, but changing how it works could be difficult.
|
I know this issue from Wc3 as well.
(For those who actually played Wc3. On the 2vs2 Map Avalanche was a expo blocked by destructible rocks. So when you directed a scouting unit towards this expo position you could often tell pretty early whether or not the rocks have been taking down by the path your scouting unit goes.
However clever players could simply build something at the destructible rocks gab which gave some players the illusion the rocks weren't taken down.)
I'd actually like this to be fixed though it's of course no bug but rather a feature. And perhaps there would be some performance decreases and other issues when it's done the other way.
E.g. on Blistering Sands. You send a Probe scouting the terrains main base. On arrival the a blocked ramp is revealed. Now how should the Probe react? Should it get as close to the blockade as possible? Should it automatically try to enter via the backdoor (destructible rocks)?
Or imagine Reaper play. You order your Reaper to jump up the cliff at this one point where the Toss built a Gateway. However if the pathing decision required vision the Reaper must try to get up the cliff at that point even though it's impossible as there's simply no place.
Just to demonstrate that it's not as simple as it may seem.
|
It definitely should NOT be fixed it allos players to do thigns like hold position wall in areas to make anyone using auto-pathing go a LONG route around instead of engaging the army. And besides, it doesn't take forcefields into consideration so players still have to pay attention to avoid forcefield traps (pathing doesn't know the forcefields exist)
|
This actually may be harder to fix than you guys realize... Where does the the psychic pathing end... does it end in your vision? Does it end in a radius around the units affected? Does it matter if there is fog of war? etc etc...
I think trying to change this may end up screwing up more things than fixing, however, if it is possible to fix it for good, without significant performance dampening, then sure.
|
I just played a game where I attack moved my army into my opponents base. His ramp was completely walled off and my units didn't even go up the ramp to attack the buildings, they just kept dancing right by the cliff while my opponents army was getting free shots at them. I had to deal with banshee harrassment at my main so I just ordered an attack move without checking on them.
|
f4hy, I agree with everything you say, except I don't think the change would be _that_ huge. I think a lot of you guys are overthinking the complexity of this problem from a developer standpoint. And here's why:
The pathfinding itself need not change at all. The only thing that would change is the map data that the pathfinding is working from. Currently all units use one and the same set of map data for pathing. Changing this so that half of the units uses one data set and the other half another isn't programmatically huge. Especially if you consider that there already exists - guaranteed - a pair of map data sets (or more for bigger games) that keep track of what buldings and destructibles are visible to each player when rendering the screen, so pretty much exactly the map data that is needed for fixing the problem should already exist in the system. So again, not huge.
And btw, I work as a software developer.
However, in the end I defintely agree that there is absolutely risk involved in implementing a fundamental change like this so it is by no means realistic to expect any fix for this soon, definitely not before release. But maybe if enough of us say what we think and show how this can and does affect gameplay, maybe the dev team will bring it back up for discussion and fix it in a future patch or episode release.
|
Isn't this only an issue now because people lack map control compared to BW? I might be too much of a noob to understand how this destroys the game that your scout realizes something is walled off when you would have scouted it anyways.
|
On July 09 2010 06:45 Na_Dann_Ma_GoGo wrote:
E.g. on Blistering Sands. You send a Probe scouting the terrains main base. On arrival the a blocked ramp is revealed. Now how should the Probe react? Should it get as close to the blockade as possible? Should it automatically try to enter via the backdoor (destructible rocks)?
In fact, in SOME ways fog of war actually does it's job. If you try to build a nexus/cc/hatch at some middle-of-the-map expansion, and when you get there it turned out your opponent had already built there, your worked will go "building blocked" or whatever. The same thing could easily work in the situation you describe: rationally the probe should follow its path up to where the building blocks it and go "path blocked" or whatever.
On July 09 2010 06:45 Na_Dann_Ma_GoGo wrote:
Or imagine Reaper play. You order your Reaper to jump up the cliff at this one point where the Toss built a Gateway. However if the pathing decision required vision the Reaper must try to get up the cliff at that point even though it's impossible as there's simply no place.
That's actually an interesting problem, hadn't thought of that.
I guess if the reaper could reverse the first half of the jump animation mid-jump, i.e. undo the jump.. or just not jump and do a "path blocked" like above.. Or I dunno, there definitely are kinks like this to work out, but it is by no means impossible!
|
On July 09 2010 02:40 theqat wrote:
Holy smoke, what a great first post. Thank you for having a positive/neutral attitude and not filling your post with anti-Blizzard tripe.
I do think the AI functionality should be changed specifically because it has given away that my opponent killed the rocks on Blistering Sands a few times.
Yes, this should be taken out at all costs, and could be a game breaker.
+ Show Spoiler +I don't find anything wrong with hating on Blizzard because of all the things they have and are currently screwing up, actually 100% of people should be hating on Activision-Blizzard right now.
Explanation:
The only game I have on Xbox 360 is Modern Warfare 2.
The only PC game I have played in the past year is the Starcraft 2 Beta.
After playing Modern Warfare 2 for over 10 days total playtime, I have to say that Activision doesn't give a damn about glitches or exploits in their game, as long as people are buying their 15$ map packs so they can get their precious money.
Starcraft 2 is shaping up to be a dissapointment. No lan, no chat channels etc, I'm actually afraid to give Activision-Blizzard anymore of my money at this point.
|
On July 09 2010 06:53 Tamlin wrote:
f4hy, I agree with everything you say, except I don't think the change would be _that_ huge. I think a lot of you guys are overthinking the complexity of this problem from a developer standpoint.
My guess though is that the part of the engine that calculates pathing does not actually know anything about the users and how much they can or can't see. The user data is certainly in a different object would have to be passed into the engine.
Granted I do not know how the engine works or how their pathing is computed, but it seems like a sane design to have the engine that computes pathing ignorant to data about what gets displayed and what the user sees.
|
Personally I think that since both players will have the same pathing there's no way an unfair advantage could be created by it. However it is very annoying some of the issue that come as a result of the pathing system, but hey BW had a pathing system you had to fight through to control the command, and this one is far less cumbersome, despite its short-comings.
|
In summary, 2 separate problems:
Psychic Path Finding:
- technically not that difficult of a problem. in terms of memory or computational costs it wouldn't be significantly more expensive. it all depends on whether the engine is structured to handle this
- there will be corner cases where this just doesn't make sense in the game - ex: reapers up a blocked cliff, though these corner cases can be handled individually
- personally I think it's dumb that your units are psychic
Real Time Path Finding:
- technically much more difficult if thought of naively. the best path finding algorithm is more expensive than expected.
- solution can be to simulate real time pathfinding. for example, isolate certain actions that would affect paths. for each action, taken, check the paths affected and recalculate as necessary. this solution might even be less computationally expensive than the current solution, which is to update paths every few seconds.
- potential problem with this solution is if someone spammed supply depot raise lower really quickly, might (but not necessarily) cause problems. this can probably be solved with some sort of decay variable on the previous x # of actions.
|
Really, all players doesn't play under the same circumstances with the psychic path finding. Terran, for example, is more prone to walling in and im sure there are more examples where this is more useful to one of the two races in various setups.
Definietly not a big problem, but should get fixed.
|
Scrap Station - another map which has this problem very noticeably. Suppose one player kills both shortcut rocks and prepares/hides his army there. The other player scans the main, sees almost no army, and sends attack. What should happen is - the attack force should take the long way, not knowing that the rocks are missing, and then the hidden army could flank him. What actually happens is - the attack force magically goes right towards the rocks and finds the hidden army as if the AI is maphacking.
|
The poll needs the option "This is clearly intentional and I prefer its effect on gameplay to the alternatives".
Edit: Which, with after more thought, I don't personally think.
Imagine Blistering Sands, you break down your rocks and have your army there and your enemy attacks your natural, not knowing you broke the rocks.
Amazing attack from behind incoming.
I would love to hear HD crying out "IdrA doesn't realize White-Ra destroyed his rocks, is he going to see it in time?!"
|
i would have to disagree with scenario one since it would reveal their sort of strat possibly. but with the walling off scouting yes
|
I totally agree! (both with what you are saying and with people thanking you for not insulting blizz )
great first post, definitely think they should fix this. For anyone who doesn't understand, heres a rough drawing I made showing what he means!
![[image loading]](http://img710.imageshack.us/img710/6308/megasoop.jpg)
*edit* thats a drone, by the way
You see, if it were BW, it would simply try to go up the ramp in both situations.
P.S. Welcome to the community!! (I'm no veteran, but welcome!!)
|
On July 09 2010 02:10 Flameberger wrote:
I think fixing it would be great. Only potential problem I can see is that they would need to change the system so that a unit's planned path is updated frequently in order to deal with previously unseen obstacles (Side effect of this: Force field would bug out unit AIs much less). It's possible that such a change would increase lag in large battles because all the units would have to constantly be re-figuring out how to move as the units around them are changing position and if things like forcefields are popping up.
This, sounds a lot like collision detection in WAR ... great idea but the side effects ruin the game.
|
Interesting. I think that the fact that scouting units don't go up the ramp is rather insignificant. However, the side-effect of this psychic pathing of units automatically finding some unblocked road is extremely important. Hiding the destruction of destructible rocks can be very crucial to many strategies and many games.
|
meh, its better than any feasible alternative. I say it should stay. Everyone can do it, so it doesn't give any one player an unfair advantage.
|
i dont think they would fix it but the question is: how can they fix it without affecting other pathing?
i think its fine if they dont fix it its not THAT much of a big deal, if you send a scout to their base and its walled off, it still heads in that direction of the base, but you wont know its walled off til it gets closer and veers away from the ramp and instead sticks to the side of the base wall
|
Not sure why they don't use the strategy every game dev in the world uses for this kind of pathing problem.
You have a long term goal (to get to the point clicked) and then have local situational behaviours that come into play given some problem - like if something obstructs the goal in some way.
Under this system the scout would not know apriori that the route is blocked until it touched nose with the blocked section of terrain. At this point there are a number of simple local behaviours that can come into play - the simplest being just sit there. If of course it gets fired upon it will automatically move away with the danger avoidance type behaviour.
These behaviours are already in the game - its why attack-move will switch goals if an enemy comes in close. This would be easy to remedy - though would probably need some time to test properly in case of further unintended side effects.
Not sure why they decided to compute an absolute path given the initial (at time of clicking) state of the map and stick to it without a bailout behaviour. Seems quite shortsighted of them tbh.
A really old but quite comprehensive intro to the issues of pathing (with cool java demos) can be found here
Edit: This post reads like garbage - but I have been up all night with the beta so meh! EU Beta!
|
On July 09 2010 02:50 Excalibur_Z wrote:Show nested quote +On July 09 2010 02:36 Chill wrote:On July 09 2010 02:05 Tamlin wrote:
Believe me, I tested it.
Thank you for testing it I'm really appreciative of people who look to bring information to us, as opposed to expecting information to be brought to them. I would have never known about this had you not made this thread. I think this is one of those issues that I would like to believe is a problem, but I can only see this causing issues in a very specific and rare circumstances on the existing map pool. With new maps, however, it could reveal a lot of information pretty readily. I would like to see it changed, but I don't feel it's a priority. This is actually pretty old information, but it's good to bring it up again. The buzzword "psychic pathing" also adds to the urgency. I think a lot of people are downplaying the potential impact that the current pathfinding system may have in some games. The way that units behave when scouting now is just not logical. You should only be able to see and act upon the information presented to you through active vision and the fog of war, and your units need to follow suit. You can unwittingly pre-empt a clever move as a result of this.
exactly. it is always best for units to follow in the path most closely associated with the player's command rather extra information or AI created strategies. Ultimately this takes away too much control when the AI decides too many things on behalf of the player.
|
|
|
It might cause an issue depending on the map. Say on scrap station the rocks are cleared open by your opponent and you decide to move to his base, but your units path through the middle. Not sure if that would actually happen though.
|
uberdeluxe: Thanks, and gj on the illustration. However that only illustrates one effect of the issue..
On July 10 2010 12:48 Ryuu314 wrote:
the side-effect of this psychic pathing of units automatically finding some unblocked road is extremely important. Hiding the destruction of destructible rocks can be very crucial to many strategies and many games.
This is probably the most important and game-affecting part of this issue.. And as many of you are pointing out, Scrap Station is probably the best example of where this could clearly dictate the outcome of a game.
A lot of you guys are arguing that since both players are equally affected by this, it is not unfair... Well, sure, but what if only one of the players even knows about this? Sure high-level players will most likely be well aware of any and all exploits and bugs, but you can't really expect that from all "normal" players. Giving a player an advantage just because he happens to be aware of an exploitable side-effect of the pathing algorithm like this is not reasonable and not fair IMO.
|
I think a simple fix would be not to take any building into path finding until it is viewable through the fog of war.
|
|
|
|
|
|
EPT
