Is this right?

[dan/california]

I think what he is trying to day in several thousand words is could we add one more los/lof check based on head position? Preferably without melting our computers or boiling Charles Jar.

[Pandur]

i think steve tried to say quiet some times now that the game has no friggin idea where the head is! it is spotting from predefined position, prone, kneeling and standing for infantry. to spot from something like the "head" or the "eyes" you need to spot form polygon to polygon, wich does not work.

better up the spotting calculations by shorter spotting intervals, this would be more important for me. this would keep vehicles from spotting each other at "exactly"! the same time after not spotting each other for 1 minute or more.

id rather see this adressed, when we talk about spotting.

[hoolaman]

Don't forget spotting is by "action spot" which is 8x8m or smaller around walls and buildings, or one per building. You can see where the action spots are by the highlights when you move a squad.

If the middle of an action spot is behind the crest it will have limited LOS over the crest for prone men. Imagine a stick with six notches on it stuck in the middle of every square. If you can trace a line from notch to notch you have LOS.

Forget about heads and torsos. All the game sees is: is everyone in the team prone? Is anyone in the team kneeling? Is anyone in the team standing?

I think there are some improvements that could be made, like being more fuzzy with spotting and less black and white, but it works pretty well I think. Partially identified units feature would add another layer of coolness to the mix too.

[Battlefront.com]

c3k,

Okay, taking what you said, above; "The height is based on roughly where the head would be for such a stance." That's the key question! When I can SEE the head of the guy sticking up from the ground (be it a slope, a crest, or a ditch), THAT guy, the head-sticking-up-guy, cannot SEE out.

Not necessarily. Probably better than 9 times out of 10 if you can see the head that Soldier's LOS is not blocked. That's because the ELOS height is roughly the height of his head. Ergo, no problemo That 1 in 10 times is if it's just a tiny, tiny bit too low compared to whatever is directly in front. In that case you might see the top of the guy's head but the ELOS height is just below the blockage. Considering there are only 3 standard graphical positions for the Soldiers this isn't a problem.

The blocked LOF (which is all a player can see) seems to start from his belly - which is behind dirt and out of sight. So, it's not an issue of "individual polygons", but, rather, the essential part of the human body which sights things - the eyeballs located in the neck-mounted head. Why does the blocked LOS not take the head location into account.

The head of a standing or kneeling soldier is directly over its center mass. So when offsetting the height to be at the head's height of that stance, the actual point being used is very close to the actual graphical head that you see. This is not true with prone because the center mass is about where the small of the soldier's back is. Or put another way, roughly half way between the head and feet.

As I've explained this is necessary for prone soldiers because, overall, the results of having the ELOS point for prone moved forward to the head means that 95% of the body is not spottable by the enemy. If you think you're distraught about the system as it is now, I think you would probably hang yourself if you saw the problems that would cause

There are complex and not-trivial CPU hits which would allow prone soldiers to have one spot for outgoing spotting and one for incoming. This is a molehill problem so we don't believe it's worth addressing for quite a while. But yes, eventually we will put in some sort of improvements. Probably call it Advanced Enhanced LOS

Again, this is a specific issue to prone soldiers with dirt between their bodies and the enemy, but their heads exposed. However, EVERY soldier likes this kind of position over the more exposed ones! Hence, this specific condition does occur more frequently than one would suppose.

No, it doesn't. The soldier laying prone has some special code in it to be a little less sensitive to blockage right in front of it. This usually compensates for the center mass issue and therefore prevents very small bumps and what not from blocking LOS. We made some further tweaks to this in v1.20 so you're definitely not seeing what the testers and I am seeing. Yet!

The above statement draws two responses from me. First, I'm aware of the non-recipricol LOF condition. Specifically, if unit A positions a member in a buildings action spot near a window, it can fire on unit B which is in the open. However, unit B traces it's LOS to the action spot center, which is blocked. Since unit A has a member AWAY from the blocked center of the action spot, A can fire on B, but B cannot fire on A.

Incorrect. There is never a situation where LOS is one way. It's always, under all circumstances, two ways. What can happen, however, is that some members of a Team may have LOS from their Action Spot to the Action Spot of the enemy, while others may not. That's not non-reciprocal LOS/LOF at all because the soldiers who can't see also can't be seen. Period.

The second response is more of a question than a statement. Since the prone soldier's LOF isn't even checked (or applied) if the LOS is blocked, shouldn't the LOS be checked from the same spot as the LOF? I'm not saying the LOS should FOLLOW the LOF (indirect fire, ballistics, etc., account for differences). I'm saying the LOS should be checked from the head/shoulder.

Technically speaking the LOS should be from the eyeballs and LOF should be from the muzzle of the weapon. Those positions are often at least .5m different. When prone this difference is more like 1m. Weapon and exact positioning of the body dependent, of course.

Blockages are less likely when kneeling, even less likely when standing compared to prone. Therefore when prone the tolerance level for blockages is pretty tight. To compensate for the lack of exact tracing of LOS/LOF from the correct polygons, the system gives the prone guys a little bit of a break. If LOS can be established from center mass the system will not allow minor blockages to interfere with LOF. The assumption being that it's too close to call and therefore err on the side of LOF being possible.

Remember, we're talking about extremely minute and extremely case sensitive situations here where what you see won't jibe with the rest of the graphical environment.

When computers get more powerful we'll likely get rid of this abstraction since it is indeed a theoretical source of problem, even if in reality it happens very rarely in a way that matters.

Let's take your 90/10% splits. This is where a misunderstanding may be occurring. If soldier A has a rifle and is prone behind a crest line/slope/ditch/etc., and his head is up, rifle is up, and he is looking to his front, I think we can agree that he only has 10% exposure, yet he also has 100% firing and spotting ability. His offensive capability does not care whether or not his left foot has LOS to the enemy. Nor his belt buckle. I DON'T CARE what enemy soldier B has exposed. If B has LOS towards A, then A's lack of exposure should be a factor keeping him from being spotted. (It's harder to spot a head sticking out from a ridgeline compared to an entire torso/body.)

Spotting has nothing to do with the polygons you see. That's what I keep saying. Having spotting to polygons is impossible for any current PC to handle, so you need to stop confusing the graphical representation with spotting. They aren't directly attached.

What instead happens is the location of the soldier establishes its horizontal plane's location. The system positions the graphical representation on that point, which becomes its center mass. The stance of the soldier determines what vertical height is used for ELOS and separately the system draws the graphical representation of that stance. That is the point in space where LOS is drawn to and drawn from. Therefore, if you see 10% of the soldier but that 10% isn't the center point for that soldier's stance... then no LOS is possible and therefore no LOF in EITHER direction.

That's the FROM portion of LOS. Earlier you stated that the INBOUND portion breaks the body into distinct portions. That's great. Really. This is all about the guy whose head is sticking up, and what he can see.

Clarification... inbound fire is constrained by the terrain that it passes through. Therefore, if a soldier's 10% that is exposed is also its' center mass then that's all the incoming fire can hit. It can't hit the 90% of the soldier which is not exposed. In other words, Spotting is a binary thing... either you spot or you don't. What you can hit, however, is situationally dependent based on many factors, primarily the level of exposure you see in the graphics.

Again, you state "the problem with this is if your soldier spots 90% of an exposed soldier, but the 10% of that exposed soldier is the head, then you won't spot that soldier". Why not? Why does MY LOS TO the enemy have to touch the enemy's head???? It doesn't work that way now, nor is anyone asking to change that.

If you want the head of a prone soldier, lying behind cover, to have a 100% opportunity (not chance!) to be spotted, then LOS must be drawn to/from the head. Which I explained above is a very bad idea.

Yeah, we're focusing on a small portion where the current system doesn't work. BUT, it's an important portion.

Yup, it is important. But much of this discussion has talked about things being "broken" which absolutely aren't.

A lot of guys under fire try to hide their bodies and only expose their head and weapon. Right now, if there's a SLIGHT elevation difference intervening between belt-buckle and enemy, the guy with his head up cannot see or shoot.

As I've said, v1.20 reduces the chances of this happening.

Center of mass as a start point for LOS is flawed; this is masked in many situations, but gets highlighted for prone soldiers with their bodies hidden in an elevation difference from the enemy.

It's not "flawed" per se, rather it is an abstraction. Abstractions are, by their very nature, imperfect. But as I've said a billion times since CM:SF was released... the amount of imperfection in CMx2 is miniscule compared to the amount in CMx1. Was CMx1 flawed? No, certainly not. It was just far more abstracted which means it was, by its very nature, more imperfect.

As technology gets better we can increase the fidelity. In a perfect world this would mean drawing LOS from the polygons which represent the eyes to any and all polygons in the entire game based on LOS. This will not be possible in my lifetime, so I'd advise people to not hold their breaths waiting for this Instead what we can do is add more ELOS points to the system as we feel the hardware can handle them. We already have 6 more than CMx1 and CM:SF originally had, so more ELOS points are definitely going to happen. The question is when does the resources it gobble up give a better bang than using those resources for something else? Right now, and for the near future, we feel the ELOS system is excellent the way it is. Therefore, we're likely to use new computing resources for other purposes before we come back and revisit ELOS in its current form.

Steve

[Beardo1976]

Steve,

Great explanation but that post made my head bleed. I'm sending you my hospital bill.

[Battlefront.com]

I don't know why people keep asking about this stuff... it's about as exciting as watching paint dry

Steve

[Chainsaw]

Well Some people actually get excited over seeing paint dry.... :cool:

[-Pv-]

I suspect the fidelity of the operational portion of the game is much higher than the visual depiction leading to misinterpretation of the details, further distorted by the player's frustrations over winning too easily or losing too easily for reasons which may not be immediately apparent.

-Pv-

[c3k]

Steve,

Just back in town. Thank you for taking the time for answering in such detail. Sounds good. (But, you KNOW I'm going to be throwing men prone into crest positions as soon as I get 1.20!!)

Thanks,

Ken

[Battlefront.com]

Indeed, customer perception of what is going on, or not going on, is key. There are tons of threads in CMx1 Forums where one person would report "this is wrong" or "this is a bug" and someone else would say "well, that's explainable because x,y,z". Sometimes that was correct, but sometimes it was in fact a flawed design or a bug. So again, CMx1's lack of fidelity allowed people to invent their own reality even when that reality was completely wrong

Steve