Jump to content

Need odd LOS issue looked at


Lt Bull

Recommended Posts

Hi,

I am back playing CM (PBEM) after taking a bit of a break from it. Reasons for the break was a culmination of the gerneral disatisfction and disillousionment with BS "official" responses (or lack of any real response) to several gameplay mechanics issues and suggestions for improvment (spotting, LOS, movement etc) that I either had experienced myself games I had played myself and/or read about in the forums.  I was hoping that with all the patches and upgrades that have been released, perhaps the kind and frequency of "this makes no sense, what the hell is going on" kind of moments I saw in games that had worn thin on me and predicatd my need to just walk away from the game would have perhaps be addressed to a point where the likelihood of experiencing those off-putting moments in game had diminished.

Well first PBEM back, latest patch, upgrade etc.about 20 turns in but only about 10 moves in from encoutering the enemy, I have had one of those same "something seems wrong with the mechanics" moments that kills/had killed my trust in the game.  :(  And no it's not the "retreat towards enemy" issue that many have posted about (hope it wont be an issue in my games).

The issue relates to an instance during the last minute of action where the LOS mechanics has been (once again) broght in to question, reinforcing doubts I wish I never had about this aspect of the game.

Do I need to bash my head explaining exactly what happened/justifyign my reasons for concern in a forum post here or is there someone at/with BF who can just review the replay file(s) and spare me the effort? I did go to support but they just seem to deal with the general tech issues.

Happy to provide PBEM files of course (will ask opponnet as well).

Regards

Bull

Link to comment
Share on other sites

I advised my opponent of the issue and this post.  I don't want to go in to too much detail in an open forum as our game is still being played and it might disrupt the FoW side of things.

I can tell you that the scenario is called "A Nasty Surprise".  The sceanrio starts at around 7:15am, weather is clear but I think LOS range is reduced due to the lower dawn light condtions in effect.   All the terrain I refer to is essentially flat. I have a unit located in rear part of the map able to trace a LOS 400m across open ground, up to and through wooded terrain about 40m deep, then another 500m across open ground on the other side.  All other units near and around the unit in question can not see through the woods at all.

And you know how I said I hope that "retreat towards enemy" thing doesn't happen tin our game?...

In the very next replay I watch of our game, an infantry unit located at the edge of wooded terrain became "rattled" (apparently caused by the incoming direct AP shell gunfire targetting and destoying a friendly tank about 10m behind it in the woods), then broke cover at edge of woods and started running across open terrain towards towards the enemy side of the map. There was some arty dropping perhaps 75m away.

Link to comment
Share on other sites

11 hours ago, Lt Bull said:

I was hoping that with all the patches and upgrades that have been released, perhaps the kind and frequency of "this makes no sense, what the hell is going on" kind of moments I saw in games that had worn thin on me and predicatd my need to just walk away from the game would have perhaps be addressed to a point where the likelihood of experiencing those off-putting moments in game had diminished.

Why would you think that? Patches fix know bugs that are recognized by BFC. They do not fix undefined issues that any random player may or may not feel is in the game. Not to mention that spotting is not meant to be easy and there is no insta-spotting feature in CM. So, you are meant to be unhappy with the imperfections of your men's spotting abilities. It is a feature not a bug.

 

You don't really specify what the spotting issue you are feeling upset about so I no one can tell if you have encountered something that might be a bug. I can only speculate about it based on the comment below. You can tell us if that might be in the ball park of your feelings on the issue:

5 hours ago, Lt Bull said:

The sceanrio starts at around 7:15am, weather is clear but I think LOS range is reduced due to the lower dawn light condtions in effect.   All the terrain I refer to is essentially flat. I have a unit located in rear part of the map able to trace a LOS 400m across open ground, up to and through wooded terrain about 40m deep, then another 500m across open ground on the other side.  All other units near and around the unit in question can not see through the woods at all.

The most important thing that we all have to remember is that darkness and fog are not nearly as dark and dense for us players as it is for the pixel troops. This is for game play-ability reasons. So, we have to temper our expectations for our units ability to spot. Also the LOS tool shows you the maximum effective range that a unit can see. It does not show the distance at which the enemy will be insta-spotted. There is no insta-spot feature in the game.

Link to comment
Share on other sites

6 hours ago, Lt Bull said:

In the very next replay I watch of our game, an infantry unit located at the edge of wooded terrain became "rattled" (apparently caused by the incoming direct AP shell gunfire targetting and destoying a friendly tank about 10m behind it in the woods), then broke cover at edge of woods and started running across open terrain towards towards the enemy side of the map. There was some arty dropping perhaps 75m away.

Not running towards enemy (who I suspect they may not have been fully aware of).....Running away from explosion (that they were all too aware of).

IMHO

Link to comment
Share on other sites

It was 'user error' to place your infantry so close to the tank. If the tank explodes (as they are often wont to do in the game) the infantry will at the very least be suppressed or rattled, or as likely panicked and wounded. Infantry considered tanks to be artillery/mortar magnets and did not like having them close by. This isn't a game mechanics bug, its you being punished for making a tactical mistake.

Edited by MikeyD
Link to comment
Share on other sites

Hate to say it, but I can confirm that both the suicidal rout behavior and bridge bug are still issues. I'm working on an urban map for CMRT with extremely plentiful cover and the ability to get out of enemy LOS quickly. However, it's a given that during any battle, a certain number of units will occasionally pass on easy cover to take the most dangerous path. In my most recent battle, I've had 6 units do it for around 20-25 casualties total. They were part of three squads, an MG and two LMG teams. MG was green, squads were reg and vet and LMG (IIRC) reg and green.

The usual pattern is for a unit in a building to panic for whatever reason and then run out of the building in a dangerous direction and get cut down. Sometimes, they exit out the door facing the enemy and get cut down. Sometimes, they run out on a safe street and then turn the corner and get cut down by the guys that the battalion has been watching for an hour. One of the LMG teams kinda' did a Benny Hill Yackety Sax thing by running back and forth across a street being covered by an MG42. Basically, the team ran out of a safe building, dashed across the street to another building, got panicked again and then ran back across the fire lane towards their original location. Eventually, they got gone.

Most, often, when units on a higher building level panic, they run to the first floor and stay there. Sometimes they move horizontally one or two AS as well while descending. Or, if they exit, they exit out the back or side.  The higher the quality of the unit, the more likely they'll make a good decision. But not always. The veteran squad got spooked and ran out of the front of a very safe building (lots of close safe action spots just to rear) towards a shellhole in the enemy's kill zone. That was also towards the enemy map edge.

From what I can tell, panic triggers a unit to displace a certain number of actions spots. The exact number appears to vary. The direction and distance is probably determined randomly with situation-specific modifiers  and unit-stat-based modifiers applied. Kinda' like Heat of Battle in ASL. So, a veteran unit is more likely to choose the smarter/safer options, but can still "roll" badly and choose dancing in the enemy's known fire lane as opposed to running backwards 8m or less into the out-of-LOS building section right behind them.

So, what some folks are calling a "bug" is code that needs massaging or enhancing. When I say massaging, I mean tweaking so that "rolling bad" happens a lot less often (like, forget about shellholes if a building action spot is within so many meters and give preference to one in the friendly map direction). By enhancing, I mean perhaps adding a layer of awareness somehow, so that when panic hits, known enemy units (like the Panther around the corner that everybody in C&C has known about for nearly an hour) are factored in.

Of course, I'm just guessing about the code here. I've no idea. But will say that the current behavior is not adequate in that, while it may get it right a majority of the time, it gets it wrong fairly often, too. And when it does, it totally blows the immersion because it cannot be rationalized as real world human behavior. Real soldiers panic, but they don't panic in the way that the code is producing when it "rolls bad."

On the bridge bug, I had a bunch of instances of it during a recent play session. The first crossing or two was okay. Then a unit got stuck. Reverse seemed okay, so backed the unit of the bridge, and then when across in reverse. That worked. But eventually, units got stuck even when reversing. None of the units got permanently stuck like in the old days, though. I was able to nurse them either across or more likely back off the bridge to where they started. I think I still have that turn save.  I'll submit it (may take a few days).

I'll also start cataloging when the suicidal rout occurs, so I can give specifics and turn saves, if desired.

Edited by Macisle
Link to comment
Share on other sites

1 hour ago, Macisle said:

Hate to say it, but I can confirm that both the suicidal rout behavior and

I don't think what you are seeing is the same as the one that got into the 4.01 patch. See below...

1 hour ago, Macisle said:

bridge bug are still issues.

Bag it and tag it. Which is to say if you have a save game send it along  you can PM me if you like.

 

1 hour ago, Macisle said:

The usual pattern is for a unit in a building to panic for whatever reason and then run out of the building in a dangerous direction and get cut down. Sometimes, they exit out the door facing the enemy and get cut down. Sometimes, they run out on a safe street and then turn the corner and get cut down by the guys that the battalion has been watching for an hour.

This behaviour is obviously sub optimal but I have see this happen in urban fighting for a long time. Do you think it is more common now? Do you have stats to show that? Not that it really matters. If you have a poor behaviour example...

1 hour ago, Macisle said:

I'll also start cataloging when the suicidal rout occurs, so I can give specifics and turn saves, if desired.

... you can do the same.

Link to comment
Share on other sites

6 hours ago, IanL said:

I don't think what you are seeing is the same as the one that got into the 4.01 patch. See below...

Bag it and tag it. Which is to say if you have a save game send it along  you can PM me if you like.

 

This behaviour is obviously sub optimal but I have see this happen in urban fighting for a long time. Do you think it is more common now? Do you have stats to show that? Not that it really matters. If you have a poor behaviour example...

... you can do the same.

Ian, I actually tagged two instances of the bridge bug in save game files. It's the same bridge, but two different battles. I'll PM you those. I should have tagged those instances of suicidal rout like I did the bridge bug, but was trying to push through as much map building and testing as possible during my week off (got a lot done!). So, I didn't follow best practices on file naming. I may have saves, though. It will be time intensive to find them (very long battle, big map and no SSD), but I might take a shot over the weekend. From now on, I'll tag game saves whenever it happens. Do you prefer bundles when collected, or singles as they come in?

I can't say on frequency. Just on feeling, it seems like the rout behavior was a little less suicidal in the first version of the patch. But that's just a feeling. All the saves should both be under CMRT 2.02.

(Edit: two bridge bug files sent.)

Edit part deux:

One potentially low-time-investment fixer idea for suicide rout that just popped in my head is, for what it's worth, this:

If unit quality has a big impact on tendency to suicide rout, which it seems to have, make all units have the maximum beneficial modifiers under rout conditions, but penalize them according to unit quality by adjusting their panic "cool down" period up or down. So, a green unit would be as smart as an elite unit as far as where to rout, but would take much longer to return to a state where it could be given orders.

Just an idea.

Edited by Macisle
Link to comment
Share on other sites

On all the commentary regarding the "break cover, run across open towards enemy map edge" issue I mentioned (incidentally and in passing really) that just happened to occur to an infantry unit in my PBEM:  My understanding (and expectation) of the "flight"/retreat/rout etc. mechanics of units is that all units on a map "know" which direction leads to the "friendly map edge" for their side. The ACTUAL direction they take however may be affected by local factors.  I understand that in the instance I explained, if the infantry unit was to simply run a path "towards a friendly map edge" it would mean they would have to literally  run towards and through the tank being attacked/destroyed which would be odd in itself.  Taking a direction laterally however would seem a more appropriate thing to do in this instance (ie. away from "local known issue", not towards "known" enemy map edge, not away from "friendly map edge", not relocating from covered terrain to open terrain).  It did mean in my game the unit got cut down and eliminated as a result of "running across open ground directly towards enemy map edge".  I will let other's decide on whether this is related to the pre-existing  "break cover, run towards enemy" issue that first started being discussed after the upgrade.

22 hours ago, IanL said:

You don't really specify what the spotting issue you are feeling upset about so I no one can tell if you have encountered something that might be a bug. I can only speculate about it based on the comment below. You can tell us if that might be in the ball park of your feelings on the issue:

The "this just seems wrong" LOS aspect of what I described is that one unit seems to be able to trace a wedge of unobstructed blue line LOS from it's position around 400m back, through a section of wooded terrain (between 40-90m deep), up to a further 400m beyond the woods, whereas other units nearby and closer to the woods itself have no visibility through the woods at all.  This "x-ray visioned" unit was also able to spot an enemy unit beyond the trees that other units with clear LOS at much closer range and with only open ground between them and the enemy unit did not spot.

I know there is a level of abstraction with modelling LOS through "woods"/trees but just one unit being able to see unobstructed through 40-90m of wooded terrain while other units LOS is totally blocked seems like something unintentional/undesirable is going on in the LOS modelling and this unit for some reason.

Edited by Lt Bull
Link to comment
Share on other sites

1 hour ago, Lt Bull said:

I know there is a level of abstraction with modelling LOS through "woods"/trees but just one unit being able to see unobstructed through 40-90m of wooded terrain while other units LOS is totally blocked seems like something unintentional/undesirable is going on in the LOS modelling and this unit for some reason.

I think you are hitting the short comings of abstraction: you can end up with some cases that are not handled well even if the aggregate is pretty good. Sorry I think you are facing one of those cases where things have not worked out for the best. In case you have not already seen it you might like to read this:

I don't think there is anything to be done. I typically set my expectations for trees that they likely will give me good blocking of LOS but I am aware that I cannot count on it. That puts me in a mind set to look for terrain features to hide behind or observer from when ever I can even if there are lots of woods around. That minimizes the number of times I am surprised. It is inevitable that surprises happen though.

Link to comment
Share on other sites

I normally try to not get involved in forum discussions, because I find they devolve into petty back and forths, however as the opponent on the opposing side I'd like to contribute my view and hope to provide some clarity and some thoughts that may have been overlooked. First off, I don't see anything from my perspective in the game, or based on Bull's explanation that points to any sort of bug or needs any review. He asked me to submit my turn and password for review, which I declined. But I'll give my perspective.

The game is early morning in winter. Visibility is very short, infantry has been making contact at less than 100m, in some cases infantry has been within 50m before being noticed. Firing and muzzle flashes change that of course, and a unit firing can be seen at a much greater distance, as can units moving at a faster or more noisy pace. (Also unit makeup, type, experience, quality, motivation, proximity to command, movement or lack there of ALL have input into how one unit functions compared to another. So there is almost no such thing as an apple to apple comparison in this game.) The LOS line tool shows a blue line going out much further than what visibility actually is, and this is in keeping with what I have seen in all other CM games. The blue line is a guide, which I use as a best case guide, not as a sure thing indicator. If a unit saw another at 900m through trees that is quite amazing and in this game I would not expect that to happen repeatedly or for that occurrence to be a tell tale for how other units should spot. Maybe that unit had a brand new pair of binocs, or there position was just right to see through a small gap in the trees and they caught a muzzle flash. Maybe the units closer didn't have binocs, or their's were fogged up, dirty or otherwise not 100%, or maybe they didn't have any. Maybe the unit that did good spotting was a very experienced unit and maybe the other one wasn't, or maybe that had nothing to do with it. Maybe they were occupied more with their immediate situation, or they were communicating their next plan since they were in close proximity to enemy forces and not paying attention to what was going on further away. Maybe the terrain, (which is not level) somehow obscured their view from the position they were in, or maybe the snow was really deep where they were and they were dealing with that. Maybe they dug themselves a snow hole to keep their heads down from the mortar shrapnel falling to their rear and couldn't see out. Maybe maybe maybe maybe.... Having myself spent a significant portion of my life in forest, wilderness, mountains, very deep snow of all qualities, I can say with 100% certainty that sometimes dealing with your immediate situation is all encompassing and in the real world there are micro features and small terrain undulations that can not be modeled in this game, but would most certainly exist in the world that this game is supposed to model.

As for the unit that broke and ran toward the "enemy side of the map", I feel there are some details in the explanations that have been left out. Like most people, I have seen this undesirable behavior before in CM games, but I also believe that in this instance there are more factors in play. The American unit that broke and ran toward German lines had few or no better options. The American attack was coming along the extreme map edge, and attacks that come against the map edge like that put themselves in a position that severely limits their maneuverability and options and puts the force in a bad spot to begin with. So out of all possible directions, 180 degrees of space was not available to the unit in question. Out of the remaining 180, there was an exploding and burning tank behind them as well as mortar shells falling in the same direction a bit further behind the destroyed tank. So take away the 60 degrees of area that would have been directly toward the American lines. The next area, off the unit in questions left flank, are buildings where contact has been made between forces and potentially might give cover to more German units. So that isn't a good direction for the American unit to run. Take away another 60 degrees on their left. This leaves the remaining 60 degrees to their front, where nothing is exploding and no enemy troops are immediately obvious or in hiding, although it is toward the German "side" of the map the area to the units immediate front does not offer good cover to enemy. Due to the unit having such limited movement against the map edge, there is very little choice (especially good choice) in which to move. The unit broke and ran forward along the map edge, which is the only path that moved them away from exploding tanks and shells and not closer to an area of cover and concealment that one could easily assume housed enemy troops. Pretty much from what I have seen the unit was in a very bad spot, against the map edge, close to enemy occupied buildings, under fire, and unsupported. With that much going against them it would have been difficult to choose a better option, and definitely not the AI bugging out from what I see.

Now I'm not saying the retreat function can't be improved, it definitely can. I just think that in many instances in which people complain about any one of the many topics that get complained about, there are many factors that get overlooked or go unaccounted for. There are so many factors in this game that relying on any one, or complaining about any one to me seems to be losing the overall focus and tactical thinking that leads to success. I am not a programmer but I would think that maybe a way to deal with the retreating forward syndrome would be to have the default fall back option to be along the most recent path of advance that the unit took, since most likely the path of advance was along the most conducive and unexposed path available to that unit. I also really don't like it when an unbuttoned tank commander gets shot by small arms fire (happens very easily) and then the standard response is for the tank to pop smoke and retreat, often into a more exposed position than they were in. That is one aspect I feel could be improved. Tanks should not retreat in the face of small arms fire. (Just saying that because I know IanL has a direct line!)

When I have one of those "WTF?" moments, I try and sit back and think about all the possible things going on I can't see in the game. I also try to think about how I could have commanded my units better and not allowed my men to get in a bad position to begin with. Usually I come up with a list of how I could do better that is longer than how the game could be better. Other times it's just straight bad or good luck depending on your perspective and luck is also a real world phenomenon, but I have found looking closer at how I could command better and use better tactics gives me less of a chance to have bad luck, or have bad luck effect me in a significant way.

I know that the game being talked about is still in it's early stages and that both sides have significant forces that have not been engaged.

Edited by Will S
Link to comment
Share on other sites

Going back to the OP. With regards to some units being able to spot, whilst others nearby not being able to, that can depend on many factors. For a start, what kind of units are you talking about? For example, a HQ with a pair of binoculars or even two pairs in some cases, will have more chance of spotting than a fire team with none. Likewise a full squad will have far more pairs of eyes spotting, than a split off fire team. Without specifying what kind of "units" you are talking about you are making it very difficult for us to judge the situation.

Link to comment
Share on other sites

On 9/10/2019 at 5:10 AM, Warts 'n' all said:

The running around blindly behaviour should have been cured with the most recent patches on all titles, except the "bocage gap charging" in Battle for Normandy.

Could this "bocage gap charge" explain my recent noob problems? 

Playing CMBN4.02 battle called CW 18 Platoon, my troops sometimes rush crazy from behind bocage toward the house, and get chopped to pieces.

Edited by lsailer
include my comment as part of quote. oops.
Link to comment
Share on other sites

Back again...

If the "rattled, break cover, run towards enemey" incident I mentioned earlier (in passing) that occured in this PBEM was not a good example of the very same issue others are reporting, then this one that just happened in the same PBEM is a very good example:

https://streamable.com/ils2p

Infantry are located behind a hedge on an slightly elevated ridge (the ground gently sloopes down behind them in to an open field in defilade). They are directly facing the enemy map edge, with the friendly map edge behind them. No movement orders were given. The only orders were to target an enemy infantry unit to the front left.  Halfway through the infantry unit takes fire from the same front left direction. The immediate reaction of some of the men in the unit is to break cover and run directly forward through the hedgeline, toward.  The remianing men initially  stay behind the hedge and cower (probably safer), but eventually get up and run through the hedgeline like the original guys did.  They are just lucky that not even one of them was hit when they broke cover as they were fully exposed (well there is always next turn)

You can see a low stone wall about 15m in front of them but they literaly have to cross open terrain, running through enemy fire to get there. It is not clear if under this situation whether there behaviour was influenced by this "alternate" (more desirable?) nearby cover.

To be honest, there are actually two other cases I could show that have happened in this PBEM that share the same fundamental characteristic: "rattled" infantry in cover breaking cover and inexplicably running towards the enemy map edge (with more disasterous results). 

If the TacAI that kicks in controlling the behaviour of this "rattled" infantry unit is meant to be a "self-preservation" reaction to the enemy fire, perhaps the code is somehow not taking in to consideration the following information: Location, distance and safest route to nearest defilade and/or cover (in this case, defilade about 6 metres behind them, albeit in open terrain ). In the heirachy of possible directions a "rattled"  unit can move to "increase it's survivability", you surely would expect the "default"/all things equal/no-brainer go-to direction to be move in would be "away" from the enemy, or towards friendly lines.  There would need to be some pretty compeling circumstatnces at play if that "safer" direction/route/path led them closer to the enemy.

Maybe the TacAI coding is placing too much emphasis on the percieved cover at "the destination" and not considering the path/route/distance and the danger to the unit in getting there.

Regardless this recent new post-upgrade behaviour we are now seeing with infantry "running towards the enemy" appears to be "a thing" not previosly seen or considered a problem prior.  I hope it gets addressed.

Link to comment
Share on other sites

3 hours ago, Lt Bull said:

Back again...

If the "rattled, break cover, run towards enemey" incident I mentioned earlier (in passing) that occured in this PBEM was not a good example of the very same issue others are reporting, then this one that just happened in the same PBEM is a very good example:

https://streamable.com/ils2p

Infantry are located behind a hedge on an slightly elevated ridge (the ground gently sloopes down behind them in to an open field in defilade). They are directly facing the enemy map edge, with the friendly map edge behind them. No movement orders were given. The only orders were to target an enemy infantry unit to the front left.  Halfway through the infantry unit takes fire from the same front left direction. The immediate reaction of some of the men in the unit is to break cover and run directly forward through the hedgeline, toward.  The remianing men initially  stay behind the hedge and cower (probably safer), but eventually get up and run through the hedgeline like the original guys did.  They are just lucky that not even one of them was hit when they broke cover as they were fully exposed (well there is always next turn)

You can see a low stone wall about 15m in front of them but they literaly have to cross open terrain, running through enemy fire to get there. It is not clear if under this situation whether there behaviour was influenced by this "alternate" (more desirable?) nearby cover.

To be honest, there are actually two other cases I could show that have happened in this PBEM that share the same fundamental characteristic: "rattled" infantry in cover breaking cover and inexplicably running towards the enemy map edge (with more disasterous results). 

If the TacAI that kicks in controlling the behaviour of this "rattled" infantry unit is meant to be a "self-preservation" reaction to the enemy fire, perhaps the code is somehow not taking in to consideration the following information: Location, distance and safest route to nearest defilade and/or cover (in this case, defilade about 6 metres behind them, albeit in open terrain ). In the heirachy of possible directions a "rattled"  unit can move to "increase it's survivability", you surely would expect the "default"/all things equal/no-brainer go-to direction to be move in would be "away" from the enemy, or towards friendly lines.  There would need to be some pretty compeling circumstatnces at play if that "safer" direction/route/path led them closer to the enemy.

Maybe the TacAI coding is placing too much emphasis on the percieved cover at "the destination" and not considering the path/route/distance and the danger to the unit in getting there.

Regardless this recent new post-upgrade behaviour we are now seeing with infantry "running towards the enemy" appears to be "a thing" not previosly seen or considered a problem prior.  I hope it gets addressed.

judging from the teams situation in the video I´d think the low wall, also beeing somewhat more down in a depression, beeing the best "cover" vs. the incoming enemy smallarms and HE from WNW. Hedge is no cover, while likely the wall and depression breaks direct LOF from that (new) threat. From a SOP´s POV it´s likely the "best" choice to get out of trouble immediately. The teams stats are fairly good actually (+1 leader, regular, +1 morale). A minus leader, green and low morale team even might have done something that looks more stupid at first sight. I.e experienced, good morale infantry under Arty fire would bug out "forward" to get out of the danger area quickly, even when it takes losses. Low experience units likely would remain in the area fully pinned and taking even more serious losses this way. Whether CM TacAI works that way, I just can guess. In my own games and when in doubt about any my units threat appreciation I use "evade" command to check on this. The automated waypoint back and face command gives a good indication where worst trouble is to be expected in that particular game minute. Many the apparent unit "retreats" in fact are self applied "evade" commands which is not the same as true retreats from broken morale.

Edit: The bits of info found in the manual "The right button tells the unit to EVADE by abandoning its current Commands, seeking immediate cover and perhaps popping smoke. Although units can Evade on their own initiative, sometimes they try too hard to stick to their Commands and need to be redirected without further delay. Instant Commands work in both Real-Time and We-Go styles of play."

Edited by RockinHarry
Link to comment
Share on other sites

I haven't been able to hunt through old unmarked game saves, but did generate a new example. This is from an urban environment. The building height of the unit is only two levels, but a simiilar thing can happen with much higher buildings, including "double walled" buildings, where the unit only needs to retreat one action spot to the rear on the same level to escape enemy line of sight. Anyway, here are some detailed screens. The unit in question is green. The behavior seems to happen more often with lower quality units, but can happen with higher as well.

The setup. The top of the image is N, which is also the enemy map edge. Bottom is S, the friendly edge direction. The unit in question is marked with the arrow. It is in C2, with the HQ unit identified via a drawn red line. The HQ unit is in the rear, in safety inside a cleared block. The suicide unit is on the second level with a previously identified enemy unit across the road to its north, and two other identified enemy units directly to the east, with LOS down the alley to the road between the green unit and the northern enemy unit. The suicide unit can reach total safety by simply dropping one level to the windowless bottom of its current building section. Or, it can move to the lower level of the attached section to the east, which is also windowless. A third option would be to exit via one of the two doors in the attached section and move south in the direction of the HQ (and friendly map edge). Next image confirms window and door placement.

suicide_example_091819_1.thumb.jpg.588ea333408b99bdd045fea8096a6e8c.jpg

 

As you can see, two completely safe, accessible lower levels are as near as can be, along with two safe exits, assuming the unit moves south, towars the HQ/friendly map edge.

suicide_example_091819_2.thumb.jpg.d709ea21d54964c37773533ac58de9dd.jpg

 

However, as the unit panics and displaces, it opts not to use the two immediate safe locations out of enemy LOS and exits out the less safe, but still potentially safe side exit. However, it opts to go north into known danger, rather than south to the safety of the HQ/friendly map edge.

suicide_example_091819_3.thumb.jpg.ca8d9b631831618e53d07a2ba5793cbe.jpg

 

The unit has now placed itself into the LOS of both known enemy threats.

suicide_example_091819_4.thumb.jpg.b1d215cdfa24feae0f48714116ddfd73.jpg

 

First, the eastern alley shooters take out a man.

suicide_example_091819_5.thumb.jpg.b90b59463769fcd335621427484f9f45.jpg

 

Then, the northern threat tosses a grenade as they pass by.

suicide_example_091819_6.thumb.jpg.0b434057c53645bf658886290fc1f6a1.jpg

 

Two more men are lost.

suicide_example_091819_7.thumb.jpg.a50e591cc10688aa4eab0e39b74e4d92.jpg

 

Finally, the unit enters a door to safety after losing a number of men. The losses were completely unneccessary and unrealistic. In RL, green units would almost certainly run downstairs and hide, or bolt back to their HQ.

suicide_example_091819_8.thumb.jpg.e870534397c0a3133cb1be69b8fbf2fe.jpg

 

While this behavior doesn't happen every time, it does happen fairly frequently. When it does, no amount of available cover seems to help.

Hope this is useful info.

Link to comment
Share on other sites

15 minutes ago, Macisle said:

I haven't been able to hunt through old unmarked game saves, but did generate a new example. This is from an urban environment. The building height of the unit is only two levels, but a simiilar thing can happen with much higher buildings, including "double walled" buildings, where the unit only needs to retreat one action spot to the rear on the same level to escape enemy line of sight. Anyway, here are some detailed screens. The unit in question is green. The behavior seems to happen more often with lower quality units, but can happen with higher as well.

The setup. The top of the image is N, which is also the enemy map edge. Bottom is S, the friendly edge direction. The unit in question is marked with the arrow. It is in C2, with the HQ unit identified via a drawn red line. The HQ unit is in the rear, in safety inside a cleared block. The suicide unit is on the second level with a previously identified enemy unit across the road to its north, and two other identified enemy units directly to the east, with LOS down the alley to the road between the green unit and the northern enemy unit. The suicide unit can reach total safety by simply dropping one level to the windowless bottom of its current building section. Or, it can move to the lower level of the attached section to the east, which is also windowless. A third option would be to exit via one of the two doors in the attached section and move south in the direction of the HQ (and friendly map edge). Next image confirms window and door placement.

suicide_example_091819_1.thumb.jpg.588ea333408b99bdd045fea8096a6e8c.jpg

 

As you can see, two completely safe, accessible lower levels are as near as can be, along with two safe exits, assuming the unit moves south, towars the HQ/friendly map edge.

suicide_example_091819_2.thumb.jpg.d709ea21d54964c37773533ac58de9dd.jpg

 

However, as the unit panics and displaces, it opts not to use the two immediate safe locations out of enemy LOS and exits out the less safe, but still potentially safe side exit. However, it opts to go north into known danger, rather than south to the safety of the HQ/friendly map edge.

suicide_example_091819_3.thumb.jpg.ca8d9b631831618e53d07a2ba5793cbe.jpg

 

The unit has now placed itself into the LOS of both known enemy threats.

suicide_example_091819_4.thumb.jpg.b1d215cdfa24feae0f48714116ddfd73.jpg

 

First, the eastern alley shooters take out a man.

suicide_example_091819_5.thumb.jpg.b90b59463769fcd335621427484f9f45.jpg

 

Then, the northern threat tosses a grenade as they pass by.

suicide_example_091819_6.thumb.jpg.0b434057c53645bf658886290fc1f6a1.jpg

 

Two more men are lost.

suicide_example_091819_7.thumb.jpg.a50e591cc10688aa4eab0e39b74e4d92.jpg

 

Finally, the unit enters a door to safety after losing a number of men. The losses were completely unneccessary and unrealistic. In RL, green units would almost certainly run downstairs and hide, or bolt back to their HQ.

suicide_example_091819_8.thumb.jpg.e870534397c0a3133cb1be69b8fbf2fe.jpg

 

While this behavior doesn't happen every time, it does happen fairly frequently. When it does, no amount of available cover seems to help.

Hope this is useful info.

To me it looks like a pathing issue from and through buildings. For a reason that retreating (or evading?) unit wants going to a certain place (another building?) likely offering better cover towards the most apparent threat (enemy in building north?). Since the TacAI almost always choose quick or fast move mode it figures path to that building takes less time when moving outside, instead of finding a viable path through inside buildings. Think this is a known (severe) limitation of path finding in/around buildings and as well causes me lots of WTF moments in my own games frequently. Surely something that BFC should think about improving. For movement purposes the TacAI prefers fast moving terrain before slow moving (maybe providing better cover) terrain all the time, unless it´s a final waypoint (into slow move cover terrain). Then there´s "Bottlenecks" (doors) and the TacAI computes paths for pixeltroopers individually. If it figures it takes too long or the bottleneck is blocked during a particular time chunk, it diverts single pixeltroopers another way to the units target AS. This is where most the "single pixeltrooper seperating from unit" issues come from IMO. Also could be the TacAI finds it can´t push the whole unit through that bottleneck fast enough, it chooses an alternative path for all pixeltroopers instead. 

Buildings in CM aren´t that strong (offering good cover) than one might think. Walls aren´t impenetrable to small arms (another shortcoming), particularly at closest range. Even if the building concerned has no windows at the basement, effective small arms fire still can pour in through windows upper story, then penetrating through the ceiling. Looks like ceilings are considered nothing more than thin wooden planks. Same in reverse. You can effectively hit & kill units in a buildings second story by pouring bullets from outside a building, through windows and up the ceiling. Staircases are fairly large dimensioned (4x4m I think) as well, offering another weak point for in building fights.

Another peculiarity of buildings is they appear having a dedicated "front door", or "front wall" (set in map editor). This one appears prefered for entering or exiting a building, even if preset window or door configurations are changed (or removed) in editor 3D view. Also changing door/window configurations oftenly lead to unpredictable effects in game, like invisible doors or odd unit movements, in and out from that building.

Link to comment
Share on other sites

3 hours ago, RockinHarry said:

judging from the teams situation in the video I´d think the low wall, also beeing somewhat more down in a depression, beeing the best "cover" vs. the incoming enemy smallarms and HE from WNW. Hedge is no cover, while likely the wall and depression breaks direct LOF from that (new) threat. From a SOP´s POV it´s likely the "best" choice to get out of trouble immediately. The teams stats are fairly good actually (+1 leader, regular, +1 morale). A minus leader, green and low morale team even might have done something that looks more stupid at first sight. I.e experienced, good morale infantry under Arty fire would bug out "forward" to get out of the danger area quickly, even when it takes losses. Low experience units likely would remain in the area fully pinned and taking even more serious losses this way. Whether CM TacAI works that way, I just can guess. In my own games and when in doubt about any my units threat appreciation I use "evade" command to check on this. The automated waypoint back and face command gives a good indication where worst trouble is to be expected in that particular game minute. Many the apparent unit "retreats" in fact are self applied "evade" commands which is not the same as true retreats from broken morale.

Edit: The bits of info found in the manual "The right button tells the unit to EVADE by abandoning its current Commands, seeking immediate cover and perhaps popping smoke. Although units can Evade on their own initiative, sometimes they try too hard to stick to their Commands and need to be redirected without further delay. Instant Commands work in both Real-Time and We-Go styles of play."

The unit is not under arty attack.  There is arty (not major calibre) falling to the units rear right about 150m away, at which range it typically does not affect units.  The direct automatic weapon fire (flak gun incl) that it is reacting to is coming in at about 45deg front left.

From the side angle view screenshots below of the same action in the video (you can see the incoming tracer fire in each of the BEFORE, DURING and AFTER the unit makes it's move), you can clearly see the terrain directly behind the unit it is more depressed (offering 100% defilade from all direct enemy fire) than the terrain in front of it. The friendly lines (safety good!) are to the left, the enemy lines (danger bad!) are to the right.

1.jpg

2.jpg

3.jpg

The infantry unit started the turn targeting some enemy infantry running away from it, but later started shooting  at a open sided 20mm or 37mm HT mounted AA vehicle that was moving laterally to the left.  It was the return fire from this AA vehicle that "rattled" the infantry unit halfway through the turn. By the time the infantry had reached the wall in front of it, it appears the AA unit had moved out of LOS anyway.  Despite any cover afforded by the wall, the infantry unit is still within enemy LOS and is now taking small arms fire from a new more flanking position from the left.

How many would disagree that, given the circumstances and threat the infantry unit was reacting to, the safest thing for it to have done was to simply follow what most would consider the most basic instinct to threats: retreat AWAY from the direction of the incoming enemy fire/location of known enemy units/enemy lines (map edge) so as to reduce/diminish, or better, break LOS completely to the enemy and hence gain defilade?  This would be immediately achieved if the infantry unit moved back 6m towards it's own lines.

Few would have reason to raise a discussion like this if the infantry unit just retreated in the basic vanilla manner described as many would expect.  Instead, it seems too frequently for it not to be noticeable in a random game, infantry units under fire are taking what can be described as more questionable "exotic" solutions that are meant to increase their safety that involve running towards the enemy fire/enemy locations across open ground, when a simple basic vanilla "just retreat away from the enemy" reaction would have sufficed.

K.I.S.S!

Not sure if it was ever brought up in the past (I am sure it was/has) but I remember questions were raised about the TacAI "self preserving" behaviour when infantry took fire typically when moving across open ground from cover to cover.  I think people called it the "crawl of death" (may have even been in CMBO/AK).  Many of the situations were like the "man who swims 80% distance across the river, then decides the remaining 20% is too far and swims back" kind of logic at play. There were cases where you would expect the infantry unit under fire to either crawl or run to the nearest cover regardless of the relative direction to friendly/enemy lines were and/or in which direction the enemy fire was coming from.  Instead, rather than continue a few more metres "to safety", they backtracked across more open ground than they would of if they had just headed towards the closest cover.

I understand coding an algorithm that intelligently handles all situations is not easy. I hope a tweak can fix what we are seeing happen too often.

Edited by Lt Bull
Link to comment
Share on other sites

24 minutes ago, Lt Bull said:

The unit is not under arty attack.  There is arty (not major calibre) falling to the units rear right about 150m away, at which range it typically does not affect units.  The direct automatic weapon fire (flak gun incl) that it is reacting to is coming in at about 45deg front left.

From the side angle view screenshots below of the same action in the video (you can see the incoming tracer fire in each of the BEFORE, DURING and AFTER the unit makes it's move), you can clearly see the terrain directly behind the unit it is more depressed (offering 100% defilade from all direct enemy fire) than the terrain in front of it. The friendly lines (safety good!) are to the left, the enemy lines (danger bad!) are to the right.

1.jpg

2.jpg

3.jpg

The infantry unit started the turn targeting some enemy infantry running away from it, but later started shooting  at a open sided 20mm or 37mm HT mounted AA vehicle that was moving laterally to the left.  It was the return fire from this AA vehicle that "rattled" the infantry unit halfway through the turn. By the time the infantry had reached the wall in front of it, it appears the AA unit had moved out of LOS anyway.  Despite any cover afforded by the wall, the infantry unit is still within enemy LOS and is now taking small arms fire from a new more flanking position from the left.

How many would disagree that, given the circumstances and threat the infantry unit was reacting to, the safest thing for it to have done was to simply follow what most would consider the most basic instinct to threats: retreat AWAY from the direction of the incoming enemy fire/location of known enemy units/enemy lines (map edge) so as to reduce/diminish, or better, break LOS completely to the enemy and hence gain defilade?  This would be immediately achieved if the infantry unit moved back 6m towards it's own lines.

Few would have reason to raise a discussion like this if the infantry unit just retreated in the basic vanilla manner described as many would expect.  Instead, it seems too frequently for it not to be noticeable in a random game, infantry units under fire are taking what can be described as more questionable "exotic" solutions that are meant to increase their safety that involve running towards the enemy fire/enemy locations across open ground, when a simple basic vanilla "just retreat away from the enemy" reaction would have sufficed.

K.I.S.S!

Not sure if it was ever brought up in the past (I am sure it was/has) but I remember questions were raised about the TacAI "self preserving" behaviour when infantry took fire typically when moving across open ground from cover to cover.  I think people called it the "crawl of death" (may have even been in CMBO/AK).  Many of the situations were like the "man who swims 80% distance across the river, then decides the remaining 20% is too far and swims back" kind of logic at play. There were cases where you would expect the infantry unit under fire to either crawl or run to the nearest cover regardless of the relative direction to friendly/enemy lines were and/or in which direction the enemy fire was coming from.

thanks for clearing up some more. My only guess remains that "cover" values are treated as absolutes, so a "wall/depression" (forward) likely counts more than "depression" (backward). Generally the TacAI search and select cover routines are bits of a mistery. Similar things happen when pixeltroopers leave foxholes to take cover in a tiny shellhole 1m away. :P At last just one of many issues in current TacAI decision cycles, when it comes to cover and move path finding.:mellow:

Link to comment
Share on other sites

6 hours ago, Lt Bull said:

Back again...

If the "rattled, break cover, run towards enemey" incident I mentioned earlier (in passing) that occured in this PBEM was not a good example of the very same issue others are reporting, then this one that just happened in the same PBEM is a very good example:

https://streamable.com/ils2p

That sure looks like a prefect example of the bug. I have questions:

Do you have a save? A .ema turn file would be good.

What scenario or QB map is this (would like to try to make it happen again)?

Did you start this game using 4.02 version of the game?

Please PM me and we can discuss getting the saved turn.

PS I wrote this upon reading your first post about this and will not finish reading the rest of this thread - apologies if you have already answered any of my questions.

Link to comment
Share on other sites

4 hours ago, Macisle said:

Finally, the unit enters a door to safety after losing a number of men. The losses were completely unneccessary and unrealistic. In RL, green units would almost certainly run downstairs and hide, or bolt back to their HQ.

suicide_example_091819_8.thumb.jpg.e870534397c0a3133cb1be69b8fbf2fe.jpg

 

While this behavior doesn't happen every time, it does happen fairly frequently. When it does, no amount of available cover seems to help.

Hope this is useful info.

That's a very good post.....But I'm not 100% sure that it doesn't reflect the behaviour of men panicking, it's annoying as hell when it happens, but as I said, it does seem plausible once the unit has 'freaked out'.  :unsure:

Link to comment
Share on other sites

6 hours ago, Sgt.Squarehead said:

That's a very good post.....But I'm not 100% sure that it doesn't reflect the behaviour of men panicking, it's annoying as hell when it happens, but as I said, it does seem plausible once the unit has 'freaked out'.  :unsure:

think there´s bits of distinction between actual panicking and the TacAI applying the evade command. When the unit concerned is still on the move one can check assigned waypoint and face, indicating it was an "evade". I remain with seeing the main issue in path finding multiple guys through restricted terrain, like buildings and bottlenecks (doors, gapped bocage etc) generally. Speed before cover, so to say.

Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Unfortunately, your content contains terms that we do not allow. Please edit your content to remove the highlighted words below.
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...