Unbelievable!!!

[agusto]

And tactically speaking, wasn't it a basic practice to knock out the first and last vehicle in a column, precisely because the others would be trapped, or at least slowed?

If you do that in CMBN the colum is slowed down. Although vehicles can clip into each other they cant do so going faster than maybe an infantry mans MOVE speed.

[thebigJ_A]

If you do that in CMBN the colum is slowed down. Although vehicles can clip into each other they cant do so going faster than maybe an infantry mans MOVE speed.

Well that much is good to hear, at least. Thanks.

I brought it up, though, in reference to a statement above about tanks blocking the way not being relevant to battles. It was an aside, in other words. I should have put a quote in at that bit for clarity.

[The Teacher]

Gee, everyone has got it wrong.

Its a new secret V weapon, to save on engine power and to give the front of the Panther more cover a new hybrid was developed but only tested under combat conditions on rare occasions.

vroom vroom

[Holien]

FYI the Panther has given up trying to get through the gap. The Half Track behind it got through OK but the Panther wasted several turns trying and failing so I gave up as it is time critical and I need the Panther on the other side.

So in a way I should be jumping up and down saying the Panther should have pushed it's way through by nudging the half track aside....

I can live with it and explain it away as some crews did insane things. I think I read a quote from Otto Carius on German Tank crews towards the end of the war that did not keep the front of the tank to the enemy and turned side on getting knocked out.

It happens and while we are gods of the battlefield real men in stress make poor choices time and time again. The game does a good job of recreating the WW2 battlefield and if it can be tweaked further that would be good.

By people raising these points these things get on the radar of the Dev team and they can make the choice as to if it is a bug or as designed.

[noob]

Not to mince words, this is total bollocks.

Please mince your words in future, to help you do this, you should write a note and stick it on your PC saying "Before i blow my stack on the forums, i should wait until i calm down, and then present the issue in a polite and reasonable way".

You need to understand that CM is such a mammoth project that it is inevitable that there will be issues, in fact one should look at CM as a work in progress, with issues being raised and fixed determined by the BF teams priorities, or by the amount of comments regarding the issue, so getting angry with CM because there are flaws in it is like getting angry with the tide for coming in.

[Rambler]

Hmm, I like this tactic. I'll have to remember it for future pbems.

1) Throw tantrum on forum over CMx2 "feature" in since CMSF.

2) Have opponent come into thread defending himself.

3) Opponent gives additional info about his situation and posts screens.

4) Profit.

Nicely done .

[Guest]

With respect, how is collision detection, something that's been in games for ages, something that would prevent features being in the game? And even if it were, how is that relevant? That's like if a bunch of textures were missing, and rather than agreeing it be fixed, one argues that putting those textures in would have taken time from something else that we'd not now have. In the strictest sense it might be true (then again, it might not. It's not necessarily a zero-sum game.) but that wouldn't bear on the validity of the complaint.

An object comes into contact with another object, it stops. It hasn't got to be perfect, or even nearly so. A little clipping on the edges is to be expected (though less and less as the years go by). Running straight through 2/3rds of an object just shouldn't happen.

And tactically speaking, wasn't it a basic practice to knock out the first and last vehicle in a column, precisely because the others would be trapped, or at least slowed?

To be clear, I don't think it's game-breaking by any measure, and I think the OP got a little hyperbolic. In principle, though, I agree. It's not too much to expect. I'm still new to this game, and am enjoying myself immensely, but I've had this effect happen and it took me right out of things. I'm here for the simulation, among other things, after all. (And to be clear in another respect, I'm not talking about any "pushing" debate. Just simple collision and clipping, which I took to be the OP's fundamental problem.)

Just an opinion, no strong emotions here.

I would definitely complain about this, and ask what we can do about it, but you're barking up the wrong tree here. With respect, what you're talking about is hard. No game on CM's scale - thousands and thousands of complex potentially colliding objects, with complex pathfinding - does it. None. We approximate some of the more attainable parts of it, but we definitely don't *solve* it. If you say this is a problem that has a straightforward solution which is possible with today's processing power, you're misrepresenting it.

Note again that I'm not saying you shouldn't complain about this, just that talking about the basic principles and then saying "it's not too much too expect" is, well, wrong.

If he hasn't already, I'll ask Steve to come in and see if he can follow up on my comment. I can talk about the technical aspects of things, but "solving" the wider problem isn't what we're after. He can speak to what we might be able to do about this more specific issue.

[JRMC1879]

I would definitely complain about this, and ask what we can do about it, but you're barking up the wrong tree here. With respect, what you're talking about is hard. No game on CM's scale - thousands and thousands of complex potentially colliding objects, with complex pathfinding - does it. None. We approximate some of the more attainable parts of it, but we definitely don't *solve* it. If you say this is a problem that has a straightforward solution which is possible with today's processing power, you're misrepresenting it.

Note again that I'm not saying you shouldn't complain about this, just that talking about the basic principles and then saying "it's not too much too expect" is, well, wrong.

If he hasn't already, I'll ask Steve to come in and see if he can follow up on my comment. I can talk about the technical aspects of things, but "solving" the wider problem isn't what we're after. He can speak to what we might be able to do about this more specific issue.

To me that would perhaps be a best case solution. IE - not perhaps determining vehicle clipping and "pass through" in every possible situation. (and I am aware of just how difficult proper collision detection must be in a game like CMx2 - what some forget I think is that it tacks every bullet. Just one burst from and MG42 could account for a couple of hundred separate objects - all of which must then be checked against everything else in the game space.) But perhaps the more specifically "business critical" side of things like vehicles blocking single lane bridges.

[Guest]

Yep, that's what I would shoot for (which *would* lead to discussion about development priorities, but at least it's a possibility).

Just a secondary comment - we do track every bullet, but they're MUCH easier. No paths to consider, just a trajectory, finite lifespan, won't be "in the world" for more than a few seconds.

Complex objects that have places they can't go, *could* potentially be anywhere they *can* go, need to live potentially as long the scenario is running, but which need to be collision-checked against all / most other such objects constantly, especially while pathfinding (which needs to create potentially thousands of paths in a very short time) are going to be a huge collision burden. We do our best, but a full solution to that on CM's scale isn't possible.

[JRMC1879]

Yep, that's what I would shoot for (which *would* lead to discussion about development priorities, but at least it's a possibility).

Just a secondary comment - we do track every bullet, but they're MUCH easier. No paths to consider, just a trajectory, finite lifespan, won't be "in the world" for more than a few seconds.

Complex objects that have places they can't go, *could* potentially be anywhere they *can* go, need to live potentially as long the scenario is running, but which need to be collision-checked against all / most other such objects constantly, especially while pathfinding (which needs to create potentially thousands of paths in a very short time) are going to be a huge collision burden. We do our best, but a full solution to that on CM's scale isn't possible.

I imagine that section of code must give you headaches just looking at it !

Personally I would like to see the "business critical" collisions high up the list. Despite the exaggeration of the op - I for one would like to see bridges being blocked by wrecks and live vehicles. Even if it was only that situation as it is quite specific. Depending on how difficult it is to do perhaps bocage gaps as well and narrow streets - which I think are most relevant to normandy - but if its only the bridges then I am fine with that. Everything else I can quite easily live with.

[A Canadian Cat]

But perhaps the more specifically "business critical" side of things like vehicles blocking single lane bridges.

Yep, that's what I would shoot for (which *would* lead to discussion about development priorities, but at least it's a possibility).

Oh, yes please. Blocking bridges roads and gaps in obstacles would just awesome. I am sure it would lead to plenty of frustrations - but the right kind of frustrations. The kind that results from your opponents doing bad things to you.

[womble]

IME, blockages do occur. At least for medium and heavier tanks. Lighter vehicles do seem to be remarkably snake-hipped when it comes to shimmying through apparently too-narrow. I've seen Stuarts and jeeps squeak through the man-sized gap in the fence you can have without flattening the bits of fence either side. Larger tanks do seem much more reluctant to try. A Sherman burning on a bridge stopped any other tanks getting across (until it went "Boom" and took out the bridge in toto) once, IIRC. I suspect sometimes it will depend on precisely where the "blocking" vehicle is.

[JonS]

I've recently completed a pair of sort-of mirrored custom-QBs, set on a map whose dominant terrain feature is a bridge across a stream that is otherwise impassable to vehicles. The goal for the attacking side was - obviously - to secure the bridge and establish a bridgehead.

While I understand the confusion and surprise over having vehicles 'ghost' through destroyed and immobilised wrecks, having in-game wrecks be impassable would make any kind of bridge scenario trivially easy for the defender to win - simply park a friendly truck or other expendable asset on the bridge and wait of it to be destroyed. While that's probably(?) realistic, it doesn't make for very compelling gameplay.

Jon

(It should perhaps be noted that I failed miserably, as both attacker and defender )

[thebigJ_A]

I would definitely complain about this, and ask what we can do about it, but you're barking up the wrong tree here. With respect, what you're talking about is hard. No game on CM's scale - thousands and thousands of complex potentially colliding objects, with complex pathfinding - does it. None. We approximate some of the more attainable parts of it, but we definitely don't *solve* it. If you say this is a problem that has a straightforward solution which is possible with today's processing power, you're misrepresenting it.

Note again that I'm not saying you shouldn't complain about this, just that talking about the basic principles and then saying "it's not too much too expect" is, well, wrong.

If he hasn't already, I'll ask Steve to come in and see if he can follow up on my comment. I can talk about the technical aspects of things, but "solving" the wider problem isn't what we're after. He can speak to what we might be able to do about this more specific issue.

Thank you, I appreciate the reply very much. And to be honest I'm not a programmer, so I *don't* know how hard these things are. I'm also nearly consistently impressed with the things that have been done in this game. The only game I've seen like it was one called Achtung Panzer Operation..... something, and that not done as well. (or perhaps it was just 'rougher around the edges', so to speak. I did enjoy it. Is there a consensus on that series on these forums?)

Please forgive me if I've inadvertently belittled your work!

But I'm not quite following you on the bit about no games doing it. We're talking about vehicles bumping into vehicles (and not even necessarily the exact edges and faces of the models), not "soldier #32's left hand mustn't clip through the windowsill of the third-from-the-left window in house B".

Lots and lots of games have vehicles that don't drive through one another. I am honestly perplexed.

[Guest]

My statement was qualified: "No game on CM's scale - thousands and thousands of complex potentially colliding objects, with complex pathfinding - does it." And they don't.

You can get decent pathfinding with complex moving objects if you've got a handful of them, like an FPS, or if your paths are extremely simple, like in an RTS (and even then very few RTSes actually use real collisions or small-scale obstacle avoidance for pathfinding - even for them it's a difficult problem that's better off not solved). CM generates complex paths and allows for vastly larger numbers of individual path-altering objects than either of those.

So... sure, some games might do it. But they're not solving the problems we'd have to (and as noted, for the levels of complexity we're looking at, "solutions" aren't really practical to attempt). Any game that *would* have to solve those same problems in some general fashion, doesn't.

[Holien]

Thanks Phil for helping to explain the coding issues behind this thread. To people not versed in the arts of coding it all seems so simple.

If it was then it would be in there already.

Anyway good to get a response from someone who knows what they are talking about. All too often (and I put myself in this category) these discussions are based on very little knowledge of how difficult it is to make things happen in the game like the real world.

We keep forgetting that real life is incredibly complex and while computers and programming have come along a fair way since the ZX Spectrum we are still a far distance from getting it "perfect" with the pixels.

Perhaps if we were able to get it "Perfect" we would be in the Matrix and not playing CMBN?

[c3k]

(snippage)

But I'm not quite following you on the bit about no games doing it. We're talking about vehicles bumping into vehicles (and not even necessarily the exact edges and faces of the models), not "soldier #32's left hand mustn't clip through the windowsill of the third-from-the-left window in house B".

Lots and lots of games have vehicles that don't drive through one another. I am honestly perplexed.

My bold.

Let's look at Company of Heroes. Great, fun game. How many maps does it have? How large are the maps? The only way to create a map, with all the pathfinding issues (definite paths, definite not-a-path, possible paths, possible not-a-path), is quite difficult. Every map is small (by CM scale), and hand-crafted. Pathfinding and collision detection are correspondingly simpler.

Compare that to the CM editor. Anyone can create a 4km x 4km labrynth of bocage in a few minutes. Drop a tank at one end and give it a single waypoint at the far end and it WILL find a way to get there...if it is possible. (You may not like the route, but that's a different kettle of fish.) That is some impressive pathfinding.

IF complexity of collision algorithms is non-linear, meaning geometrically more difficult as you increase objects and terrain, then the CM series might be hundreds of times harder to solve collisions than CoH.

As far as I know, there is NO LIMIT to how many tanks I can toss down on a map in CM.

(I do know, based on experience, that there is no limit to how many tanks I can order to their deaths in CM. )

Ken

[agusto]

My bold.

Let's look at Company of Heroes. Great, fun game. How many maps does it have? How large are the maps? The only way to create a map, with all the pathfinding issues (definite paths, definite not-a-path, possible paths, possible not-a-path), is quite difficult. Every map is small (by CM scale), and hand-crafted. Pathfinding and collision detection are correspondingly simpler.

ArmA 2 has maps larger than all maps available in CM and allows battles with several thousand AI that wont clip into one another. 'Tac AI' in ArmA appears to be noteably less sophisticated if it even exists and user generated missions need more accurate AI plans than in CM though, but i think its still prooving that pathfinding on huge maps with thousands of AI can be done without clipping and with hardware available to the average customer today.

On the other hand it must also be mentioned that at least in my experience AI vehicles in ArmA2 get stuck more often than in CM2 and that the AI needs much more accurate movement paths defined in the editor by the user in order not to get stuck, so pathfnding algorythms in CM are probably more complex and resource demanding.

[Guest]

Yes, I figured ArmA 2 would be used in response. I've actually played the game pretty extensively and I've made some informed guesses about what they're doing on the back end, as well as (more importantly) seeing how they handle particular problems.

In short they're not solving the problems CM would have to (in some cases they're not solving the problems that CM already *does* solve), and their back end is pretty complex but still not nearly as complex as CM's. They have some significant advantages in terms of the game's structure and assumptions they can comfortably make that we can't. It proves that they're good at finding tricks and approximations (of course they have the team and budget to do so). They're not "solving" the problems that I'm talking about, though.

So... nope! You're talking about apples and oranges.

Guys, I'm not some hermit who only plays CM. I play the games that are out there. And I've been writing pathfinding code for a long, long time, and know how pathfinding systems work. If you are a programmer that's solved the problems I'm talking about - pathfinding with moveable colliding objects on an essentially unmanaged scale and scope- please stand up so the community can applaud you (and then you can tell me how you do it!). But when I say "no games of CM's scale do this" I'm not just blowing sunshine.

[JRMC1879]

ArmA 2 has maps larger than all maps available in CM and allows battles with several thousand AI that wont clip into one another. 'Tac AI' in ArmA appears to be noteably less sophisticated if it even exists and user generated missions need more accurate AI plans than in CM though, but i think its still prooving that pathfinding on huge maps with thousands of AI can be done without clipping and with hardware available to the average customer today.

On the other hand it must also be mentioned that at least in my experience AI vehicles in ArmA2 get stuck more often than in CM2 and that the AI needs much more accurate movement paths defined in the editor by the user in order not to get stuck, so pathfnding algorythms in CM are probably more complex and resource demanding.

From my experience - the tac AI both friendly and enemy - appears to be absolutely non existent in Arma - the primary reason I stopped playing it. I played a scenario with the british forces module and when ambushed in a Scripted mission - the 8 AI guys with me stood around and did nothing. Not very little. Absolutely nothing. When I compare that to what the Tac AI does in CM - the two are incomparable.

[Guest]

Thanks Phil for helping to explain the coding issues behind this thread. To people not versed in the arts of coding it all seems so simple.

No problem! If there's one thing I've learned, it's that if I think a common problem has a simple solution, then I probably just don't understand the problem. Which is pretty easily remedied as long as you're open-minded. In that vein (and because I like talking to you guys!) I'm happy to pass along what I know. And I don't mind debating the ways that some games might be doing it, either - I find this subject pretty interesting.

If it was then it would be in there already.

This is perhaps the most succinct and accurate statement about this that I've seen yet (I wish I'd thought of saying it!). For something this fundamental, if there was a real, practical solution, it would be in there. It would be *much* easier on us than trying to invent workarounds and approximations.

Anyway good to get a response from someone who knows what they are talking about. All too often (and I put myself in this category) these discussions are based on very little knowledge of how difficult it is to make things happen in the game like the real world.

I'm glad I can help, however little I might!

We keep forgetting that real life is incredibly complex and while computers and programming have come along a fair way since the ZX Spectrum we are still a far distance from getting it "perfect" with the pixels.

Perhaps if we were able to get it "Perfect" we would be in the Matrix and not playing CMBN?

Heh, perhaps.

[TheBlackHand]

Threads like this are useful and entertaining. They're useful because they provide the game designers with insight as to what may or may not need fixing as well as informing the players of the processes involved in a fix.

They are entertaining because you get to see the fanboys and the frustrated go at each other tooth and nail.

[jep]

It is indeed suprising that they couldnt figure out a better solution for vehicles passing through blocked gaps. But for me, it is not a game breaker because situations as you described it, where the clipping has an actual tactical impact on the game are quite rare, even if it is not nice to look at.

This is problem because it denies otherwise valid tactics like using at-mines to block narrow corridors.

[Guest]

It's definitely a problem, I don't think anyone is denying that. What we're talking about is the realities of a total solution to the problem (for CM, it's simply not possible), about what workarounds might be possible in lieu of that, and how those workarounds fit into BFC's development priorities. I've asked Steve to step in and comment on the latter two issues.

agusto, just in case you miss it, I replied to your ArmA 2 post a few posts up.

[Georgie]

With the upcoming Market Garden module the "ghosting" will be more annoying with some scenarios. The only thing that I can think of that might work is to impose a time delay on the first vehicle that "ghosts" thru a destroyed vehicle. Maybe three or four turns or so. This would kinda simulate moving the destroyed vehicle. It would normaly take much more time to move a destroyed tank but at least there would be a penalty.