Here is how bad the movement code actually is, and it's not just "pathfinding"

[Elmar Bijlsma]

Originally posted by Redwolf:

For what it's worth, I have my expectations about how much and what part of this problem is fixed in 1.03. I won't post here because it's too late to influence 1.03 and I don't want further flamewars. But I am willing to mail my expectations to somebody now so that I later have an independent voice telling me whether I was full of it or whether I predicted accurately. Would be an interesting experiment. Any takers?

And if proved right, we will all line the streets for the ticker-tape parade held in your honour.

[Huntarr]

Originally posted by Moon:

First of all, "the patch" only refers to Paradox customers. If you bought a Paradox version you will need it to sync up with our Battlefront release version. Therefore it's not quite correct to speak about the patch as if the game was out already. It's not and our release version will be complete without a patch.

Secondly, the list of issues that were addressed in the past 5 weeks includes something like 300-400 bugs from our bug tracker. It's impossible to list them all. It would be fairly useless, too, because unless you know the problem the fix might not make much sense to you until described in great detail.

The way we develop games is different from many (most) other big developers. We spend the first year or two with designing the base engine, and then polish it, while the bigger labels first add the art and then the game (if any). This means that during the last 1-2 months you see the game change much more dramatically than during the first 1-2 years!

From the top of my head, some of the bigger issues that were addressed in the past weeks:

- unit pathfinding (especially for getting through small wall gaps that were blown during game)

- soldier animations

- reactive TacAI (especially for firing ATGMs)

- corrected AI scripting in some scenarios

- added resolution settings

- fixed a Dual Core bug

- optmized LODs

...and much more. It's really a huge list.

Martin

PS. However, don't take my word for it. If in doubt, grab the demo once it's out in the next day or two. That's why we release demos.

Originally posted by Battlefront.com:

Weeeeeeeelll I would not say our pathfinding is perfect. However, that Berm thing you're talking about is likely not a bad decision on behalf of the pathing system. I'd be curious to know where you were clicking. To the pathing AI the Berm is nothing more than a piece of raised terrain that can be traveled on. It therefore, quite rightly, thinks it is legitimate terrain to travel on. So you must have done something to indicate you wanted it to go up there, even if you didn't.

Either that or it is a bug. You do know that bugs can be fixed, sometimes rather easily, right? CMBO was patched 12 times, CMBB not quite as much, and CMAK a few times too. I know we had things broken in the game that eclipse a vehicle going in the wrong direction.

And anybody that would give up on an entire game based on less than one turn can't possibly be a supporter. Someone that SUPPORTS someone doesn't cast them aside so easily.

Steve

Did you use Search? These are just a few of the responses. BFC has acknowledged back in July that the pathing was being worked on.

[Lurker765]

A niggle on the facts in the quote. CMBO was not patched twelve times, only eight.

versions 1.1, 1.2, 1.3, 1.4, 1.5, 1.10, 1.11 and 1.12

Also, if you are giving him a hard time for posting his pathfinding bug by showing how they were working on it in July then it is good that he posted it again to show that it is still a problem in August. They July fixes should have been in for v1.02 and thus it is still a problem.

[Elmar Bijlsma]

Originally posted by Lurker765:

Also, if you are giving him a hard time for posting his pathfinding bug by showing how they were working on it in July then it is good that he posted it again to show that it is still a problem in August. They July fixes should have been in for v1.02 and thus it is still a problem.

To be fair, Redwolf has been making an ass of himself hinting that BFC hadn't spotted the pathing issues until he pointed them out. He's not being given a hard time for posting on the issue, he's being given a hard time because his chest beating is so laughable.

[Lurker765]

Well....if we are going to continue to give Redwolf flak for posting a detailed bug report and pulling quotes about how BFC has already fixed it we can find counter examples from BFC that indicate they believe the user is responsible for many of these issues. For instance:

"If you refuse to use the waypoints the way they must be used, you will get suboptimal results. There is no work around for that on our end. Instead, if you listen to others you'll see that you will get the results you want if you do what they do." Note this quote is somewhat out of context, but the whole reply is a bit much to quote so here is the link:

http://www.battlefront.com/cgi-bin/bbs/ultimatebb.cgi?ubb=get_topic;f=52;t=002086;p=2

From my viewpoint it looks like Redwolf posted a legit bug in great detail and then defended his posting against many counterarguments and flames. I don't see him as the instigator of this thread going downhill.

It doesn't seem like calling someone an ass for trying to give the developers solid details on a still existing bug (non beta testers don't have 1.03 yet) is a good customer service policy.

But, Elmar, I guess we sort of had this type of discussion on one of the other threads when a new player had difficulty with setup zones and I guess our opinions of customer service differ.

[Huntarr]

Originally posted by Lurker765:

Also, if you are giving him a hard time for posting his pathfinding bug by showing how they were working on it in July then it is good that he posted it again to show that it is still a problem in August. They July fixes should have been in for v1.02 and thus it is still a problem.

no I am not giving him a hard time about posting. I am pointing out that there is no excuse for starting another thread after there was more than a dozen threads clearing stating pathfinding issues. I am also calling him out that he stated that BFC has never made a public statement acknowledging the issue. He is clearly a veteran of the boards and it could be forgiven of a Newbie not understanding what search is for, but he is clearly trying to state that he and he only found this bug. I mean come on.

[Elmar Bijlsma]

Originally posted by Lurker765:

Well....if we are going to continue to give Redwolf flak for posting a detailed bug report and pulling quotes about how BFC has already fixed it we can find counter examples from BFC that indicate they believe the user is responsible for many of these issues. For instance:

"If you refuse to use the waypoints the way they must be used, you will get suboptimal results. There is no work around for that on our end. Instead, if you listen to others you'll see that you will get the results you want if you do what they do." Note this quote is somewhat out of context, but the whole reply is a bit much to quote so here is the link:

http://www.battlefront.com/cgi-bin/bbs/ultimatebb.cgi?ubb=get_topic;f=52;t=002086;p=2

From my viewpoint it looks like Redwolf posted a legit bug in great detail and then defended his posting against many counterarguments and flames. I don't see him as the instigator of this thread going downhill.

It doesn't seem like calling someone an ass for trying to give the developers solid details on a still existing bug (non beta testers don't have 1.03 yet) is a good customer service policy.

But, Elmar, I guess we sort of had this type of discussion on one of the other threads when a new player had difficulty with setup zones and I guess our opinions of customer service differ.

Because you have difficulty reading, I'll try again, with bolding.

He's not being given a hard time for posting on the issue, he's being given a hard time because his chest beating is so laughable.

[Rocky Balboa]

Posted by Redwolf : But I am willing to mail my expectations to somebody now so that I later have an independent voice telling me whether I was full of it or whether I predicted accurately.

Red, The whole your digging is vast and wide indeed. Please save us the theatrics of climbing your mountain of vindication and proclaiming " I Redwolf predicted that there will still be pathfinding and AI issues after patch 1.03". If you really are a programmer of any stripe then you will of course know that a simulation this detailed will never be perfect and that there will always be things that need to be fixed. CMX1 still has issues to this day ....

[Michael Dorosh]

Originally posted by Lurker765:

But, Elmar, I guess we sort of had this type of discussion on one of the other threads when a new player had difficulty with setup zones and I guess our opinions of customer service differ.

Just a word to the wise. Elmar doesn't provide "customer service". While he may have opinions just like all of us Moon, Madmatt, Kwazydog, rune, Steve and Charles set company policy and make official representations of that policy. The comments of the little fish should be taken with that in mind. Real issues vis a vis matter of policy are best discussed with BF.C in private if there is truly a large issue.

I have a great deal of respect for Redwolf and think he has done a great service for the community by attempting to bring issues to light and discuss them respectfully. I've been fortunate to experience patching and testing from both inside and outside perspectives; I can relate to frustrations felt (at the time) when lobbying for changes in a CMAK 1.04 patch which didn't materialize. Looking back, I can see more clearly now, years later, the reasons why that never in fact occurred, and hazy memory had obscured the fact that some issues I thought were never addressed by BF.C in fact clearly were, even if not to everyone's total satisfaction. I certainly wasn't happy at the time with the way the conversation went, and like Redwolf's thread, the discussion degraded to the point of insults, which was silly, since everyone had the same endstate in mind. I suspect Redwolf has the same general goal as anyone else on this forum does. Why it should descend into name-calling then is anyone's guess. I suppose ego is a tough thing to dial down through a keyboard and mouse.

Redwolf, I find the best advice in these situations is patience. I think at this point any further discussion will simply be labelled chest-beating - and ironically, that would refer also to the act of calling someone else a chest-beater. Rest assured your concerns have been noted; coming from a luminary such as yourself who has been with us so many years, I simply can't imagine them not being heard or taken seriously. That you took the time to post your concerns in such a detailed and dignified manner reflects well on the fanbase as a whole. I'd suggest waiting for the 1.03 patch and seeing if your concerns have been addressed - I know the effort will be appreciated, even if not tangibly, should you choose to do a comparison of findings at that time.

[ August 29, 2007, 02:08 PM: Message edited by: Michael Dorosh ]

[MarkEzra]

Michael has given some sage advice that perhaps benefits all who have been posting on this thread.

[Melnibone]

Originally posted by MarkEzra:

Michael has given some sage advice that perhaps benefits all who have been posting on this thread.

Bah! One mans sage advice is anothers pompous waffle

[thewood]

mmmm, I like waffles, especially canadian ones.

[B00M$LANG]

Originally posted by thewood:

mmmm, I like waffles, especially canadian ones.

Don't you mean their Parliament? :cool:

[MarkEzra]

I like Canadian bacon...their waffles I am not sure off. But I do like French Toast...although I can't say I have ever tried them with sage.

[Battlefront.com]

Lurker765,

Also, if you are giving him a hard time for posting his pathfinding bug by showing how they were working on it in July then it is good that he posted it again to show that it is still a problem in August. They July fixes should have been in for v1.02 and thus it is still a problem.

Well, that would true if:

a) we said that v1.02 was supposed to fix the pathing issues

that we did not say that v1.03 was supposed to fix the pathing issues

The fact is that we had other fish to fry for v1.02 and we said so up front. The fix list that comes with the v1.02 patch confirms that. We also said that v1.03 was set aside to deal with a couple of very important issues, one of which was pathing. I seriously doubt Redwolf was unaware of this.

Still, there was nothing wrong with Redwolf detailing a particular problem that he found. It wasn't at all useful to us since we already fixed it, but there was no way for him to know that ahead of time. That's really all it should have been... one guy posting an observation and us coming on and saying it was fixed. But for some reason a few people didn't want it to be that easy.

The fact is that we saw this overshoot bug as we were launching v1.01 and it was too late to do anything about it. It was fixed, or mostly fixed, by the time Redwolf started this thread. Whether it is 100% fixed right now I can't say for sure. Pathing is a very difficult thing to get right for a game like this, so I'd not be at all surprised to find out that we have to revisit this issue for v1.04. And with that...

Redwolf,

1.03 is the first, and possibly only, major gameplay polishing patch.

You are half correct. It is the first major gameplay polishing patch, but it is not the only one. Since you've missed me saying that I guess it's worth repeating. We expect to patch at least as many times as we did for CMBO which, as was pointed out above, was 8 times (I forgot we skipped some numbers!). I don't know exactly what the next few patches will bring to CM:SF since we handle these one at a time, but I can say for sure that v1.03 is not the end of it. I do think it will likely be the most important patch (on the whole) of any CM:SF will ever see, but that's more because of the functionality not because of the quantity of fixes.

Oh, and just for the record (again) we consider the first version 1.01, not 1.0, because Battlefront never released v1.0. So in our minds we've only patched CM:SF one time so far and are about to patch it for the second time.

Steve

[Lurker765]

Steve,

Thanks for bringing some sanity to this thread.

[Redwolf]

I don't think this problem is much improved, if any, in 1.03.

Run the appended 1.03 savegame which has two paths plotted, one quick as my original report, one move with more waypoints (way more waypoints than you can plot in real-time mode).

Be advised you probably want to delete one of the plotted paths to prevent collisions. The slower vehicle usually overtakes the one on the fast route.

Download savegame

[Hawkmek]

to Redwolf . . . and the password is?

[Redwolf]

Originally posted by Hawkmek:

to Redwolf . . . and the password is?

No password.

[Hawkmek]

Interesting Battle you set up Redwolf. LOL

As for the pathing, was it Quick or Fast versus Move?

I believe I read somewhere that slower speeds were better for the urban environment. I know your test had no buildings, but there were many turns that one would not normally make in the open. I am guessing it comes down to managing the vehicles' speed, you can't go sliding around corners in your armored vehicle. You have to decelerate, turn, then get on the gas again. The move command is much smoother in this case with the constant speed. Remember the tortoise and the hare?

I would like to see some changes made as well, and I am sure it will improve with future patches.

Armored vehicles pivot well. Why do they have to make those ridiculous Stryker-like turns?

[Redwolf]

Originally posted by Hawkmek:

Interesting Battle you set up Redwolf. LOL

As for the pathing, was it Quick or Fast versus Move?

I believe I read somewhere that slower speeds were better for the urban environment. I know your test had no buildings, but there were many turns that one would not normally make in the open. I am guessing it comes down to managing the vehicles' speed, you can't go sliding around corners in your armored vehicle. You have to decelerate, turn, then get on the gas again. The move command is much smoother in this case with the constant speed. Remember the tortoise and the hare?

I would like to see some changes made as well, and I am sure it will improve with future patches.

Armored vehicles pivot well. Why do they have to make those ridiculous Stryker-like turns?

Let's not re-cap the previous 10 pages. There is no urban environment, or any environment at all on this map. In the savegame I provide one vehicle moves fast with few waypoints and one moves at standard speed with many waypoints. Both vehicles show problems and it is clear that "move" and "quick" at not at all handled differently by the engine when it comes to the "re-pathing" (I think that's a good term for what is going on here).

Since the old savegame doesn't work anymore I cannot compare 1.02 and 1.03 precisely. What I can see is that the extreme "turn-ass-to-movement-direction" behavior is not as prominent in 1.03 as it was in 1.02. Instead of a 180 degree wrong tumble you now get a 120 or somesuch. But the overall path followed, direction taken and speeds chosen by the engine for "re-pathing" in 1.03 are not close to realistic or acceptable.

In 1.03 there is still a lot of tumbling that is not realistic, including turning the rear to the overall movement direction (although all waypoints are basically in one direction), the sand next to the road is hit although no corners were close to the edge of the road and the speed (actual, not user-selected) is pretty much random, and way too many waypoints are required for even halfway smooth travel.

I can only re-voice my opinion that this whole "re-pathing" business should be abandoned as long as the user can only draw straight lines. If you want curves for movement, give the user bezier curves to plot. If not, then follow the path, slow down before turns and pivot on the turns. There are enough construction sites elsewhere in the code where energy is better spent. I don't see what this "re-pathing" gets us except maybe more purdy looking moves at the expense of realism. Just my already stated opinion.

[Redwolf]

I spare you the screenshots, but this behavior is unchanged in 1.05.

The savegame is simple enough that it actually works on 1.05 so you can run it yourself.

[thewood]

Here is my post from another forum.

I just had an open, flat test of Stryker reactions. I had a T-72 move closer to them. I gave it a straight line to within 100m of the Strykers. The 72 actually turned right for over 20m, then curved back to the left about 40m, getting about 40m closer to the Strykers. It then went from a straight line about 20m to left and 40m in front of its start point. From there it went straight to its destination. I have no idea in the world why it would do that.

Looks like some kind of over correction going on. The 72 was pointed straight at the Strykers when the command was given. If a Bradley was there, it would have had several good flank shots.

[mazex]

Well, the completely insane pathfinding is actually my biggest issue with CMSF so it's depressing news that it has not been fixed. As it has not been fixed after so many posts about it there has to be something fundamentally wrong with the whole movement code in the game. This has to be a prime candidate for 1.06 even though it should have been fixed in version 0.7!

I've posted a few images way back when of M1A1:s beeing given a long "single" ~30 degree movement order in a plain desert and they go 45 degrees until passing the point where they can go 0 degrees to get to the destination. In a flat desert and a long "leg" the deviation from the plotted course can be rediculus - and they expose themselves to the enemies I tried to sneak up on using MY EXPLICIT ORDER. They have to be taken out and shot in an ad hoc court marshal if they survive...

As I don't have the 1.05 patch yet (living in Sweden and bought the boxed Paradox edition), I haven't been able to test it. I guess my 30 degreee test will fail then to. Until a simle order like that is executed correctly it doesn't matter how nice the dynamic lighting is - for me at least!

/Mazex

[Drusus]

I don't know much about path finding algorithms. Still I claim there _must_ be a way to check if the movement command given by the user is a valid one in the sense that it is possible to follow the path _exactly_. If it isn't then fire up the path finding algorithm.

If it isn't possible to check this for some reason, then let there be a button "micromanage this movement command" and if the user then does something stupid, it is his problem. Yeah, I know this isn't going to happen. But the system can be so unbelievably brain dead sometimes (v1.04 at least was) that even this would be good.

Now, once more: Having a T-72 or M1A1 turn to show it's flank when you give it an order to advance straight on towards the enemy is unbelievably frustrating. The main reasons for the frustration are:

1. This would never happen in the real life. That is, the vehicle crew will not pick a longer route, nor will they pick a route exposing their flank.

2. There is _nothing_ the player can do to play better. That is, the _game_ ends up ruining the plan. The plan fails not because of clever enemy, bad luck (as in lucky shot) or because the plan was bad. It fails because of a bug in the game.

I would like to mention that the path finding is generally good. But there are those rare occasions it ****s up simple movement commands, and when it does, it does so in a manner that usually results in "alt-q".