Jump to content

4.0 Upgrade Looks Fantastic --Thanks, BF!


Macisle

Recommended Posts

10 hours ago, JoMc67 said:

This is Good to hear...and I for one was hoping for something like this to happen in CM, and it appears 4.0 is a step in the right direction.                                                                    

Now, it seems at least Firefights, in your opinion, are a little more realistic with reduced casualties due to shorter Firefights, because TAC AI is making better use of cover and advance/retreat routes...I wonder thou, do you think the TAC AI will attempt to get to the closest cover, wither it forward or reverse..? ex, a Team is advancing towards a House 5 meters to the front of them, but started to receive Small Arms fire...Do you think that Team will continue to advance to that House (closest cover), or Fall back to a House in the rear that's 15 meters away (thou, I would imagine it depends on Moral, Motivation, Suppression, etc).

This New TAC AI Logic may also make players re-think their Tactics some what when playing against other Human Opponents.   

Joe

It's hard to say on better use of cover. As there are no new animations and we cannot see the interior layout of buildings, we have to accept a certain level of abstraction in terms of what our eyes see the soldiers doing at the time that bullets are hitting or missing. I think most of what is going on is a reduction in small arms accuracy in combination with more dispersed soldier spacing, interacting with a tweaking of suppression levels and morale-based behaviors like rout/evade.

My AI Plan attempts to maximize approach paths for safety, but it was made under the pre-Upgrade limit of 16 Orders (Area fire has been added,though). I could add new orders to further try to curb AI carelessness (like "leading" them to a building side with one Order and then putting the next Order inside the building), but I'm not sure if it is worth it. The AI is fighting much better than before and that could end up putting AI troops in bad spots as they hang out in the street longer when there is no way to know where the player might have troops placed to fire on them. I think it would be a good idea for me to let others playtest with the current moves as a baseline to see if I should add more movement Orders to help the AI keep casualties down.

The AI troopers in the firefights I'm mentioning are operating under Active stance, which means they will stop and shoot more readily than under Normal stance. IIRC, they are also usually using Assault movement, as opposed to Advance. On the question of how they respond to incoming fire vis-a-vis trying to close the distance to cover nearer the enemy, I can only give limited impressions, since the code surely has curveballs up its sleeve and it's still early days. I've seen AI troopers who encounter fire not too far from their start point fall back, try again, rinse and repeat a few times. Others who are more than halfway to their destination have continued on. And, the same group that fell back a few times eventually started crawling across while their buddies in the building at their start point provided covering fire that achieved fire superiority over my guys attempting to keep them at bay.

Here is a screen of the first crossing attempt by an HQ unit and squad. The small cloud of smoke is from my LMG, with a rifle team also firing from another buidling just behind and out of the picture. The transparent building in front only has a casualty in it, with a Soviet tank Area firing on it.

31313748424_53586235eb_b.jpg

 

The Soviets turn back quickly and are able to fall back with no casualites.

31313748324_77aabece3c_b.jpg

 

Again, they tried this (sometimes only the HG unit, I think) a few times, eventually making it by crawling under overwatch with fire superiority.

One other thing to note: despite much lower casualties for the AI, (aside from one very dumb decision that cost me half a platoon, my casualty rate is about the same), the scenario is playing faster than pre-Upgrade. I've had to fall back faster and am running out of real estate while waiting on my reinforcements--who may not arrive in time.

I'll post more pics soon.

Edited by Macisle
Link to comment
Share on other sites

  • Replies 156
  • Created
  • Last Reply

Top Posters In This Topic

Here are two shots of a firefight that has been going on for 5 minutes and counting. This was earlier on. So far, I've lost 2-3 men and the Soviets have lost about twice as many, with a few attempts to push through the courtyard towards me being repulsed. Some teams on both sides have fallen back and been sent back in after recovering their nerve.

32006839242_cf8a724f10_b.jpg

 

Closer up. Lots of SMG ammo has been burned up and some are running low now.

32006839112_0fecd7c096_b.jpg

 

The AI also attempted to flank along the E-W road on the southern side of the Soviet-held building block. I have a veteran HMG team covering that and it quickly wiped out the first squad that attempted it. Here are pics of the second attempt:

31780672580_4cfa48e40d_b.jpg

 


My HMG had to change barrels as the Soviets started their run and stopped firing for a number of seconds, allowing the enemy to cover some ground.  However, I had six riflemen and 2 SMGs also covering this and the Soviets did a quick about face.

31780672490_34434f4117_b.jpg

 

No casualties were taken until my HMG came back online. It inflicted 3 casualties with the remaining Soviets running back into a building (I think the same place they started from).

31780672350_b200850bfe_b.jpg

Edited by Macisle
Link to comment
Share on other sites

Okay, here are some rout shots. First, two showing that units don't always go for the front of buildings/independent buildings.

Sorry about the yellow, I was going to show the team status, but Fraps doesn't seem to capture the stress level graphic correctly, so I went ahead and cropped the shot. This forward team comes under fire and the turn ends with an evade dash waypoint plotted by the TacAI. I change the order to a quick move across the park to the team's rear, ending in one of the buildings in the block on the other side.

32117492696_73e5641fde_b.jpg

 

However, the TacAI ignores my order and has the team run into a commercial building along the way I wanted it to go. The street in front of the building was an enemy fire lane, so I was very thankful that the TacAI chose the rear entrance. I got the team to the destination I originally wanted the next turn.

32037478971_498da0ed0b_b.jpg

Edited by Macisle
Link to comment
Share on other sites

Here is a more troubling, potentially-suicidal rout example. This was from a turn that was saved pre-Upgrade, but processed post-Upgrade. All the other examples and pics are from my current playtest, which is a fresh playtest under 4.0.

First, some context: my German platoon recently fell back a block to the south, to its current position. Prior to falling back, they were in the L shape of buildings immediately to the north of where they are in the pic below. You can clearly see directions of threats, as well as the protected inner courtyard of their block, which is being kept further warm and cozy by the HQ unit there.

32155976415_5ea742789c_c.jpg

 

The SE team takes intense fire from the SE and goes into rout mode as a result.

32155976255_f799a1b362_b.jpg

 

As you can see, the rout path is far from ideal. Ignoring the obvious safe zone just to the rear of their building (the one where both the Plt and Co. HQs are waiting with cigarettes and open arms), they decide to run across the street to the north, potentially exposing themselves to fire during movement and placing them right back in the block that the platoon just fell back from. This is the kind of thing that is hard to rationalize, even from a panicked unit.

32155976095_ebd4798935_b.jpg

Edited by Macisle
Link to comment
Share on other sites

Here's another example from my fresh playtest. This one is easier to rationalize than the last, but still seems a bit counter-intuitive to trained soldiers and self-preservation. The German platoon is falling back under the cover of two lines of smoke, as a Soviet Maxim has a fire line down the street they must cross. Things go well and no casualties are taken by the initial group of teams. However, the HQ, who stayed behind to place smoke, lags behind.

32156301115_6d1ba8296d_b.jpg

 

The HQ team's destination is the closest door, but just as they approach it, the Maxim sends a burst past their ears and they become flustered.

31781312580_79400cdc7b_b.jpg

 

I would expect soldiers in a similar situation to dash right into the door, as fast as they could. However, with the smoke starting to clear, the HQ team decides to keep running down the street covered by the fire lane to a door further away. The second door down offers escape as well, and is out of the LOS of Soviets flanking the street from the west. But, the HQ team decides to keep running to the third door down.

32156300835_ae313e6edd_b.jpg

 

At which point, the smoke clears enough for the Maxim to get a bead on them and drop a member of the team (thankfully not the Platoon leader, but the team's morale drops to broken).

31781313070_1105215d61_b.jpg

Link to comment
Share on other sites

Here's what lead up to the desperate, smoke-covered fallback in the previous post. I was going to try and hold part of a strongpoint, ambushing any Soviets who came through the center. Although it has been stated that the new Peeking Around Corners behavior does not apply to wall openings, I thought I'd try to use them anyway, as I've had some recent luck with them. However, like a complete idgit, I forgot about my exposure to enemy units moving up to the second floor of buildings with a view of the courtyard. Note the enemy positions and threat directions.

31781774560_792c285efb_b.jpg

 

Needless to say, I was a sitting duck. The northernmost team took four casualties and the remaining man ran down the L to the south, turned W and ran out the wall opening at the bend right into the Soviet fire (you can still see the cross for him). Then the next team took a casualty and ran out the wall opening it was covering, again right into Soviet fire, getting wiped out.

31781774490_c22a3cfef7_b.jpg

 

At which point, the LMG team ran out the building heading E, getting mowed down in the street, while the HQ team ran S (a smart move), and the remaining team, rather than running into and out of the building to their S, followed the Blaze of Glory exit paths of the other teams, heading N out the wall opening right into Ivan's loving arms and death.

31781774450_e3c2c75f7a_b.jpg

 

However, the HQ team still didn't take the safest route, which would have been the closest door right across the street. They decided to risk more and longer exposure to go two doors down the street around the corner.

31315375824_bf8ebd6287_b.jpg

Edited by Macisle
Link to comment
Share on other sites

Here are two shots of a vehicle rout. The Opel took a burst of fire on its way to the building to pick up an HMG team and the Co. HQ. At turn end, there was no damage to the Opel and it's morale seemed okay, too. So, I gave it a face command order (in place, no movement) and told the infantry to run beside it and then mount up.

32118901906_884318c5c0_b.jpg

 

However, when the turn ran, the Opel decided to drive around the corner to the S and the infantry followed it. The turned ended as below. Luckily they got away, but it could have been bad.

31346930203_bb7d0d732a_b.jpg

 

Alright, that's it for the screenies. These are food for thought and conversation. Like I said, I'm not seeing too many howlers as I put in more hours. Usually, what happens is reasonable, and sometimes, the rout path is quite good. It's only occasionally that something happens that is very hard to rationalize.

Edited by Macisle
Link to comment
Share on other sites

Nice frame action report and really food for thought. Could you relate any the rout paths to the recently mentioned "front side" alignment of buildings? From my playtests in CMBN at least I could encourage some refugees to use the back door/backside of a building more if it´s also set up with the front side facing away from the enemy frontline.

Routing units generally appear to choose a rout waypoint, that lead them away from the threat that caused the route, while considering a minimum standoff range of 3 or more AS. The final blue face command line, gives at least an indication, from what threat the unit is routing from. Though unfortunately not considering any other threats, either known or unknown along the rout path. Non panicked/shaken units do appear choose for more clever evade paths and appear to be shorter generally. In multi threat environments it´s as good as impossible to predict likely retrograde moves, so it´s usually better to anticipate the right moment to back off from possibly untenable positions, before the TacAI takes the matters out of ones hand.

I also tend to change AI generated evade waypoints, when there´s some tactical usefulness to it. If the majority of pixeltroopers is still on the way, drawing the waypoint to better cover terrain is one option. Deleting it entirely and replotting it could lead to loss of time, as well as making the situation even worse than just letting the TacAI do. If there´s just few leftover stragglers on the way to the evade WP, I sometimes change order to crawl when there´s cover terrain, or breaks LOS to the enemy midways.

Would be interesting if one could find some rules of thump for evade moves, when they happen and where they lead to, considering all the soft factors, terrain and enemy threats. Think I´ll toy around some more with the evade move as it most likely applies the same logic as when used by the TacAI on friendfly and AI player side.

So far I have no major problems with V4.0 and the enhancements are throughout positive I find. One just needs to get used to battles played in V3.x now play somewhat differently, particularly when opposing an AI player. There´s still lots in the hands of the scenario maker when creating plans to make for a good AI opponent, but V4.0 is a big step forward in this regard. So thumps up and 10$ (or 9,87 Euro) well spent. :)

Link to comment
Share on other sites

Very interesting read. I actually didn't notice significant changes to behavior. Apparently I was not watching close enough. I was attributing the changes I saw to the more staggered and spread out stance of teams. Less total wipe out of teams so more guys still left to retreat.

One question though how sure are we that teams favour front doors over other doors? I have a few questions in this regard. Is there a statical difference that shows they favour the front door? In all cases or only certain geometric configurations? And is it significantly different from the v3 engine?

Link to comment
Share on other sites

7 hours ago, IanL said:

...

One question though how sure are we that teams favour front doors over other doors? I have a few questions in this regard. Is there a statical difference that shows they favour the front door? In all cases or only certain geometric configurations? And is it significantly different from the v3 engine?

That´s the big question. I´m still testing on my urban map where I know building configurations and where the interesting actions take place (routing, moving from particular buildings). After turning around some buildings, I noticed some change of behavior, or at least some pronounced tendency of routing units taking the back door into a neighboring building, instead of what formerly was the front door (at what I think the game considers to be the actual front side), leading straignt to the main street killing zone. The buildings look still the same, as I matched the walls and doors of those buildings, as they were before. The buildings were formerly placed side by side with front sides towards the street, as one usually does during city block construction in the map editor. The variations had them turned around 90°, one clock wise the other cc wise, so that the front sides face each other, building the internal walls and connected with doors.

I just tested this isolated single case and particularly watching for changed routing paths (toward the waypoint set by the Tac AI). If more cases can be created for a routing unit prefering different rout paths through a particular door, by changing building alignments, then we might have something interesting to think about. For now it´s just an interesting observation, that needs confirmation by other players or BFC.

A related and even more interesting question would be the decision of the TacAI to select a particular AS for its rout or evade waypoint. Thus far I had the evade (arrow down) button paid no attention, but wow.... I just found it to be an as valuable tool as the targeting command for determining LOS/LOF. Pushing the button repeatedly (and waypoint canceled) gives a pretty accurate prediction of where a unit will be moved to by the TacAI, when a rout or evade occurs during following turn execution phase. So basically it´s no different for a player using the evade button or the TacAI. Majority of cases it´s the same AS chosen as destination and same follow up face command. That off course could change during follow up 60 seconds, when new enemy threats might appear, but the evade destination AS gives an idea if it´s time to relocate a particular unit, before the TacAI does.

Evade destination AS. Obviously an AS that gets the threatened unit out of LOS and LOF of the currently biggest threat. Farer away vs an armor threat than soft/infantry threat it seems. I see no indication that location of formation HQ plays a role or friendly/enemy map edge, although for the latter I´m not sure. Evade (tactical retreat) usually is shorter than a rout. The TacAI sometimes changes evade/rout destination mid move if a morale change takes place, ie from rattled to worse or better. Needs more testing, but hitting evade button (and canceling) during orders phase gives an idea what might happen in the very near future.

Link to comment
Share on other sites

Good information on Evade, RH! I used to use it a lot after Normandy came out (is it in CMSF? That's what I cut my CM teeth on. I think I used it there a lot, too), but haven't bothered with it for many years.  I used to think I knew better than the TacAI, or at least knew how I preferred to retreat. I'm going to rethink all that. 

Link to comment
Share on other sites

Does this upgrade do anything with the game performance? I found out about Combat Mission recently and downloaded the Red Thunder demo and it seems really nice - but the performance is just so horrible. Been reading about it here and that just seems to be so, no matter what PC you are running it with.

Link to comment
Share on other sites

First, no there are no significant changes to performance.

Second, the vast majority of people do *not* have performance problems. Yes, there are some machine configurations and graphics cards that have issues. Here are a few threads to read to see if their suggestions help you (if you get the demo to work satisfactorily then the full game will too)

 

Link to comment
Share on other sites

I think the new evade AI has made the game easier for scenarios where you have to attack the defending cpu (which is most scenarios). If I know enemy troops are in buildings or foxholes I call in a medium strength artillery barrage and put troops on overwatch of the area. In my experience with 4.0 the defenders will now leave their fortified positions and run out into the streets to be taken out by the artiy or picked off by my troops in overwatch. Before 4.0 you know you'd have to go into the town and weed out the stubborn survivers.

Not saying this is necessarily a bad thing but it does change a lot of existing scenarios difficulty.

Link to comment
Share on other sites

1 hour ago, bodkin said:

I think the new evade AI has made the game easier for scenarios where you have to attack the defending cpu (which is most scenarios). If I know enemy troops are in buildings or foxholes I call in a medium strength artillery barrage and put troops on overwatch of the area. In my experience with 4.0 the defenders will now leave their fortified positions and run out into the streets to be taken out by the artiy or picked off by my troops in overwatch. Before 4.0 you know you'd have to go into the town and weed out the stubborn survivers.

Not saying this is necessarily a bad thing but it does change a lot of existing scenarios difficulty.

Just Wonderful !...This is something I'm not looking forward to, and for the most part agree with akd. 

Edited by JoMc67
Link to comment
Share on other sites

Thanks for the detailed analysis, RockinHarry. I follow a similar SOP regarding the altering/cancelling of evade/rout paths. In recent months, I have also been using the Evade button to see where the TacAI might go, but only occasionally, not systematically.

I can't give any new information on the issue of building facing and how it impacts rout paths. In recent turns of my playtest, I've had several units rout intelligently within their blocks. On the other hand, in my last two turns, I had a unit that had been given a safe, multi-WP fallback path take fire and choose the worst path possible, being wiped out in the street. I may post pics of that later. It's probably the worst instance so far among the pics I've posted.

While I'm loving 4.0, there is a creeping feeling that the Attacker's job, which I believe general opinion thinks is somewhat easier in CM than in real life, has gotten still easier under 4.0. Since the first half of the scenario I'm testing (SP only) is an intense, defensive situation, it is a useful tool for analysis here. I'd say the scenario is playing about 10-15 minutes faster than pre-4.0. Meaning, I'm having to give up positions that I would have held onto longer before because I am following the line of thinking that RockinHarry mentioned--falling back before the TacAI does it for you. (Edit: tank fire under the new AI Area Fire feature is a significant part of this as well, since I can't use some key positions as effectively as before--or at all).

Using that technique, I can avoid excessive casualties, as along as unexpected fire doesn't throw a team into suicidal rout/evade and I have real estate available to fall back to (currently running out). However, this still means that a Defender may not be able to defend key terrain as well as his real life counterpart can. And, it also may mean that the defensive AI is now at a greater disadvantage because it lacks the intelligence to shift units to plug gaps and restore a defense as routing/evading units recover.

I'm also wondering if attacking units might be a bit too brave now. While their not taking casualties as quickly as before is a good thing, I'm noticing some fairly impressive ability of units to shrug off defensive fire when crossing streets, etc. That may just be a product of my feeling under the gun in my defense since the cavalry is probably going to be too late now. But maybe not. I could grab some screenies of that, as the last turn offered two examples.

It may be a day or two, though.

Edited by Macisle
Link to comment
Share on other sites

Another quick post to say that, having cooled down from a bad turn, it's too early and there are too many variables at play to make a judgement on problems with Defense vs. Attack balance yet. This is, after all, only one playtest and my troops are at the mercy of the decisions I'm making, which may not always be the best.

Link to comment
Share on other sites

I think the problems with the AI trying to fall back and running into the street is due to the pathfinding being unable to plot a path through buildings.

You can see the same happening if you have a squad in front of a long block of buildings, and you plot a single movement order to the street on the other side of the block. The guys will not go through the buildings, they will run around - even if it's a long detour.

So basically the evade AI is trying to do the same thing in your example with the courtyard and walls, @Macisle.

Link to comment
Share on other sites

13 hours ago, Macisle said:

While I'm loving 4.0, there is a creeping feeling that the Attacker's job, which I believe general opinion thinks is somewhat easier in CM than in real life, has gotten still easier under 4.0. Since the first half of the scenario I'm testing (SP only) is an intense, defensive situation, it is a useful tool for analysis here. I'd say the scenario is playing about 10-15 minutes faster than pre-4.0. Meaning, I'm having to give up positions that I would have held onto longer before because I am following the line of thinking that RockinHarry mentioned--falling back before the TacAI does it for you. (Edit: tank fire under the new AI Area Fire feature is a significant part of this as well, since I can't use some key positions as effectively as before--or at all).

Yeah, I also feel in CM the Attacker (with only a 3 to 2 Odds, let alone the more realistic 2 to 1 needed in RL) has an easier job against Defender compared to RL, and now seems it's going to get a little easier w/v4.

If you think it's bad now, just wait until you play the game where the Attacker lays down a Barrage on your troops in Buildings/Foxholes/Works (doesn't matter) just to watch them skedaddle loosing Real Estate & more Casualties (instead of suppressing, and taking low casualties pre v4).

Yeap, CM already plays out relatively quick, and v4 is going to make that a little quicker...I always envisioned that 1 minute in CM is equal to like 2.5 to 5 minutes in RL.    

Edited by JoMc67
Link to comment
Share on other sites

2 hours ago, JoMc67 said:

Yeah, I also feel in CM the Attacker (with only a 3 to 2 Odds, let alone the more realistic 2 to 1 needed in RL) has an easier job against Defender compared to RL, and now seems it's going to get a little easier w/v4.

If you think it's bad now, just wait until you play the game where the Attacker lays down a Barrage on your troops in Buildings/Foxholes/Works (doesn't matter) just to watch them skedaddle loosing Real Estate & more Casualties (instead of suppressing, and taking low casualties pre v4).

Yeap, CM already plays out relatively quick, and v4 is going to make that a little quicker...I always envisioned that 1 minute in CM is equal to like 2.5 to 5 minutes in RL.    

And, the above is only against the AI. Just think how quick and how it will play out against a competent Human Opponent...The Defender barely had a chance pre v4...and now :-/

Link to comment
Share on other sites

The effects against a human player will be very different, especially in regards to artillery.  Human player will proactively hide or displace troops before effective fire is delivered, which means they don't take casualties as easily, which means the TacAI HE evade behavior is not triggered.

Link to comment
Share on other sites

8 hours ago, Bulletpoint said:

I think the problems with the AI trying to fall back and running into the street is due to the pathfinding being unable to plot a path through buildings.

You can see the same happening if you have a squad in front of a long block of buildings, and you plot a single movement order to the street on the other side of the block. The guys will not go through the buildings, they will run around - even if it's a long detour.

So basically the evade AI is trying to do the same thing in your example with the courtyard and walls, @Macisle.

Evade uses the fast move command toward the evade waypoint. That means the path chosen between the original and destination waypoint would be one that have least possible obstacles (or bottlenecks) present. I´ll set up a test to see if buildings are overally treated as an obstacle, or if it´s just the doors providing a bottleneck. I´ll remove walls (front and back) for this purpose.

It seems the path finding routines do attempt finding paths for each single soldier seperately and not quite in organized ways for the whole unit. That´s were bottleneck terrain (doors, bocage gaps...) causes squad members when not finding a path in limited time choose an alternative route to the destination waypoint. That´s also where squad members get seperated most the time. A better SOP for this situation would be for single soldiers stand up and move through the bottleneck one after the other and not all (attempt moving) the same time.

Link to comment
Share on other sites

I decided to test the new engine with a QB.  I gave some infantry units a Move order. They started walking across an open field towards a small forest area maybe 200m away. The AI started an artillery strike to this forest. After some rounds had landed my infantry did exactly the opposite of " AI PROACTIVELY AVOIDS HIGH-EXPLOSIVE FIRE ".

IMO once the first round exploded in their movement direction they should have stopped and hit ground.
Instead they increased speed and ran to the forest where rounds were exploding and suffered the first casualty soon after.

Since I just started testing it I don't know how this new feature works in other situations, but in this case it didn't seem to be working.

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...