spotting changes in 1.11

[c3k]

Hmmm,

It seems that the only way to test this would be for you to take a machinegun up to the top of the cupola on St. Paul's and perform some spotting and firing "tests". Please let us know what the results are.

Ken

[MarkEzra]

Get it! There aren`t any problems regarding with spotting, nor will there be any!

I had a hard long time to understand this after hours of indoctrination but now i really start seeing it. Not just stupid "feeling" or "witnessing" but really "seeing" the hard facts...

Good Lord, Taki...put down the kool aid! That stuff will kill us all.

Spotting is an algorithm with a heavy dose of chance and a dash of probability. To simulate "Real World" vision in a 3D environment on even the scale CM mimics would require a great deal of programing I'm sure. But more importantly it's gonna require me winning the lottery so I can afford the computer to play it. And ya know what...it still won't be right.

My personal view as a guy testing the game:

I want it to be the best it can be. I read every comment that players make...both positive and negative. But especially the negative. It is always very helpful when a player with a specific viewpoint provides some kind a facts. In my tiny slice of the game that means a save file (WEGO is best) that demonstrates whatever's not right.

Each of us, as wargamers, has some internal vision of "how it OUGHT to be". And it's always good to voice our views...with or without supporting data... It's how the hobby improves. If that weren't the case we'd all be pushing around cardboard markers and using slide rules. What I have learned in this hobby is: I don't have a corner on what's the right way to make ANY game be better...but my opinions do matter. Relax...enjoy the ride!

[Taki]

Good Idea with that Picture flamingknives. I hope that noone brings up a Picture of a M1 standing 50m down the Road

That would be awesome!

[dan/california]

Steve has at least briefly mentioned that the spotting subroutines do not run every single second since it would bring the processor to it knees. So some, perhaps much, of the lag in spotting newly visible units could be as a result of this.

The good news is that once quad core+ processors have enough market penetration it is a problem that Moore's law can fix.

[Other Means]

Steve has at least briefly mentioned that the spotting subroutines do not run every single second since it would bring the processor to it knees. So some, perhaps much, of the lag in spotting newly visible units could be as a result of this.

The good news is that once quad core+ processors have enough market penetration it is a problem that Moore's law can fix.

I don't think that's it. AIUI* there's a few things going on here. There's a LOS routine that will run against the underlying 8*8m grid to see if LOS is possible. Then there's a higher fidelity 1*1m + height above ground LOS check.

That will happen, to all intents and purposes, instantaneously.

Layered on top of that is the spotting routine. That will take the fact that LOS is possible and then simulate other factors - smoke, dust, unit experience, thermals, binos, fatigue, facing, number of eyes looking etc.

It's this that will give the delay. It's a way of simulating how people would actually be able see on a battlefield.

So any delay you see isn't a lack of CPU resources - how would that work with two opponents on vastly differing systems - but an attempt to make the spotting more realistic.

Given flamingknives illustration above I think spotting movement at 600m would take some time. Especially over hot ground.

*this is all public domain + my guesses so make of it what you will.

[dan/california]

I had not considered heat shimmer, that would make the first hints of motion much more difficult to see.

[Thomm]

Might as well throw this in here: I keep seeing indications of a bug that lets a squad fully identify an enemy squad (no question mark, but unit type icon) but not target it. I am currently playing the campaign, so save games are huge, but perhaps you guys can keep an eye on this.

Best regards,

Thomm

[Kwazydog]

Hi Guys

Hi Dan

Very interesting figures. Was there wind?

Default . In this case it is light from the W, which is behind the MG team.

Whoa! Dan, how are YOU able to get exact round count figures from your M240 test? Are you guys withholding some cool developer's tools? Obviously, this game is broken until we can all get that software!

MG teams have an ammo count from their last reload, and I counted the rest, hehe.

With two men running at the MG team in the building: why would it take 50 meters of running for the MG team to spot them?

Actually I said that they were spotted and *fired upon* at this range, which is quite a difference from just being spotted . As I was testing the lethality of the M240 I wasnt too specific with spotting to be honest, but here is some more detailed info...

The guys actually start movement at 576m (I have a small dip at the end of the 600m map to hide their HQ team). As such in the example I mentioned thats 25m of terrain covered before movement was spotted, the sniper team was spotted and aquired, the information was communicated within the MG team and the guy with the MG opened fire.

After running the test a few times I found that the ? symbol appears at the snipers location within around 1-3 seconds, the squad is properly spotted with 5-7sec. On average the MG guy opens fire with 5-15 of the turn starting.

To put it in perspective, your 600 meter distance is equal to one and 1/2 track laps. If I am watching something at that distance, barring any LOS obstacles, I will IMMEDIATELY see 2 men running.

And so does the engine if you run the test. It just takes a few more seconds to actually identify the target, the infomation to be passed around the team and for the MG guy to aquire and fire at the target. Overall 5-15 seconds doesnt sound like an unreasonable amount of time for this to happen?

Dan

[SlapHappy]

I'm not so sure about engine time requirements to acquire spotting. Intentional time requirements to spot outdoor targets at more extreme ranges might be feasible.

However, troops moving into even small buildings still require a few seconds to acquire spotting of enemy combatants just a few meters away. It is hardly instantaneous. Now, this could be intentional, to simulate other factors - Or it could be a delay caused by internal game check restraints. Only BFC would know the answer to this for sure.

[Kwazydog]

Get it! There aren`t any problems regarding with spotting, nor will there be any!

Taki I have seen you complaining a lot in various threads, but I havnt yet seen you take the time to run some specific tests and to do some research to back up your assumption that the engine is getting it wrong.

We have already done this during the last few years of development and what you see is the current result of that research. We are open to further input though, as we always have been with the CM series. If you feel that something isnt working as it should, run some in game tests, dig up some real world data to show where the engine is getting it wrong and pass on the results.

Dan

[Peter Panzer]

Here's a pictorial update on our rooftop friends from the scenario I described in my original post. I hope this helps.

A quick recap...

UNCON veterans on a flat rooftop 50 meters distant from a veteran M1A2SEP and nine spotting, veteran US Mech Infantrymen.

Below is what the five rooftop ninjas see about 10-15 seconds into the first turn. Insha'Allah.

Martyr Vision:

The next image is what the BLUFOR crunchies and tankers see after more than five minutes of direct observation.

Infidel Vision:

The third screenshot is what one would hope the BLUFOR would observe on a time scale similar to the REDFOR.

See that big ole knob on the left of the M1 turret? Inside it are optics that allow the TC to magnify his field of vision in the following increments: 3X, 6X (wide angle), 13X, 25X and 50X for acquisition. The Gunner and Loader have their own fancy harware to utilize in addition. Of course, there are also the nine pair of eyeballs wired into the infantrymen's wetware plus the binoculars the squad leader is toting.

What the BLUFOR Should See:

Just for kicks, this is what the M1 gunner would see using his GPS-LOS (Gunners Primary Sight-Line of Sight) at 10X magnification. "Hey Grabowski! I can see the hummus stain on that dude's keffiyeh. Second contact from the left, watch where I put this burst of coax..." That RPG tube might command a bit of attention too.

What The BLUFOR M1 Gunner Should See:

What's more, those rooftop guys have a habit of raining down some virtual discontent like a chorus line of Rockettes armed with a case of frags. Five guys on a flat rooftop pitching hand grenades 50 meters(+) at a M1A2SEP/Mech Inf. squad and remaining unobserved for over five minutes, yeah, that squares up.

Guys, please, an adjustment to the manner in which infantry on rooftops is spotted is needed here. The testers will likely want to see how troops on balconies are treated in this regard as well.

I have a savegame file ready, but I do not know how useful it would be. Wouldn't you want to be able to observe the situation over a matter of several turns rather than using a "snapshot?" Perhaps the testers could easily mock up a similar scenario. I would be glad to answer any questions. Other Means what is your e.mail address?

Taki:

Hang in there man. BFC has a long history of making adjustments to their games so long as the evidence to warrant it is supplied, we just have to do our part - they really do listen. The screeshots and savegame file don't lie.

[Drusus]

Layered on top of that is the spotting routine. That will take the fact that LOS is possible and then simulate other factors - smoke, dust, unit experience, thermals, binos, fatigue, facing, number of eyes looking etc.

It's this that will give the delay. It's a way of simulating how people would actually be able see on a battlefield.

So any delay you see isn't a lack of CPU resources - how would that work with two opponents on vastly differing systems - but an attempt to make the spotting more realistic.

I am pretty sure this is wrong. The reason the check is made only once in a while is that there are a lot of them, and the operation is quite heavy.

If you have 200 soldiers on both sides then you would have about 200*200 = 40000 checks to make. If you do this every second, there would probably be no resources left for anything else.

CM:SF uses a lot of tricks to ease the burden on the CPU. But as far as I know, it still has to limit the rate of LOS checks. The way to make this work on vastly different systems is to limit the rate so low that the checks can be run on any system above minimum specs.

BTW I would not be surprised to see multi-threading in CM:SF soon. The return of the blue bar tells me that the graphics calculations and actual turn calculations are perfectly separated already. If LOS checks could be ran on it's own thread and multi-core machines made the minimum, then there would be a lot more resources to throw at this problem...

[Other Means]

Peter Panzer - yep, that looks plain wrong to me.

You can email me through the forum software.

Drusus - LOS checking is a major pull on the CPU.

It speeds things up considerably by doing checks against the 8*8m grid before doing higher fidelity ones.

And bear in mind; when LOS is established you don't have to do the reciprocal check.

[c3k]

Dan,

Aye, perhaps I'll allow you that delay now that you've filled in those details. You may have future in this whole computer game business; keep it up.

Thanks!

[Paper Tiger]

Hi Pete

like the old saying goes... a picture is worth a 1000 words. Yup, that doesn't feel right. I suspect we're seeing this because those are uncon units and they're on top of a building and the civilian density setting is quite high? Further, their stance indicates that they're not actually using their weapons and so the game is treating them as if they are not actively doing anything that threatens the troops. The game doesn't show us civilians so we can assume safely that those US units do actually see those guys on the roof but they're not doing anything suspicious that would make them think "Uh oh, danger". But, they're lobbing grenades? :eek::eek: That'll be what's wrong but I guess this'll get fixed for v1.12 shortly after the Brit Pack arrives.

[SlapHappy]

Easy test. Set up the same situation using Syrian regulars. If they are spotted easily, you have your answer.

[akd]

Hmm...anyone hanging out on top of a roof in the Middle East during the day in summer should be immediately flagged hostile/stupid. Nighttime is a different matter.

[Flanker15]

Easy test. Set up the same situation using Syrian regulars. If they are spotted easily, you have your answer.

I did that:

Clear day with no civilians, same layout as above.

Syrian Rep G normal settings troops on a one story building with no wall and the HQ unit on a 2 story building with a wall. Both in normal stance with a target arc so they don't shoot. US has a M1A2SEP, a recon squad and HQ all normal settings. No order just pointing at the building 100m. 30min game, building guys spot enemy in seconds, ground guys spot nothing in 30min. Gave the rifle squad a move fast order to the ground flaw and back repeatedly one turn, didn't get spotted. Gave them a move order out the door, they got spotted a few seconds after leaving the building. Aslong as I didn't shoot and stayed in the building I was invisible (I could move around on the open roof quite alot). I also ran one of the US squads up and they got pelted with nades but that didn't reveal the enemy (no ? or anything just nades from god!). They got spotted when they started shooting after running out of nades.

I've got some screens and the mission in a zip file, link 7Mb:

http://www.megaupload.com/?d=9Q7CZ9NI

[Flanker15]

Ok just to make sure I switched the side posistions and the Syrians have no better luck spotting the US on the roofs. So it's not just a Syrian bug, I gave them a T90 aswell to make it even.

This and experience with the game in general give me the conclusion:

There is a extremly high chance that a unit in a building will remain undetected unless: it shoots it's gun (grenades are ok) or it leaves the building. An enemy unit is able to move right up to the unit (tricky to make them not attack) and won't detect it even if it is standing on the unit (touching distance). Skill and equipment seem to be ineffective here.

Basically a unit in a building is invisible until it shoots, you can even move around the building. The result is: it is impossible to avoid ambushes, the guy in the building will always get the first shot. Even when he is standing on a roof in plain sight or running around in the building.

What I haven't tested is whether this ability requires the unit to start in a building or if they can be invisble after entering a building from outside.

Edit: ok I tested that and aslong as the unit isn't spotted going into the building it will be invisible once inside.

[Taki]

Basically a unit in a building is invisible until it shoots, you can even move around the building. The result is: it is impossible to avoid ambushes, the guy in the building will always get the first shot. Even when he is standing on a roof in plain sight or running around in the building.

Some ppl really that that this should be realistic. For me its just anoying. Agruing from one side or another dint do the trick.

So maybe i should start a Topic like:

"What is the US Army doing to prespot ambushes wich we can simulate in the Game well?"

[c3k]

Is the game engine treating rooftop locations the same as locations INSIDE buildings, as far as spotting is concerned?

[MarkEzra]

Is the game engine treating rooftop locations the same as locations INSIDE buildings, as far as spotting is concerned?

Good question...I think not yet. BFC is the obvious final authority there.

My Personal Opinion based solely on long time play of All CM's games:

I seem to recall that All roof tops simulate clutter that acts as cover. When I see Pics that demonstrate a perfectly unhindered view of troops I try to remember this. The Old CM engine was a total mass of abstracted graphics. CMx2, with it's 8 meter grid and near 1x1 Inf Graphic approximation has gone a long way but is not 100%. Nor will it be for all the well known technical reasons (like my cheap PC and old graphics card...Oh...and yours, too.. )

But buildings remain largely abstracted. The doors and windows work nicely and are a huge step forward. But the immediate exterior, the interiors, stairwells, and roofs remain abstracted.

I have seen it mentioned often in the past, by those who were unhappy with CMx2 graphic sim, that their preference was to play the highly abstracted graphics with it's combat resolutions hidden "under the hood" and just use their own mental imagery to represent the actual combat. Good enough. I like radio plays, too. But movies are kinda cool...especially the "talkies". BFC seems motivated to move forward in representing as much of the 3D world as Computing power allows for the average guy. But as many vets have noted: No game, no simulation, Can EVER represent the pure hell and confusion of Humans at war. And that may well be a blessing.

Again, just a personal view and not meant to hinder conversation on improving spotting methodology.

[Peter Panzer]

Basically a unit in a building is invisible until it shoots, you can even move around the building. The result is: it is impossible to avoid ambushes, the guy in the building will always get the first shot. Even when he is standing on a roof in plain sight or running around in the building.

...and we have a winner.

Other Means:

Your e.mail address is not available in your profile. Given Flanker 15's link, do you still require my example or does his suffice?

[slug88]

...and we have a winner.

Other Means:

Your e.mail address is not available in your profile. Given Flanker 15's link, do you still require my example or does his suffice?

I disagree with this. Just now I played a scenario where a US scout squad was spotted by a T-72M1V from about 50m as they were running up flights of stairs. The squad had yet to fire any weapons at all, and there had been no blue fire originating from the building for a good 5 minutes or so. Despite this, the T-72 spotted the squad and annihilated it.

[Other Means]

If you click my username you can send me a message.

Or it's othermeans, My ISP is blueyonder which is a company in the UK.

Spam - me hates it.