Improbable spotting

[bodkin]

I never noticed this in CMSF but unless the code has changed I see it quite often now with the more densely vegetated landscapes of CMBN. I'm talking about the ability to spot and identify a target through thick forests because the game code can draw a direct line to the target even though the actual visible area is probably only a few pixels.

It's like looking through 100 slices of swiss cheese and being able to possitively identify something at the other end through the holes.

In the below example my tank is engaging an ATG through thick forest the actual area of the gun that would be visible must be about the size of a football. This is at a range of 250m.

I'm not saying it's a game breaker as it doesn't advantage one side or the other but it is pretty unrealistic. It is something that is common in other games like the Arma series where it is more of a problem if playing against the AI because you have to spot and shoot yourself.

Maybe it's meant to be like this as it simulates magnification through the optics or something but it seems unlikely and could be something to look at for future editions.

[Chris Ferrous]

That really doesn't look realistic given the apparent density of the trees near the ATG.

However, you've left a lot of potential questions unanswered:-

Did this occur after patch 1.01 was applied?

I assume the ATG was unlimbered and stationary?

What terrain type was the ATG in?

Had the ATG already fired on the tank?

Just how illogical is the 'straight line view' through the forest beyond what we can see?

Was the tank moving and buttoned when it spotted and fired on the ATG?

Could any other of your units already see the ATG?

And possibly an important one, what is the predominant ground cover in the intervening wood?

It would be nice to know there were some mitigating factors.

[Georgie]

From the screenshot that you posted we, the observers, are looking through the canopy of the trees whereas the tank commander is looking through the trunks of the trees and there appears to be no undergrowth. Take a look at the loading page in the game that pictures an Elephant parked in what looks like a managed pine forest and you can see that there is quite good visibility for several hundred meters.

[JSB]

Is there some wind which could make the foliage move? (I don't know if the game takes it into account though).

To have a better idea of what the troops can see, I often zoom from their supposed eyes position, and it's generally a good help.

[hoolaman]

Yes tree canopies do not block LOS like they appear to visually and should do in real life. I'm not sure how it works exactly but it doesn't give anything like realistic results. You don't notice on a flat map but situations like this are really bad.

[LJFHutch]

I'd say that's a very rare situation. I just set up two battalions of armour facing each other and after 2 minutes one of my tigers had spotted an M10 through the woods and knocked it out but that was the only action, some (?) icons (about 5 around the knocked out M10) but that was it. Forces were facing each other stretched across the roughly 500x500m map with a forest running down the center, 4 thick of the thickest forest, a mix of mainly the low tree but also a line of the high.

The tiger that got the kill actually did have a line of sight through the trees. From what I can tell at the moment the vegetation actually blocks too much. Many of my tigers should have been able to see under the foliage as there was no undergrowth - visually they did have LOs to their targets. Then again, this was from about halfway down the turret so that may explain it, did tigers have optics for the driver? Would they have been able to get a good view from 450m?

[siffo998]

i expirienced the same during a quick battle yesterday but i`ve got no savegame or screenshot... a stug IIIg spotted and killed a sherman ca. 800m away. i followed the path of view from the stug to the sherman and the sherman clearly was behind a treeline so that the stug should not be able to spot it. (the sherman was not at ground level so that the stug could spot him under the trees but was at a slight slope behinde the tree canopies.)

i`ve also experienced it in 1.00 times but it seems thats its still persistant now with 1.01

[Blackcat]

Tree foilage does block line of sight but very occasionaly there are gaps which due to the limitations of graphics are not obvious to the player. This has I am sure been the subject of a previous thread.

The situation above would appear to be an example of this. If you look closely the first trees encountered are low and the Panther would be able to see over the top of them. Behind those there is a low spot between two tall trees and in front of that spot it seems like there is a junction between two smaller trees. To the human player it would seem that there is no gap. However the internal modelling is more refined than the graphics display and so there may well be a clear LOS.

Such situations are aggravating, but in my experience, rare enough not to be a major problem.

[siffo998]

here is a videolink thats reffering to the same problem...clearly there is no chance that the pak would see the tank in real life...

http://www.youtube.com/watch?v=Jvc_DW5iG0M

@blackcat:

the problem is if there is a discrepancy between intern calculations and engine modelling then it is very difficult to the player to decide where his tanks or troops are safe... for example i place my tank behind some trees and think hey nobody can see him here and in the next moment my tank goes bang because the internal calculations say something different... i think this should be fixed by bfc in the upcoming patches because would be really unfair if its really this way.

the only other logical explanation why this would happen is because the crew/infantry saw the tank/pak/soldiers before it/they moved to cover and estimated where it would be. so they fire a round blindly through the foliage... dont know if those behaviour is modelled in cmbn

[Blackcat]

"the problem is if there is a discrepancy between intern calculations and engine modelling then it is very difficult to the player to decide where his tanks or troops are safe... for example i place my tank behind some trees and think hey nobody can see him here and in the next moment my tank goes bang because the internal calculations say something different... i think this should be fixed by bfc in the upcoming patches because its really unfair."

Ok, I can understand your view. Clearly it would be ideal if the internal "map" was matched exactly by the graphic display. I don't know how feasible that is but I suspect its is impossible due to hugely diverse numbers of graphics cards and settings that are available.

The internal calculations are though available to the player in the from of the target tool so it is possible to check if the position we think is safe from view is actually safe.

Finally, how often does this situation actually occur and of those occasions how often does it make a difference? I suspect it is very rare.

[siffo998]

"

The internal calculations are though available to the player in the from of the target tool so it is possible to check if the position we think is safe from view is actually safe.

Finally, how often does this situation actually occur and of those occasions how often does it make a difference? I suspect it is very rare.

hm might be true that it would be very hard to fix...

in case of how often: i experienced it around five+ times during play...

and in case of how severe: if you loose a tank (especially during a h2h) this way than it would make a difference i think... because owning a operateable tank or not makes a huge difference especially when its the only tank on the battlefield and you placed it somewhere were it should be invisible to the enemy

but i guess you are right and its to complicate to change it so we must assume that the crew estimated where the enemy is and fired blindy in the foliage or something like that... **** happens in war and especially in peacetimes.

[Chris Ferrous]

I recall similar discussions about CM1 and at the time felt frustration equivalent to that some posters have mentioned here.

There were two things which frustrated me with CM1 at first:-

Slot fire between buildings seemed very gamey (and often it was) but the fact remains it DID happen in RL.

The scattered tree tile seemed (to me then) a bit variable as to when or how many were needed to block LOS. I suppose in many ways that's what we're discussing here, the CM2 version of a scattered tree tile, it's just graphically far superior and much more variably represented (and designer controllable) than with CM1.

Again, in RL, tanks DID get hit being targetted through treelines, broken walls/buildings* etc. so CMBN may well be doing a very good job of this.

Personally, although it's only a game, I wouldn't want a spot to be determined as 'safe' with something like a green spot on it!

* I have read of successful 'indirect' tank versus tank fire when the gun-layer would actually use the curvature of the trajectory to fire behind an obstruction, e.g. wall, embankment, or low ridge, without the target ever seeing his vehicle. This could be achieved by firing at pennants, aerials, exhaust plumes, waving foliage, or even the noise or by some brave spotter on foot waving to say fire straight over my head! We don't seem to have any of these featuring in CMBN so I'm going to sulk.

[siffo998]

yes chris of course it happened: actually the most famous case i can recall might be the pershing gunner at cologne who reported that he shot inside the house (the panther was hiding behind it) because he guessed that he could kill the tank even through the house (or at least get the panther out of his hiding place). so its good i guess that those "impossible" hits are modelled nonetheless its very frustrating to loose a tank (or anything else) this way...

maybe we should learn out of this cases that you should be really carefull with your tanks and not feel to safe anywhere...

[Chris Ferrous]

True, Siffo, it's not whether these 'strange' things can happen at all, it's just how often that matters!

[Battlefront.com]

Yup, frequency can take something realistic and turn it into something unrealistic by either happening too many times or too few times.

We're going to look into what specifically might need a little tweaking. We did a big round of tweaking right before we released v1.00 so we might have missed something.

Unfortunately for us game developers, "outliers" can only be identified after the average behavior has been correctly established. For a game of this complexity it takes years to do that. It also usually takes a lot more than a couple dozen testers to find some of them simply because in one day more games are played by our customers (post release) than our testers could possibly do in several months worth of play. Outliers are all about the numbers.

Steve

[LJFHutch]

Well here's my test, the second time I spotted another M10 and a M20 truck, but as you can see, I had LOS there, I was surprised there wasn't more shooting going on considering there was a battalion of TDs on the other side of the trees.

[Battlefront.com]

Yeah, the problems manifest themselves with complicated combos of different foliage types and some degree of elevation changes. It's pretty darned specific and we've not quite figured out why it sometimes happens.

Steve

[bodkin]

From my understanding there isn't anything wrong or a bug with what's happening, if the code can draw a direct line of sight through tree trunks, foliage, buildings etc, it will see it as direct spotting. The problem is in real life the actual visible area of the target would be a tiny patch of paint visible through a small window in the environment, this is unlikely to happen in real life.

[Battlefront.com]

Yes, that's it in a nutshell. It's an example of 1:1 sometimes being too literal. The trouble is if we abstract the 1:1 associations too much then we get into a different set of problems. Therefore, we need to proceed with caution and be as narrow with our changes as possible. And to do that we need to have as narrow an idea of what to tweak as possible before we start messing with stuff.

Steve

[Chris Ferrous]

Is it true that trees have two blocks on LOS; a cylindrical trunk and a spherical crown?

If so, would it be possible, and might the solution be, to increase the size of the trunk cylinder slightly (visually only, i.e. not for movement) so that the chances of drawing a line through a patch of trees would decrease?

I was thinking that such a solution just might replicate what really happens, i.e the difficulty of spotting a patch of painted metal in a sea of green.

[Thomm]

Is it true that trees have two blocks on LOS; a cylindrical trunk and a spherical crown?

No. The tree is treated as a homogeneous cylinder from top to bottom.

This is why you can see through the foliage.

This is how I still understand it. If this is wrong, Steve will hopefully correct me.

Best regards,

Thomm

[stoex]

You have mentioned this twice now, Thomm...I hope you are not right about this, because if you are, it is simply horrible. It would skewer spotting so bad where there are trees involved that it hardly bears thinking about. And it would be the burial of realism in a lot of circumstances. A line of trees two or three deep might as well be considered a brick wall for spotting purposes, which obviously is not the case IRL, underbrush or not. Either that or trees might as well be trunks only and we could seriously save on FPS by just not showing the foliage at all by default, as it woudn't really exist for game purposes. Either way - don't even get me started on how wacky i would consider the decision making process that led to such a design.

I beg for Steve to show up and say you are wrong. If you are right, I will lean out of the window here and call it a fatal flaw in the game design. Not kidding.

[Battlefront.com]

Stoex,

I don't undertand your fear. You have the game, you see others playing. In the end it doesn't matter one bit how the code works, rather it only matters what the results are. And the results are nowhere near as bad as you are worrying about, so there obviously isn't a big problem here. If there was, we'd all know about it through direct experience with the game that's in our hands.

How are trees simulated in code? I don't know/remember. I don't think Thomm knows for sure either. But it's not really important as long as the end results are as they should be

That being said, whatever computation savings techniques Charles is using to handle trees (a framerate killer even in CMBO, remember) does have some shortcomings. Hopefully they can be worked around if we can isolate specific conditions that aren't performing as expected.

Steve

[stoex]

Steve,

I completely agree with you...maybe I should have phrased my earlier post differently. I really can't imagine that what Thomm indicated is really the case. As you say, we all would have noticed that long ago. I guess I was really making fun of Thomm in some obscure way because I thought his idea was patently silly or impossible. Apologies, Thomm!

I am also aware that beyond the trunks, trees cannot be 1:1 for reasons of computing power, and that in itself is not a problem.

That said, I also tend to agree that LJFHutch's picture above indicates that something isn't working quite the way it should realistically be. That number of trees shouldn't be blocking LOS or LOF as dramatically and uniformly as they appear to be doing. Unless undergrowth is a LOT more abstracted than it appears to be.

One slight issue I have with CMx2 trees in general is that they always have the foliage beginning at the same (relatively low) height regardless of other circumstances. What I mean is, in a dense Central European forest the space from the ground to about 5 meters is generally pretty open (at least in summer) due to the fact that very little light falls through the foliage of the tall trees. That means there is very little undergrowth except on the forest's edge and on possible clearings, and there aren't a lot of low branches with leaves on them to block view either. This counts especially for deciduous trees. There isn't really a good "dense forest" deciduous tree in CMx2 as far as I'm concerned, but I live with it just fine .