Steel Beasts vs Combat Mission t-72 visibility test

[domfluff]

8 minutes ago, Bozowans said:

I always assumed that spotting in CM worked like dice rolls. Like perhaps the game engine constantly draws lines in the 3D world between friendly units and enemy units at all times. If that line is intersected by another object in the 3D world, like a building or hill or a tree's "hitbox" or something, then nothing happens. If the line is not obstructed though (or partially obstructed), then the game engine gives you an X percentage chance to spot the unit at Y distance every few seconds. So like a 10% chance every few seconds to spot a unit at 2000 meters. Sometimes you get lucky and sometimes you won't, but if you have 10 tanks all lined up in view of something, chances are that one of them will spot something very quickly even though most of them won't. And then the tanks will talk to each other and share the info. 

Maybe I am terribly wrong about how all that works.

Things we know:

The Target tool is based on a lookup table, generated by the map. This takes five heights (crawling, standing, small vehicle, vehicle, tall vehicle) and finds each square to each other square.

This is why the Target tool can be so fast, and why it's always been fast, even when the engine itself wasn't as hot.

Spotting and Shooting (line of sight and line of fire) are two separate calculations. Spotting is done from the model's eyeballs, and a firing solution is plotted explicitly - the actual chap represented plots true line of sight when trying to make a shot. The shot itself has a degree of randomness, of course, and one factor that is  factored into that are a "hinderance" value from intervening terrain tiles (much like ASL, etc.).

Spotting uses a similar method, but doesn't use exactly this method. Doing this would be expensive in CPU terms (and especially was in 2007). Instead, each unit is given a spotting cycle. This defaults to seven seconds (i.e., there is a check every seven seconds for what this unit can see), but this time is flexible and will vary based on aspects like distance to the enemy.

This spotting cycle is often a cause of complaints - it's possible to force some weird results if you charge a Jeep directly at a tank along a road, moving as fast as possible. If you happen to catch this at a weird step in the spotting cycle, then this will not update until the Jeep is in an apparently ludicrous position.

So there are many factors at play. Even if there were no RNG elements involved at all in the spotting calculations, the resultant calculations are far more complex than something human-observable. Since variance, hidden information and sufficient complexity can all be mathematically equivalent, even if there were no random elements at all to the spotting system, you'd still end up with something which is to all intents and purposes, "random". If you can't possibly predict the outcome from knowing the starting conditions, that's "random" whether or not it involves any dice rolls.

[domfluff]

There's some more, but that's a good start.

[Sgt.Squarehead]

3 hours ago, Artkin said:

Even after getting blasted by the other M150. Strangely the game even says "enemy target" despite me not having a spot on the vehicle whatsoever.

Missed that bit.....That does sound odd! 

@The_Capt  Did you notice this?

[The_Capt]

11 minutes ago, Sgt.Squarehead said:

Missed that bit.....That does sound odd! 

@The_Capt  Did you notice this?

Huh?  Not sure what is happening there.  So Artkin is saying that the M150 is lit up even though no one can see it?  My advice for all Soviet tanks is to open up, they spot dramatically better (except maybe for the T80) opened up...learned that one the hard way during the AAR fight.

A lot of this sort of behavior is what makes CM realistic to be honest...and frustrating at times.  The player is god, all seeing and floating over the battlefield.  The player can clearly see the M150, or even if it reduces back to a "?" spotting, you know it is there.  The TACAI is not really self-aware, and that is ok because troops in combat often are nor either (scope eye is a real thing).  So CM models a bunch of weird stuff under the hood that often leaves the player screaming "but they are right there".  More realism = less control, (still too much control if you ask me but it is a game after all).

So looking at Artkins screen shots, is the tank out of C2?  It is buttoned up, so if it is also out of contact it is about as bad as it can get for a Soviet tank. [went back and checked and in one shot, yes the tank is out of C2 with platoon and company which is not good]

[chuckdyke]

8 hours ago, Sgt.Squarehead said:

Where?  It was always my impression that more spotting attempts were made within the covered arc.

Yes and besides the unit or units look in the right direction. 

[Artkin]

Yeah well their CO is alive but their PLT leader ate it just like every other T64A.

Opening up is against soviet doctrine for obvious reasons

[zmoney]

@akdso if spotting to the rear is the worst 90 degree cone and I place a cover arc to the rear, would it not increase spotting to the rear 90 degree arc?

 

I have seen many instances of my atgm teams not spotting until I narrow their, with cover arc, zone to scan. Perhaps it was just a coincidence. I do believe like you have stated that it can also be useful to give HE units cover arcs so they won’t hit close friendly units as well. But just because you give a cover doesn’t mean they disregard everything outside of the arc, I have many times witnessed my units with cover arcs spotting things outside their respective arc. And on occasion seen them fire on something outside the arc as well.

Edited October 17, 2021 by zmoney

[akd]

11 hours ago, zmoney said:

@akdso if spotting to the rear is the worst 90 degree cone and I place a cover arc to the rear, would it not increase spotting to the rear 90 degree arc?

Your unit would turn to face that direction making the center of the arc the new front facing.

Think about this: micromanaging arcs to improve spotting would create a game within the game that the AI cannot play at all.  Every one of you (except Sgt. Squarehead; he’s special) that believes arcs improve spotting is using your god knowledge of enemy unit locations to place narrow arcs on the locations of enemy units that you have knowledge of, but that the unit does not (if it had C2 knowledge it would receive a ? spot that would itself increase chances of spotting).  That is certainly a game to be played if it worked that way, but it is in no way realistic in overall outcome.

Edited October 17, 2021 by akd

[ASL Veteran]

[Sgt.Squarehead]

14 minutes ago, akd said:

Think about this: micromanaging arcs to improve spotting would create a game within the game that the AI cannot play at all.  Every one of you that believes arcs improve spotting is using your god knowledge of enemy unit locations to place narrow arcs on the locations of enemy units that you have knowledge of, but that the unit does not (if it had C2 knowledge it would receive a ? spot that would itself increase chances of spotting).  That is certainly a game to be played if it worked that way, but it is in no way realistic in overall outcome.

No.

I never place a target arc for a vehicle (with the intent to directly kill another) unless it already has a tentative contact.

To clarify, I do place target arcs for coverage etc. but I don't focus them on known enemy locations unless the vehicle in question already has a tentative contact.

[dbsapp]

9 hours ago, Combatintman said:

Ok ... so are those results consistent with what you are seeing in CMCW?

Yes, I found that M60's behavior in CMCW is much more consistent with Steel Beast test, which is in stark contrast to T-72 behavior.

I ran several M60 test in CMCW swapping M60, who now became a hunter, and t72, who became a prey (well, has always been in fact). 

On average it took about 1 minute for M60 to spot t72 and destroy it (the fastest result was 30 seconds which is very close to Steel Beast, the longest about 2 minutes).

In all tests M60 killed t72. 

The start: 

The end:

 

The conclusion: where as there are some differences in M60 behavior between CMCW and Steel Beasts it's in more or less acceptable range. At least in both games the tank acts as expected when it has clear sight on the target and advantageous position.   

It is T-72 behavior in CMCW that produces heavy divergence from Steel Beast.

The Scenario mission link.

[chuckdyke]

On 10/16/2021 at 11:49 AM, dbsapp said:

Steel Beasts is a tank simulator that is used in several countries to train military personnel. 

Combat Mission Cold War and Steal Beasts have a lot of things in common, so it would be interesting to compare how two games simulate combat. 

In order to do that and to make as precise experiment as possible I put t72A (m1) tank that is featured in both games against M60 TTS tank that is also present in both of them. 

To keep experiment clean I used default "flat map" and the same weather and time conditions - clean weather, the time is June the 1st, 12:00. 

In CMCW I had to use additional "formation" units of observers, but I put them behind tall walls, so they didn't interfere in the process. In CMCW skill level of both of the tanks was put on "regular". Steel Beasts doesn't have skill level feature. 

Under created conditions t-72A looks directly at M60's side. It is oriented from South to North, M60 - from West to East. The distance between them is 2 km. The conditions are the same in Steel Beasts and Combat Mission Cold War. 

That how it looks like in Steel Beasts: 

How it looks like in CMCW:

What T72 gunner sees from his position from the very start:

As you can easily notice, M60 is immediately and perfectly visible from gunners sight.

The same with T72's Commander's sight:

The results:

In Steel Beasts t72's AI spotted M60 almost immediately, which is not surprising, taking into account that it has perfect view on the target. It took t72 about 2 seconds to spot the opponent and about 18 sec to hit and destroy it. 

In CMCW something opposite happened. I ran several tests and t72 couldn't spot m60 once.  Its optics was not enough to spot the tanks directly ahead of it at the distance of 2 km during clean daylight.

In fact, every time M60 spotted t72 first and killed it. It took about 2-3 rounds and from 1,5 min to 5 min to kill t72.  t72 didn't see the opponent despite m60 was firing at him. 

How it ends in Steel Beasts:

How it ends in CMCW:

The CMCW test scenario is attached.

You can make your own conclusions. 

 

 

T72VISION TEST.btt 32.37 kB · 6 downloads

The examples here would result in a contact in CM but not in a full contact in which you can identify the spotted AFV. Result in CM your AFV won't open fire. 

Edited October 17, 2021 by chuckdyke

[S-Tank]

T-72 vs M60A1, both tanks are regular with + 0 leadership, facing each other at 2000 meters.

The T-72 spots first, fires first, but misses. Note it has only been 25 seconds since mission start. Pretty good eh?

The second round impacts the ground just as the turn ends. Getting closer!

The M60 finally spots the T-72 and returns fire at 28 : 16.

A full minute later the the T-72 fires its third round. Almost there.

First blood to the "inferior" tank! The driver of the M60 is killed and the engine destroyed. It took the T-72 four rounds to find the range, which seems perfectly acceptable performance for a tank without a laser range finder at that range.

The rattled crew of the M60 pops smoke.

 

A few minutes later the smoke clears and the T-72 finishes off the hapless yankee imperialist machine with its fifth round.

 

Edited October 17, 2021 by Demo_

[Combatintman]

6 hours ago, dbsapp said:

Yes, I found that M60's behavior in CMCW is much more consistent with Steel Beast test, which is in stark contrast to T-72 behavior.

I ran several M60 test in CMCW swapping M60, who now became a hunter, and t72, who became a prey (well, has always been in fact). 

On average it took about 1 minute for M60 to spot t72 and destroy it (the fastest result was 30 seconds which is very close to Steel Beast, the longest about 2 minutes).

In all tests M60 killed t72. 

The start: 

The end:

 

The conclusion: where as there are some differences in M60 behavior between CMCW and Steel Beasts it's in more or less acceptable range. At least in both games the tank acts as expected when it has clear sight on the target and advantageous position.   

It is T-72 behavior in CMCW that produces heavy divergence from Steel Beast.

The Scenario mission link.

How different do you think this thread would have gone if you'd run those tests and included them in your original post?  A lot of these 'XXXX is broken - fix it Battlefront' threads go the way this one has gone because of the way the original assertions are made.  As others have pointed out, and yes it may be a misinterpretation of your intent, your posting history makes people think that your intent with these threads is to further an agenda that there is a Battlefront conspiracy to make the Soviets deliberately weak.  This is why I talk about the presentational aspect of your initial post.  You may be right about T-72 spotting, but I'll let others who have greater knowledge and like doing these sorts of tests continue the discussion.  I will also be pleased if you are right and a fix is applied in a future patch.

I personally have not seen much of the T-72 in-game because it wasn't in GSFG at the time and so the stuff that I've produced, which I prefer to bear some resemblance to reality, doesn't have them.  I think I tested an early version of MikeyD's Between Two Autobahns which had them but didn't see anything that looked particularly odd because of the ground layout and how I put my plan together.  I did not expect the Soviet force to have many opportunities to get eyes on and target me and pleasingly enough that is pretty much what happened.

Personally I agree with others that Steel Beasts vs Combat Mission is an apples and pears comparison.  Of course there is some common ground in what they're simulating but they are fundamentally different beasts  .  Steel Beasts I think is probably peerless in its simulation of being an AFV crew member, without the joys of track bashing, getting knocks and bruises from solid bits of metal, living and working under someone's armpit and a heck of a lot of noise, and its sweet spot is AFV vs AFV combat.  It does other things less brilliantly.

[kevinkin]

Six pages of posts in <48 hours, is that a record?

 

[Bufo]

18 hours ago, akd said:

Targeting arcs do not increase chance of spotting inside the zone of the arc beyond:

1. Their effect on unit facing.

2. In the case of turreted vehicles, their ability to set different facings for turret vs. hull crew.

I believed the same but then I found a post from Steve:

"there is a bonus for spotting things within an Arc"

Also:

"The spotting bonus is only within the colored area of the arc. Outside of that there's no bonus."

Links:

 

Edited October 17, 2021 by Bufo

[MOS:96B2P]

49 minutes ago, Bufo said:

I believed the same but then I found a post from Steve:

+1.  Nice find.  Now, I wonder if target arcs still have a small spotting bonus or if something was tweaked since 2011?

Same as @akd I have believed that there was no spotting bonus within a target arc.  Now I'm not sure........ 

[Sgt.Squarehead]

1 hour ago, Bufo said:

I believed the same but then I found a post from Steve

+1 from me too.....Knew I'd read it somewhere, but wouldn't have had a clue where to begin looking.

[ASL Veteran]

Well there was a very lengthy discussion about arcs (behind the veil) and he never once mentioned that there was a spotting bonus to arcs.  I don't recall that he specifically said there wasn't a bonus or not, but that was a part of the discussion and he never said there was either (as far as I remember).  So if there is a bonus then perhaps he forgot (entirely possible) or perhaps it was altered later (also possible, but probably unlikely).  I would guess that the bonus is probably very slight assuming there is one.

[akd]

Did you all just skim right over Steve’s last post in that thread?

Quote

OK, I poked Charles into giving me some more details about how things work under the hood. Turns out I got this same info from him back in May and posted it here (somewhere). Guess I should post it again:

 

 

  Quote 

The spotting is not tied directly to the arc. However spotting success is related to your facing. You spot stuff to the front much better than to the rear. And when you have a target arc, you face right down the center of it, putting the arc in your "best front angle". So you will spot things better there than outside of the arc. The arc itself isn't the driving force though -- it's the spotter's facing. But the target arc enforces facing discipline so you can be sure your men are looking right where you want them to look.

 

So the "bonus" is, as I mostly remembered correctly, primarily due to the Facing of the unit and not a special bonus artificially assigned. This should clear up all questions being asked in the past page or two. Specifically:

 

 

1. The primary spotting advantage of a CA is to keep the unit from shifting it's Facing due to other tactical distractions. Meaning, if you absolutely want to make sure you keep a narrow portion of the battlefield under observation, CAs work better than non-CAs in theory. But reality means it comes down to distractions because...

 

2. A unit looking in the same exact direction in the same exact situation will spot exactly the same whether it has a CA or no CA (and not Hiding, obviously).

 

3. Units that have a 360 deg CA don't see any benefit from it other than limiting engagement range.

 

4. As I've said several times now, the "bonus" isn't that big of a deal. People trying to "game the system" by using CAs to increase spotting chances are wasting their time and possibly causing themselves some harm.

 

Hope that helps.

 

Steve

 

Edited October 17, 2021 by akd

[Thewood1]

Spotting has always been at the crux of complaints around the CM games.  Its more the inconsistency in it and the fact that no one seems to really know how it works in detail.  I think the LOS scanning cycle is part of the perception issue.  I used to bump into issues infrequently where I could drive a Sherman right up to a Tiger and the Tiger wouldn't spot for a half a turn.  It doesn't happen often, but it happens.

Not sure anything can get at those infrequent outliers.  Steel Beasts, which I think represents cold war battles better than CM, has its own issues with insta-spotting of infantry and other things.  I do find it more consistent in its spotting, good or bad.  And it does some very good things around combined arms.  There is a wargame mode for it and its still my go to for cold war and near-modern tactical combat.

[ASL Veteran]

2 hours ago, Thewood1 said:

Spotting has always been at the crux of complaints around the CM games.  Its more the inconsistency in it and the fact that no one seems to really know how it works in detail.  I think the LOS scanning cycle is part of the perception issue.  I used to bump into issues infrequently where I could drive a Sherman right up to a Tiger and the Tiger wouldn't spot for a half a turn.  It doesn't happen often, but it happens.

Not sure anything can get at those infrequent outliers.  Steel Beasts, which I think represents cold war battles better than CM, has its own issues with insta-spotting of infantry and other things.  I do find it more consistent in its spotting, good or bad.  And it does some very good things around combined arms.  There is a wargame mode for it and its still my go to for cold war and near-modern tactical combat.

Trust me, BFC is very aware of all the spotting complaints.  Personally - just speaking for myself - I think there is a combination of factors that skews things a bit.  For one thing I'm not sure that gamers realize how difficult it is (per veteran accounts as well as various 'spotting tests' done in the 19th century - the human eye hasn't changed since then) to spot anything at all on the battlefield - up to and including stuff that's firing directly at you.  I'm also not sure that gamers fully appreciate what a human being looks like at 700 meters (for example).  On the one hand you have to make spotting such that a game can function (the empty battlefield) and on the other hand you have to try and make something that an average gamer can reasonably associate with reality and that's a difficult wire to walk (gamers who may or may not have an appreciation of what something actually looks like 800 meters away).  I think probably the biggest spotting 'hole' in the game (if you will) is probably anything movement related.  Movement draws the eye and assists in spotting something.  Binoculars will bring something a lot closer through magnification, but of course your field of view is way more restricted than it would be with the naked eye.  So typically I would expect that if something was stationary (even sitting in the open) it would be difficult to spot at various ranges (for example, an infantryman standing in the open might not be seen with the naked eye beyond 700 meters if he isn't moving).  However if something is moving you should notice it even with peripheral vision with the naked eye at reasonable ranges (not at 2000 meters for example) and then binoculars could be used to scan the specific area where movement was detected in order to firm up the 'spot'.  Infantrymen could walk through an open field at 2000 meters and it's unlikely that you would even know they were there if you were looking just with the naked eye. 

Edited October 17, 2021 by ASL Veteran

[Vanir Ausf B]

The inconsistency is mainly a function of the high variance. At 2000 meters tanks can spot each other in 5 seconds or 5 minutes. The average may be realistic but when a player rolls snake eyes on the spotting dice they think something is broken, especially if they are testing on a "firing range" which is not the environment the spotting model assumes. The other issue is that because spotting checks are by far the most CPU intensive task in CM they are every 7 seconds rather than continuous.* This means fast moving units can occasionally move over open ground without being seen for several seconds or even longer if first spotting check is snake eyes.

*There are exceptions to this, I think mostly when units are firing.

[Sgt Joch]

So....

talking about the difference between spotting “on a range” and spotting in combat when you know you could be dead anytime...

Gabby Gabreski, the US WW2 Ace recounted his 1st time in combat in his autobiography. He is flying as wingman, all of a sudden calls of “Bandits!”  are all over the radio. He starts looking everywhere all around him, but spots nothing. All of a sudden he hears on the radio: “check your 2 o”clock!”. He looks and sees a 109 as big as a house, barely 200 meters away that he was slowly gaining on. Before he can react, the 109 spots him and dives away...

[Artkin]

Thanks for the anecdotes, it really seems like I rolled a snake eyes about 100 times in a row. Certainly a first for me.