Bahger

There certainly is a pathfinding issue. The OP is quite right. Various vehicles will skyline themselves on the berm and get shot up even when you plot a course for them on the opposite side of the approach. On top of that, the gaps in the berm are not recognized as gaps by the vehicles (or by the camera from certain angles) and the AI will show its ass to the enemy after meandering aimlessly and uncommanded along the ridge. There sure as sh*t is a pathfinding issue.

I'm not a hater and I resent being called that. Quite the opposite, I'm a supporter of BFC and their product but, even leaving aside the performance issues, which I have not experienced except for the Ati-related stuff, the tac-AI is simply not equal to the task of recreating the precision and mobility of modern-day MOUT, which depends on these abilities (plus effective comms and superior intel) in order to seize and retain the initiative in a hostile environment against a numerically superior enemy. With the less sophisticated weaponry, more evenly-matched forces and limited mobility of the WW2 battlefield, the severity of the tac-AI problems is not quite so critical. But if I'm fighting a Stryker company, I want vehicles to go exactly where they are ordered to go -- without doing doughnuts or turning their ass-end to the enemy -- and dismounts to find cover intelligently. I do not want vehicles to be skylining themselves on berms, making uncommanded hail-mary charges at the enemy or otherwise behaving as though they are being driven by drunken rednecks on a Saturday night. Until the tac-AI can live up to the basic concept of the game, CMSF is an absolute failure for me, and saying so does not make me a "hater". In fact, if I did not admire the folks at BFC I would have sold this grievously hobbled game on eBay a long time ago instead of waiting for patches. I've played too many complex combat sims that, for all their issues, do basically live up to their own conceptual aspirations not to be able to recognise that the opposite is the case with this one, unfortunately.

I'm not sure what's in more urgent need of a patch, CMSF for AI or M. Dorosh for I. (Sorry about the above, Michael. As the other poster said, your heart seems to be in the right place but you really should give it a rest.)

Which was the one in CMx1? </font>

There should be some tool available to survey terrain undulations. At the very least, it would be good to be able to get vehicles into defilade rather than hoping for the best when we maneuver them into what might just be some kind of cover...or might not. In short, there is not yet any useable "military horizon" in the game as there was in CM.

Originally posted by The DesertFox: Look, you're right, it's beyond a "matter of opinion" that the game is in a lamentable state. On top of the lobotomised AI and pathfinding I do not understand, for example, how they could not have known that the group select "box" function and the adjust fire commands are useless (for those with Ati cards, I believe). However (and this is a big however): We have to be respectful and constructive in our criticism. CM is not a corporate product. It's the product of hobbyists and enthusiasts whose motives are honorable and their dedication genuine. Now, I happen to believe that they may have made certain commercial calculations that, on top of the bugs and a standard of AI that is just not equal to the task of replicating the mobility and precision of modern-day urban combat, have put the game in a rather precarious place at the moment. I do not understand how Paradox was allowed to go out with an essentially Beta version of the game, or why the devs seemed genuinely to believe that version 1.01 was in a fit state for release. However, one thing you cannot accuse these guys of is cynicism. They are hobbyists, enthusiasts and good guys and there is no other game out there with CMSF's level of ambition or purity of purpose. Now, I believe that somehow their reach far exceeded their grasp with this version of the game but if we can communicate our support for the devs alongside our candid views on the extent to which it is broken, there will be a much better chance that the team will approach the next patch in the spirit of optimism. I have to say, I have never (i) spent $70 on such a dysfunctional piece of software yet (ii) I have never regretted it less because I'm looking at my payment as a contribution to further research and development work. I think BFC will come through. Let's hope.

Cavtroop, your review was right on the money. It was thoughtful, fair, well-researched and took no cheap shots at CMSF. The game is seriously broken in its current state but if anyone can fix it, it's Charles, Steve, Madmatt and everyone else at BFC. They are a great group and I remain optimistic. The idea of a high-fidelity simulation of company-scale combat against a numerically superior enemy fighting on its own turf against an ultra mobile U.S. Stryker force with full support is such an exciting one and I really admire BFC for not copping out and placing the conflict in some imagined state ending in -stan. CMSF has so much going for it but its deficiencies cripple gameplay and immersion to a critical extent and I think the game is in triage at the moment. I believe that deep down the dev team knows this and that they will fix it. Frankly, I'd pay another $50 for Combat Mission: Shock Force with overhauled AI and pathfinding. Thanks for the endorsement, Paul. [ August 01, 2007, 06:55 PM: Message edited by: Bahger ]

If they are chasing the commercial bucks, they are somewhat deluded, as, at its core, CMSF is still a niche product. Unfortunately, attempts to satisfy the COH RTS market have compromised the game's playability (along with AI that is not up to the task of simulating contemporary MOUT). I think CMSF is a noble, but almost absolute failure in its current condition and I'm hoping like crazy that the effort behind the patches will be focused on returning CMSF as far towards its roots as is possible given the irreversible nature of some of the design decisions that have been made. I cannot imagine that the game has sold so well via Paradox that the attempt to grab the commercial brass ring (if that is indeed what happened) can be justified in terms of sales. It seems to me, therefore, that BFC have nothing to lose by doing everything they can to improve AI, pathfinding and WEGO gameplay once they've fixed the technical bugs like Vista, the Nvidia cards that don't work and things like the broken group-select box, shift and ctrl hanging the game and the non-functioning artillery adjust mode for Ati cards. On top of all that, Paradox was allowed to release a Beta game that got hammered by certain influential critics and was even worse than BFC's "final product". What a mess.

Yeah, me too. Anyone?

I just can't play it right now because of exactly these problems. The game is badly broken but the devs are great and I'm hanging in there.

Steve has said there are severe limitations in what they can do to improve pathfinding but as I've said in the thread responding to the Eurogamer review, the problem is very grave because the nature of the type of warfare being modelled in CMSF (requiring speed and precision to compensate for numerical inferiority) has the unfortunate effect of showcasing the games biggest deficiencies (tac-AI and pathfinding). So...It's time for some lateral thinking. Something needs to be done in terms of map design, force size or calculation resources in order to simulate greater battlefield SA on the part of the U.S. forces if pathfinding improvements cannot actually implement it. The devs are facing a conceptual problem here, which is that in its current state, the game cannot meet the standard set by its own concept of warfare.

Oh I really hope not. This is a wonderful assembly of developers with a dedicated team of beta testers with genuine military experience and their product, while terribly flawed, shows real promise. The problem is this, in my opinion: They are all so familiar with the CM series that working around certain critical weaknesses of the game's tac-AI and pathfinding is like second nature. Now I'm not saying you can't play like this (i.e. by integrating problem workarounds into gameplay), and maybe that's how I'd be playing if I'd progressed further through the CM series instead of enjoying the heck out of the first one for two years and then moving on. However, there are two reasonable expectations of non CM die hards that are not being met: (i) People expect that innovations will be matched by a corresponding evolution in the basics. A lot of people will not be thrilled by a vastly improved arty model if a vehicle can still not be relied upon not to stray into a line of tracers for no reason and then dash towards the enemy FLOT like a suicide mission. (ii) MOUT and asymmetric warfare requires a particular combination of speed, mobility, swift deployment and superior battlefield situational awareness. It also requires, at least for offensive operations, that the player seize the initiative and retain it. Every one of these tactical considerations is undermined each time a vehicle breaks formation for no reason just before contact is initiated or a squad runs around a building getting picked off instead of entering it through the door that was plotted in its movement orders. As I say, people immersed in the series might look at this as "the price of doing business" with CM, and if these one step forward, two steps back gameplay innovations had been applied to a game in the WW2 theater, one's ability to ignore their shortcomings would be much greater, as would be one's level of satisfaction with the game. But BFC has taken a big leap in relocating its theater to contemporary MOUT ops and the highly specialised nature of this type of warfare has had the unfortunate effect of showcasing every one of the game's enormous deficiencies in the worst possible way instead of camouflaging them within a theater, like WW2, where comms were weaker, vehicles slower, intel less omniscient and therefore the combat more chaotic. There is a surgical professionalism to modern MOUT operations as conducted by outnumbered U.S. quick-reaction forces and this is what is not integrated into CMSF's current gameplay because of its debilitating AI and pathfinding limitations. Steve has been very articulate in my pathfinding protest threads and I'm both grateful for the interaction and thrilled with the potential of this game. However, through no fault of the dev team, the nature of the combat they are simulating has conspired with the specific nature of the game's weakest aspects (AI, pathfinding) to produce critical mass. These problems are systematic and their resolution is essential to the minimum acceptable level of gameplay in a sophisticated sim like this. The Eurogamer review is right on the money, sadly, and a massive effort will need to be made to patch the game to a point where it lives up to its basic concept. After that they can concentrate on getting it to live up to its potential. The game is very broken and I do hope they can fix it.

As Steve says, the issue here is abstraction, but I'm going to deploy the term in a slightly different way: I can accept vehicles getting snarled up with each other and fired upon as an abstraction of fog-of-war, even when I've plotted good paths for them to contact. That's an abstraction I can just about live with. But it's when perhaps 20% of my vehicles are displaying absolutely no battlefield SA and driving around in circles or driving uncommanded into the midst of the enemy that I can no longer use abtraction as a necessary means of renmaining immersed in a game. I value precision. I know very well the old saying quoted in the manual about the best plans not surviving first contact with the enemy but in 80% of the first turn of missions I've attempted to play, a movement order has produced anomalous results, exposing a vehicle needlessly, turning its rear end to the enemy or otherwise sabotaging the initiative I started out with as the offensive force. No offensive plan can be fun to play out if the initiative is squandered via coding glithces rather than player error; a tactical game loses its integrity that way. If my orderly plan is to fall apart, I want this to be the result of a tactical mistake on my part (I've made plenty) or smart enemy AI rather than persistently crazy pathfinding AI. Last night I fired up GRAW 2. It's not tactically "deep" like CMSF, but the biggest (and necessary) improvement the devs made between GRAW 1 and GRAW 2 is that your AI squadmates go where you send them and face the right way, every time, assuming intelligent cover positions. And you can do this from either the tac-map or using HUD placement symbols. CMSF is potentially a much brainier, more tactically challenging and rewarding sim than GRAW but only if you can rely on your vehicles to go where you send them and face the right way, every time, assuming intelligent cover positions. Sadly, you cannot right now and the result breaks the tactical credibility of gameplay, a hugely destructive thing.

Steve, well the feeling's mutual. Every time you respond quickly, thoughtfully and at length to my comments, the next thing I know I'm firing up CMSF again, hoping for the best... In previous CM games there was very little need to get a vehicle to face in a particular direction at a particular street corner. The failure of the unit to do just that was less critical because there weren't that many street corners. In CMSF, however, entire battles are fought from street corner to street corner and if a vehicle cannot be relied upon to take every separate, driveable route from point A to point B that the commander can legitimately order it to, the player's reasonable tactical options become critically degraded. In Fallujah I imagine a company commander with intel of an ambush beyond visual range would have his vehicles flank by cutting through alleyways and improvising with the urban terrain. I don't want to be fighting the limitations of the pathfinding; I want to be fighting the enemy, but this is where CMSF falls down. However, thanks to you, I'm going to give another mission a shot this evening. I promise not to start a separate thread if I experience the same thing.

My experience entirely, right down to the Steel Beasts comparison. I keep on losing units that wander, sometimes miles, into enemy fire, uncommanded, and because I like to play WEGO, there is no way to correct it. You have to spend an entire minute watching your carefully-plotted movement to contact utterly ruined by this persistent AI behavior. I just don't have the time or the ability to suspend the disbelief that it would take to put up with this nonsense and, quite frankly, the game should not have been released in this condition.

I agree with this and with everything else in your post. I'm a keen enough gamer that I will make allowances for inevitable glitches in otherwise good games, especially in AI and pathfinding. But in CMSF, lobotomised pathfinding utterly cripples an otherwise really well-conceived and executed game and because I have so many other claims on my time, I just cannot persevere. I am committed to CMSF, however, so instead of playing it I'm going to be here, lobbying for the one issue to be fixed that has sabotaged the experience for me. Example: If you design a CQB sim that showcases the tactical advantage that NVGs give our guys in darkness against most militia-style enemies, then you have to be sure that your virtual spec-ops guys can navigate around the maps with realistic eficiency in the dark. It's the same with CMSF: If you want to model a form of combat (near-future, asymmetrical urban warfare) realistically, then you have to implement a Stryker force's speed of deployment effectively because it is a vital force-multiplier. I'm prepared to excuse the occasional glitch and regard others as unintended fog-of-war moments, but whatever you say about your experience of the game, Boom$lang, vehicle pathfinding in CMCF is a complete and systematic failure. Perhaps CM diehards will work around it, but tactical realism diehards will not.

Exactly how am I pouring virtual gasoline on myself and setting myself on virtual fire? Do you really think I'm trying to commit virtual suicide in order to attract attention? Exactly how does this analogy work? It's very rarely that most reasonable people can be completely confident of an argument they wish to make, but this is the case here, for me. I have many years of experience with battlefield simulators and much knowledge of the tactical arena in which modern infantry fights in built-up areas. I know that there is a crippling problem with the game's playability here and I'm committed enough to what CMSG could be to attempt to start a polite, but firm effort -- hopefully aided by others -- to redeem the game before it's too late.

I understand, Martin, but the showstopping aspect of this issue stems from the asymmetry of contemporary urban conflict and the tactical significance of speed and accurate navigation. It would be better if BFC put big green lines on paths through which we could navigate our vehicles with confidence and in formation rather than leave us to hunt and peck our way through the maps wondering if a vehicle is going to balk and break formation, causing chaos at exactly the point of initiating contact with the enemy. My argument is that this is a design issue as well as an execution issue, and much more destructive to gameplay than it was in a battlefield context that didn't rely on speed of maneuver. Can you imagine anything at all being salvaged from the urban battle in Somalia portrayed in BLACKHAWK DOWN if, once the choppers had gone down, the Humvee convoys got lost in the city instead of being able to navigate accurately and at high speed? It's the only way you can fight RPGs on rooftops. For the gameplay dynamic in CMSF to be satisfying, we need to be able to move vehicles with precision and confidence because the U.S. forces have to leverage their mobility and superior training in order to overcome numerical inferiority and win. This pathfinding anarchy (sorry, it's not too strong a word) makes a mockery of the very concept of battle behind the game. [ July 30, 2007, 11:32 AM: Message edited by: Bahger ]

It's a great game so crippled by one thing that playing it is like trying to drive a Ferrari on its rims. That thing is vehicle pathfinding.

As you can tell from the length (and hopefully the articulacy) of my post, I'm passionate about how great this game could be, and how long-lived. But this is not a subjective issue for me, nor would it be for any other serious military gamer I know; if they cannot improve the pathfinding radically the level of frustration will outpace the pleasure of playing CMSF by an enormous margin, I'm sad to say. Thank you for allowing the thread to remain.

If it's okay with you, I started this thread, having decided to stop playing the game, in the hope that there may be more chance of the devs paying attention to a fresh thread than revisiting an old one. As long as you don't mind, that is.

I tried the first mission of the campaign for the fifth time last night and, yet again, a vehicle strayed onto the berm, uncommanded, and got killed before I could move to initiate contact. I also tried one of the standalone battles and spent more time babysitting vehicles through wide urban roadways than fighting my Stryker force. Sadly -- because I spent some serious $ on the deluxe edition -- I'm done with this game until they overhaul the pathfinding. I really admire the Battlefront guys, their concept, most of their execution and their obvious passion for their product. But as I said originally, it's wishful thinking to simulate MOUT with such awful vehicle pathfinding. I'm spending most of my time in this game mouse-clicking my way around corners, micromanaging vehicles into cover, figuring out where the "no-go" zones are, stopping them getting tangled up in each other. You shouldn't have to micromanage vehicle pathing in this way. And vehicles must maintain military formation, especially when moving to contact. I'm sorry but they just don't. They wander all over the place and even my active imagination (along with a reserve of goodwill towards this game) cannot excuse this lobotomised vehicle behavior as "fog of war". It's very simple: if a Stryker can drive up an alley in real life because it is, after all, a very mobile concept of offensive urban warfare, and if there are alleys in the maps, I expect to be able to drive it up that alley, place it in cover, dismount my squad and position them towards a threat. I challenge any one of the devs to prove to me that this can be done right now. It cannot because of both appalling pathfinding and anomalies in the design of the sequencing of orders. It gives me no pleasure to say that the way CMSF plays right now is more like having to be a kid pushing a three-wheeled truck around a sandbox than it is being the commander of the most highly mobile, technologically sophisticated company-level strike force in the history of warfare. It's a crying shame. I think this problem was much less of an issue in WW2 CM titles because the battlefields were less cluttered and speed was not such a tactical issue. However, when fighting assymetrical battles in hostile, built-up terrain, whether city or village, speed is required to gain and keep the initiative. I need to know that if I send a platoon of Strykers there and then have them face that, they will do both, in formation and using exactly the route plotted. This is what we can expect from vehicle pathfinding AI in Steel Beasts, Company of Heroes and even GRAW 2. However, CMSF's vehicles get themselves all turned around even when moving to contact and by the time an enemy is spotted, I have to play traffic cop to a bunch of drunken teenage joyriders rather than take advantage of a fast and efficient deployment. The tactical asymmetry is often to the Syrians' advantage in MOUT; a Stryker needs to flank a Syrian tank so that the infantry can be dismounted with Javelins for a fast, lethal shot. However, such is the chaos in vehicle pathfinding that every attempted flanking maneuver is a pure crapshoot. It's sad. I've bought and played about a hundred high-quality military-themed games across all genres in the last ten years and I was expecting to hunker down with this title, working my way through the single player campaign and playing PBEMs. I was disappointed to learn that PBEMs were saddled with 14 MB swap files but I understand this to be the consequence of the dev team's genuine technological advancements. However, never have I played an otherwise promising game whose very design concept is so completely sabotaged by one all-important aspect of gameplay, which is the need to recreate precise, orderly and military movement and deployment of vehicles. It's an awful shame and quite possibly nobody's fault but I'd have to side with Tom Chick about the unplayability of the game in its present state. It will stay on my hard drive, unused, until the next patch.

I tried this mission again last night and one of the standalone battles. Sadly -- because I spent some serious $ on the deluxe edition -- I'm done with this game until they overhaul the pathfinding. I really admire the Battlefront guys, their concept, most of their execution and their obvious passion for their product. But as I said originally, it's wishful thinking to simulate MOUT with such awful vehicle pathfinding. I'm spending most of my time in this game mouse-clicking my way around corners, babysitting vehicles into cover, figuring out where the "no-go" zones are, stopping them getting tangled up in each other. You shouldn't have to micromanage vehicle pathing in this way. And vehicles must maintain military formation, especially when moving to contact. I'm sorry but they just don't. They wander all over the place and even my active imagination (along with a reserve of goodwill towards this game) cannot excuse this lobotomised vehicle behavior as "fog of war". It's very simple: if a Stryker can drive up an alley in real life because it is, after all, a very mobile concept of offensive urban warfare, and if there are alleys in the maps, I expect to be able to drive it up that alley, place it in cover, dismount my squad and position them towards a threat. I challenge any one of the devs to prove to me that this can be done right now. It cannot because of both appalling pathfinding and anomalies in the design of the sequencing of orders. It gives me no pleasure to say that the way CMSF plays right now is more like having to be a kid pushing a three-wheeled truck around a sandbox than it is being the commander of the most highly mobile, technologically sophisticated company-level strike force in the history of warfare. It's a crying shame. I think this problem was much less of an issue in WW2 CM titles because the battlefields were less cluttered and speed was not such a tactical issue. However, when fighting assymetrical battles in hostile, built-up terrain, whether city or village, speed is required to gain and keep the initiative. I need to know that if I send a platoon of Strykers there and then have them face that, they will do both, in formation and using exactly the route plotted. This is what we can expect from vehicle pathfinding AI in Steel Beasts, Company of Heroes and even GRAW 2. Instead they get themselves all turned around even when moving to contact and by the time an enemy is spotted, I have to play traffic cop rather than take advantage of a fast and efficient deployment. It's sad. I've bought and played about a hundred high-quality military-themed games across all genres in the last ten years and I was expecting to hunker down with this title, working my way through the single player campaign and playing PBEMs. I was disappointed to learn that PBEMs were saddled with 14 MB swap files but I understand this to be the consequence of the dev team's genuine technological advancements. However, never have a played an otherwise promising game whose very design concept is so completely sabotaged by one all-important aspect of gameplay, which is the need to recreate precise, orderly and military movement and deployment of vehicles. It's an awful shame and quite possibly nobody's fault but I'd have to side with Tom Chick about the unplayability of the game in its present state. It will stay on my hard drive, unused, until the next patch.

Steve, I really appreciate your courteous reply to my slighly testy post. I'm going to give you a detailed AAR here so that you can understand the showstopping nature of the problem. Let me state also that I'm an experienced PC wargamer who understands the universal need to work around the imperfect current state of tac-AI but the point at which workarounds become impractical is when AI behavior is so lobotomised and egregious that it disrupts all immersion in the game. - The map looks like there is a gap in the berm, but rotate it and suddenly what looks like a gap isn't a gap. Try to send vehicles through it and they will career off along the crest, getting shot up before turning around and coming back. Along the crest, fully skylined. Understand that all this is uncommanded movement. - Re. the above: I've read my history books about the first day of offensive infantry ops in both the first Gulf war and the 2003 invasion of Iraq and I know this much: If you are going to blow a gole in a berm so that a large force of vehicles can launch into defended territory with the maximum firepower available for the offense, the hole has to be big enough to drive tanks though at least two-abreast. If the game cannot accomplish this, then Rune should not have put a hole there. Neither the graphics engibne nor the vehicle AI seem to recognise a hole as a hole. - Rather than deal with a phantom hole in the berm through which to mount a pincer-like assault on the compound, I decided to drive my task force to the end of the berm and approach from the road. Again, the killer here was uncommanded AI pathfinding. I used "formation move" commands to get the force to the first turn in some semblance of military formation because the shft-left-click command is broken and hangs the game. In four out of four attempts, a vehicle found its way onto the berm when it had not been commanded to go anywhere near there. The Stryker MGS is a particular culprit. In my last attempt at the game, one climbed the berm and the other raced around the edge of the berm and launched a hail mary assault on the compound all by itself, getting shot up at the entrance of the compund when the rest of my force was still keystone kopping its way around the angle between the berm and the road. If it hadn't been so frustrating it would have been funny and give me some credit; I never mouse-clicked a waypoint for that vehicle so far forward of my assaulting force. - The road has sustained damage and in certain places there is a median. This is fine in theory but the pathfinding is so crude at the moment that my tanks responded to these navigation challenges by doing preposterous things like breaking formation, turning their backs to enemy fire and jamming into each other like bumper cars. - At my last attempt at the mission (my fifth) I gave up when, having cleared the path to the compound entrance by moving in bounds and having the tanks overwatch the Stryker convoy, I could not get a recon force of two Strykers through the entrance to the compound without the two of them becoming all snarled up with each other like two women in Volkswagen Beetles fighting over a parking space at a shopping mall. - I always use "Move" and "Quick" movement commands unless the unit is under fire. Guys, it's awful, I'm sorry to say. I will absolutely persevere because so much commmittment and expertise has gone into the game and one day it might be great. But it's nothing but wishful thinking to believe you can simulate MOUT with such woeful vehicle pathfinding and AI problem resolution that makes a mockery of basic military movement discipline. The goofy, uncommanded movements are just dreadful immersion-killers. It's one thing to accomodate one's tactics to the limitations of current-state AI; I've been doing that for many years in just about every sophisticated PC battlefiled simulation I've played. But when tactical control of the AI collapses into chaos that is beyond even the power of babysitting to resolve, it makes a mockery of the game's tactical intent and destroys all immersion. I will try the mission again today. I ordered the deluxe edition of the game and look forward to getting the extra materials. I will stay on this board, playing the game if I can face it, and maintaining a polite, constructive tone at all times. But you guys have to know the scale of the problem you have here. The pathfinding is so bad that it makes a mockery of your very design concept itself. If you can fix this, you may have a great game but not until then.

Yes, the order sequencing is one of the weakest design points in the game, and surprising, given the team's attention to detail in general. It should be possible also to copy and paste command sequences so that vehicle formations might be properly organised. ATM the AI is incapable of maintaining vehicle formations.