1:1 abstraction workaround - split syrian squads too

[KNac]

Watching my own screenshoots (the gridline i drow was wrong BTW) and reading what I had written soemthing came to mind so I made a quick test...

So... editor squares are 8x8 but do not fit with action spots? or are fire does nto snap to the center of an action spot as we though? mmm rare

[thewood]

This is why Dorosh thought designers notes were critical. All the information is scattered around the forums and is just confusing the heck out of everyone. A detailed understanding of the grid and action spot system is needed to let people (especially me) understand what is supposed to be happenning.

[Cpl Steiner]

Having done a few tests of my own I am becoming seriously concerned now that LOS is not a two-way thing. Logic dictates that if A can see B then B can see A, but when I did the test illustrated below, my squad never spotted the enemy even though they were literally feet away and blazing away at them.

First image: Shows the layout in the map editor. The blue dot is the US squad and the red dot is a group of Uncons. The purple arrow shows the direction the squad took using the "Hunt" command.

And now the aftermath. All US soldiers are casualties and not one even returned fire. They just ended up doing a "crawl of death" trying to enter the building next to them for cover.

The view from the jubilant Uncons a few feet further up the street!

I hasten to add that I believe the US squad didn't fire back because it had no LOS to the Uncons. If it was due to some other factor such as panic I would still expect the Uncons to become spotted but they never did, even when I could see mussel flashes and spent cartridges flying from invisible weapons.

This is worrying, to say the least.

[dan/california]

if i am interpreting the lat few post correctly Cpl Steiner and Knac have been kind enough to prove me at least basically correct, or maybe i was already reiterating something that they had already figured out. but regardless....

The single biggest problem is that the game is computing incoming fire to squad members that do not have outgoing LOS because the rest of there squad, and there fore their action spot for LOS purposes, is on a trailing action spot relative to their direction of movement. That is a BIG problem. I think it is a fixable one, and it should not break the frame rate too badly. I would personally rather play a smaller scenario with this fixed than a larger one with it still broken.

Can Battlefront give a yes,no, or maybe to the conclusions in this thread.

[hoolaman]

Originally posted by Battlefront.com:

P.S. CMx1 behaved with even less fidellity, you just didn't know it. So let's not get started on yet another off base thread about how 1:1 in CMx2 is less realistic than abstracted stuff in CMx1. It isn't.

Originally posted by Battlefront.com:

OK, so what about the behavior problems with CM infantry teams "sticking together"? Keep in mind that CMx1 was even worse than CMx2, so please don't even start with me about how it was better before because it most definitely was not in terms of this discussion (i.e about the whole unit being exposed or not to do something). Here's an example:

But nobody here was talking about CM1 or comparing it to that game but you Steve.

They were talking about a specific issue where you currently have better playability with the US side but not the Syrian side via splitting squads.

[Panzer76]

Indeed Steve, its sarcastic, because this title is supposed to be the most realistic yet, but somehow we are supposed to think that the Syrians are incapable of shoting a RPG without exposing every single man in the sqd. This is ofcos absurd and as such, any suspense of disbelief goes out the window.

Thing with cmx1 was that it might have been very unrealistic, but since it was abstracted to such a high degree, we imagined what were happening, instead of watching our 1:1 soldier behaving like idiots.

[Battlefront.com]

KNac,

I'm very interested to the question Cpl Steined asked, is soemthing I though the other day, maybe if a building corner it's spread between 3 action spots, checking corners etc w/o overexposure would be easier. Some testing requiered, but each square in the editor is 8x8m so is an action spot.

It looks like you answered this yourself already, but I'll confirm that buildings are their own Action Spots overlayed on top of the underlying grid of Action Spots. There are other tailored spots like this in the game, such as shell holes (large ones at least), trenches, and probably some other stuff I'm not thinking about.

As for the basic problem that sometimes a couple of soldiers "spill out" into a dangerous location and two problems arise from it:

1. They do not effectively return fire. They are supposed to, but various they are at a serious disadvantage in the same cases you guys are noticing, so it generally doesn't amount to anything.

2. They do not reposition themselves well enough when an adverse situation crops up.

Building corners, that are not in line with the grid, appear to bring this out more than other situations for the reasons you illustrated. Currently it is a limitation, but I think it can possibly be worked around. We have a number of things like this lined up for v1.05 so my hope is that you'll see some changes very soon. At least Charles will be spending time on issues like this very specifically.

So... editor squares are 8x8 but do not fit with action spots? or are fire does nto snap to the center of an action spot as we though? mmm rare

It's hard to tell for sure, but I am fairly certain that your grid lines are off. Those two soldiers are a single HQ unit, correct? Then they should be contained within the same Action Spot, yet your green lines have them in two different ones. So it looks like something is wrong with your lines.

Hoolaman,

But nobody here was talking about CM1 or comparing it to that game but you Steve.

Right... because if someone is going to complain it has to have some sort of proper context, otherwise the complaint itself is meaningless. Arguing that CMx2 "isn't realistic" is a pathetically stupid argument to make if you enjoy wargames (any wargame) since it is obvious that ALL wargames ever made, and those not yet made, fail this test. Nobody, including myself, is idiotic enough to say that CMx2 is "realistic" when "realistic" is implied to be a complete reflection of everything and anything in the same combat setting. That's why I'm saying is that it is "more realistic than what came before it" (CMx1), which in turn was "more realistic than what came before it". Since that is the context, trying to avoid it (as you would like done) makes the entire discussion pointless because without context there can be no point.

Obviously the reason why people, like yourself, object to me brining CMx1 into this discussion is that you want to think of it as a Sacred Cow that is beyond criticism. Because afterall, if you are arguing that CMx2 is not good enough because it isn't realistic enough, and I can show how CMx1 is less realistic than CMx2, then by carry through of logic you must also find that CMx1 wasn't good enough either. Since you don't want to face the reality of your own line of reasoning, it bothers you that I keep bringing CMx1 into discussion when someone is trying to argue that CMx2's realism is crap.

They were talking about a specific issue where you currently have better playability with the US side but not the Syrian side via splitting squads.

As intended. Syrian Squads are inherently "less playable" in real life than US Squads, so that is reflected in the game. I outlined the logic about this already, so I guess I should ask if you bothered to read it? If you did, clearly you didn't understand it.

Panzer76,

Indeed Steve, its sarcastic, because this title is supposed to be the most realistic yet,

If you have a glass that is half full and another that is three quarters full, and someone points to the one that is three quarters full and says "that is more full than the other glass" it would appear that you'd reject that statement because the glass isn't full. That is an illogical position and a pointless one knows ahead of time the glass will never ever be completely full.

I don't understand why it is SO VERY HARD for you to understand that CM:SF is not perfect, but compared to CMx1 it is more realistic. You can argue that you don't like CM:SF as much as CMx1, or at all, and I'll not offer any resistance to that opinion. But when you make illogical comments about CM:SF's realism compared to CMx1, or anything for that matter, yeah... I'll be right here to kick your bad logic to the curb.

but somehow we are supposed to think that the Syrians are incapable of shoting a RPG without exposing every single man in the sqd. This is ofcos absurd and as such, any suspense of disbelief goes out the window.

I won't argue with this at all. For you the suspension of disbelief is more important than the realism of the game itself. Fine, that's a rational argument to make and I accept it. But you can't have your cake and eat it too... if you want to argue that CMx2 is less realistic than CMx1 you had better rise to the occasion and honestly debate that instead of coming up with childish quips.

Thing with cmx1 was that it might have been very unrealistic, but since it was abstracted to such a high degree, we imagined what were happening, instead of watching our 1:1 soldier behaving like idiots.

Again, I have no argument against this. You find the balance of abstraction:realism in CMx2 unappealing compared to the abstraction:realism of CMx1. Fine, that works for me. Just keep to that argument and don't continue to get into discussions of realism in a vacuum because you'll dilute your own argument in doing so.

Steve

[ September 29, 2007, 09:11 PM: Message edited by: Battlefront.com ]

[C'Rogers]

Thing with cmx1 was that it might have been very unrealistic, but since it was abstracted to such a high degree, we imagined what were happening, instead of watching our 1:1 soldier behaving like idiots.

The argument seems to be you didn't actually like CMx1, you were just tricked into liking it. The things you didn't like just weren't noticeable. Now there are less of those things that you shouldn't like (abstractions) but they are more evident because of the increases in realism.

[dan/california]

Given how many of these problems are driven by building corners, would the best short too medium term fix be to create a specific "corner" animation and maybe even a command to go with it. That could abstract if not simulate having the sense to poke your head or even a mirror around the corner before jumping into the street?

[Battlefront.com]

C'Rogers,

That's the sort of logic flaws I've been trying to point out here! Glad I'm not the only one seeing it. If the arguments of Panzer76 and Hoolaman were simply "I don't like CMx2 as much as CMx1" there would be no debate. But for some reason this sort of thing has to be justified so that it becomes a point of fact that CMx2 is crap rather than an opinion. That's where we get into problems

Dan/california,

Given how many of these problems are driven by building corners, would the best short too medium term fix be to create a specific "corner" animation and maybe even a command to go with it. That could abstract if not simulate having the sense to poke your head or even a mirror around the corner before jumping into the street?

Certainly no new Commands for this. The TacAI should be able to do something with this using a player's assigned Waypoint as guidence. In my mind (which doesn't matter much since it isn't Charles' mind ) the fix would be that a Waypoint near a corner would be treated as staying on the near side and not going beyond it until regrouped. Then move to the next Waypoint (if there is one) or stay put if there isn't. No Waypoint at the corner means simply walk right on by it. At least that is what I am going to propose to Charles! It hinges mostly on his ability to sense what a corner is. That could be problematic, I don't know.

Steve

[Guest]

It hinges mostly on his ability to sense what a corner is. That could be problematic, I don't know.

Ever thought of branching out and managing development in less... interesting industries? That's one of the best things I've ever heard a non-dev development manager say.

KNac, love the thread. Steiner... those screens have awakened a serious hankering to play CM:SF again. I'm really looking forward to 1.04. Happily I've cleared the Marines module with my wife. Yeeeeah.

On topic, now...

I would personally rather play a smaller scenario with this fixed than a larger one with it still broken.

Have you guys (BFC) given any thought to picking a lower upper limit on scope for future iterations of the engine? Would free up some power for Charles to use elsewhere.

It seems as though you were shooting for battalion+ when you developed the engine, dunno if anyone is actually playing at that level (especially with RT factored in... I don't know if I could handle an entire battalion).

Eh, I'm rambling, it was a long week. We just patched our very large DB/web system after a grueling three-month cycle, *and* I wrote a half-dozen new production support apps in addition to managing my guys and doing my deadlined R&D. Sometimes I wish I were in a nice socialist country. 35-hour weeks indeed. At least developers in California get overtime.

So... I may not be making sense. Grain of salt and all that.

[dan/california]

Certainly no new Commands for this. The TacAI should be able to do something with this using a player's assigned Waypoint as guidence. In my mind (which doesn't matter much since it isn't Charles' mind ) the fix would be that a Waypoint near a corner would be treated as staying on the near side and not going beyond it until regrouped. Then move to the next Waypoint (if there is one) or stay put if there isn't. No Waypoint at the corner means simply walk right on by it. At least that is what I am going to propose to Charles! It hinges mostly on his ability to sense what a corner is. That could be problematic, I don't know.

The real key would be for the unit stopped at the corner to get the ability to look around it, even if their action spot did not normally allow this. The ability to, at least mostly, look before you leap is the primary issue here. I am pretty sure that is SOP in the sandbox. In fact various robotic means of doing seem to be ever more popular, but that is another thread.

If the same look before you leap idea could be applied to the crest of a hill that would be even better. :cool:

[Battlefront.com]

Phllip,

I think I understood everything but your compliment about my post Did you like the simplistic nature of my summary of a very complex code issue, or just that I'm honest that I don't know if we can manage to do something that appears to be simple from a lay person's standpoint? Inquiring minds...

We have not thought about putting an upper limit on the game system. In fact, we were explicitely "requested" to not do that by gamers. Seems like the better way to do things so we'll keep on doing it that way. But as you say, a Battalion+ force in RealTime is rather difficult to manage, though I think some people have done it and enjoyed it! I don't understand how that is possible any more than I understood how people played 2+ battalions in CMx1 games

Congrats on the DB fix. One small victory in a war against buggy software won!

dan/california,

Yes, a "look before you leap" feature would be quite nice. I have NO idea if that is possible, though I do have some ideas to float by Charles. Could come to nothing, but even I don't know until I ask!

Steve

[dan/california]

Could I propose the rather simple idea of extending a units LOS check to the surrounding action points if and only if it is an infantry unit using the hunt command, and maybe something to represent the highest state of alertness in a fixed position, and not otherwise engaged, suppressed or exhausted.

This would seriously limit the number of units the extra LOS checks that had to be run for and also simulate the benefits of moving very deliberately in bandit country. If Charles can detect a corner reliably all the better but I think the idea would help either way. And it could be ridiculously difficult to explain the concept of "corner" in C++, jsut a guess on my part.

You put the code in 1.05 and follow up with the animations later.

[ September 29, 2007, 11:22 PM: Message edited by: dan/california ]

[Guest]

Originally posted by Battlefront.com:

I think I understood everything but your compliment about my post Did you like the simplistic nature of my summary of a very complex code issue, or just that I'm honest that I don't know if we can manage to do something that appears to be simple from a lay person's standpoint? Inquiring minds...

Well, one man's honesty is another man's wisdom. Knowing what you don't know is very respectable in my book, and quite rare among non-dev development managers in my experience.

Glad my post was readable.

Originally posted by Battlefront.com:

We have not thought about putting an upper limit on the game system. In fact, we were explicitely "requested" to not do that by gamers. Seems like the better way to do things so we'll keep on doing it that way. But as you say, a Battalion+ force in RealTime is rather difficult to manage, though I think some people have done it and enjoyed it! I don't understand how that is possible any more than I understood how people played 2+ battalions in CMx1 games

I guess I don't understand how people do that either.

Very seriously, though, if you haven't capped the engine I'm flabbergasted by what Charles has accomplished. I don't think I would attempt that by myself... nor would I want to. You've got the potential to have huge maps, with all units in view, doing very low-level things on a very rapid basis.

I shudder. Might be something to think about if Charles needs more CPU time / memory, though by all appearances he is a magical genius, so that may never be a problem.

Originally posted by Battlefront.com:

Congrats on the DB fix. One small victory in a war against buggy software won!

Thanks. We also made a whole bunch of fixes to the user-facing piece of the thing (a gigantic web-based system with billions of records). My part in it mostly involved keeping people (including myself) off of the pertinent developers' backs and writing apps / directing my folks to keep our production people from going insane. My own long-term projects (basically AI/NLP/clustering work) are generally pretty orthogonal to the main branch of our code.

Anyhow, thanks for the answer. Cheers!

[Guest]

Hey, time for some totally-for-my-own-amusement speculation!

If they've really got building-based "special" action points it may actually be really doable to represent / "explain to C++" (I'm gonna use that one on Monday if that's okay, dan) corners in the CMx2 code.

Beyond telling C++ about corners, though, the logic of corner-handling has always been an interesting puzzle to solve whenever I've tackled it. It's a combination of all of the other major problems of AI (spotting / complex coordination / pathing / positioning / behavior), all happening at once.

[Other Means]

I suggested this:

Originally posted by Other Means:

I wonder if the whole 1 - 1 thing could get a quick win by the squad action point being given by the soldier on point. That way no-one gets killed before the LOS check is done.

Of course, tail end Charley gets into more danger but I think that'd be less of an issue (except to Charley).

http://www.battlefront.com/discuss/ultimatebb.php?ubb=get_topic;f=52;t=003080;p=4

Here, and Steve said it might work, so that could address many of these issues.

[Other Means]

What I'd like is the Hunt command has the point man check round the corner and check towards the side of the map opposite where they entered first, then the opposite way, automatically at every corner.

He should do this with 25% exposure if possible. An animation would be nice.

Re-positioning AT teams towards AFV's should also be automatic, if possible.

[Cpl Steiner]

Just had a couple of thoughts regarding how to handle corners.

1. I've noticed that a MOUT squad of three teams actually occupies three action spots. The game uses the "team" as the basic unit, even when a squad is not split, so a three-team squad takes up more action spots than a two-team one, which takes up more action spots than a one-team one. This can make it quite hard to avoid having men spill past their intended building corner, as the action spot picked for the squad is actually only for one of the teams in the squad.

You can avoid this manually by splitting a squad into its respective teams and ordering each team to the same action spot. The effect is that the squad reforms but in a smaller area. Ideal for MOUT!

A better, less micromanagement way of doing this would be to have some sort of status toggle for infantry like "spread out/close up". If you issue a "close up" status order, all teams reorganize to the same action spot and move like that from then on until a "spread out" status order is given. That way we could safely move a whole squad to a building corner.

2. I have an old Avalon Hill board game called "Gunslinger". It is a hex-based tactical combat game set in the wild west. In that game, if you put a guy physically at a building corner, you are allowed to use a command to place his "head" in any adjacent hex (using a counter). Shooting is always traced from the guy's head rather than from his body, if a head counter is in play. Shooting at someone's head rather than his body also incurs a "to hit" penalty (as the hit system of the game selects a body part first, and anything other than "head" is ignored).

What this effectively gives you is a neat and simple system for looking and firing around corners or over low walls under the restrictions of a hex grid. It even takes into account the cover effect of only exposing one's head to the enemy.

To replicate this in CM:SF, here's what you do. First, implement the "special" corner action spots that Steve alluded to. Next, if a squad moves to a building wall action splot ("closed up" as discussed above) it stops just short of the special corner spot and deploys two men there, one standing and one kneeling. This is the equivalent of placing a "head" counter in that action spot. Now, LOS/LOF can only be traced to the men at the corner, who should also receive a fairly good cover bonus to direct fire. From what I've read, building corners are some of the best cover there is in RL, because of their thickness.

How does that sound?

[Cpl Steiner]

One more thing regarding the screenshots I showed before. Looking at the map editor layout again, it is clear to me that the squad in serious trouble should be the Uncon one. In RL, the US squad would peek around the corner and spot a group of bad guys literally standing in the open with no cover. The US squad, on the other hand, has good cover provided by the building corner. In RL, the Uncons would all be shot within seconds and the US squad would emerge from the side street unscathed. At the moment the exact opposite is happening. This is why, to my mind, it is so important to fix this for v1.05.

[Drusus]

Would it be possible to get an explanation to the following question?

Why there isn't a immediate recheck to the other direction when the uncons get LOS to the US squad? The thing I have difficulty to understand is that because of the positive check from uncons to US squad, we already know there must be a LOS to the other direction! Why isn't this information used?

Probably there is a good reason why this information can't be used. I am only trying to understand what reason that is, so we all get to know the LOS system a little better...

[Bigduke6]

Not really big guy, we just happened to be reading the same thread at the same time. Took me another 12 hours or so you were speaking of the Devil, so to speak.

Kinda skeery, in a Twilight Zone sort of way.

Originally posted by Abbott:

Hehe, you mention that guy and he shows up!

[Other Means]

Originally posted by Drusus:

Would it be possible to get an explanation to the following question?

Why there isn't a immediate recheck to the other direction when the uncons get LOS to the US squad? The thing I have difficulty to understand is that because of the positive check from uncons to US squad, we already know there must be a LOS to the other direction! Why isn't this information used?

Probably there is a good reason why this information can't be used. I am only trying to understand what reason that is, so we all get to know the LOS system a little better...

I think it's because by the time the *squad* is in LOS, the front *soldiers* are exposed, giving the uncons free shots.

That's why having the soldier on point be associated with the action spot may change things.

[KNac]

I think having the soldier on point associated with the action spot would solve some issues quite fast indeed. This combined with what dan/california and myself were suggesting of activating adjacents action spots (all the action spots a squad occupes) would make things even better.

Steiner suggestions are nice too, but somewhat I see them harder to implement so would take more time. Personally I prefer the second option (involves less user input); when you order a squad to move on the wall to the corner, they would move carefully and then soem sort of special action spot would be activated to trace LOS. Other option combining various of the ideas thrown around, when you order the squad to move to the corner, they stop in cover and the adjacent action spot is activated for check LOS, with the men being on cover (an animation of someone checking moving the head would be nice but we can do it with the abstraction). If there is a positive contact then you can see the enemy squad via interface.

Now there is an other 'minor' problem, once we have solved on how corners are checked, is how do you attack the enemy behind the corner. Once there is a positive contact, first it would be for both sides unless a special rule is included, that would be a disadvatage for the enemy. And later would be a matter of squad repositioning so they have positive LOF agaisnt the enemy but seeking major cover using the corner.

[Wiggum]

What about to give Syrian Squads only the possibility to split a AT-Team from the rest of the squad ?