The Wrong Left Turn and the Uncanny Valley

[Battlefront.com]

moneymaxx,

For me comparing CMx1 and CMx2 is like comparing a matured technology with the possibility for only marginal improvements with a new technology that has much more potential but in its initial installment is inferior (like the first TFTs compared to CRT monitors, the first digital cameras, cars and carriages ....).

Very well put.

Hoolaman,

I agree that 1:1 modelling is a lot more unforgiving, and much more difficult to get right, but if you agree that it could be done right, the problem becomes the execution, not the concept itself.

Also very well put. This is the downside of 1:1 modeling and it is the one we recognized way back before it was even practical to attempt to implement it. The problem was, as noted in my previous post, the days of us being able to avoid rising to the challenge were growing smaller.

Sorry guys, the squad/team abstractions in CMx1 were unappealing to many. They were initially put up with because of the novelty of CMBO. But with each passing year less and less people were interested in repeating that game experience.

We made a decision to not do a "Talonsoft" and continually milk the same small group of people over, and over again with the same engine. That was a deadend route to take and they indeed deadended. The alternative was to push beyond the CMBO paradigm and embrace 1:1, challenges and all.

As stated above, it was the right choice to make and we already know that for sure.

Steve

[Battlefront.com]

if BFC re-abstracts squads I'll feel like a five year old who just had his birthday cake stolen by a gang of evil clowns.

Don't worry Konstantine, the evil clowns aren't don't have control of the party

This discussion is interesting, but it is extremely blindsided and a tempest in a teapot. We're not going to undo 3 years worth of work to go back to a system that we identified as dead 2 years before we even started work on it.

1:1 is here so live with it or move on to something else.

Steve

[Runyan99]

Word.

1:1 is flawed in CMSF, but it still is the right way to go looking forward. Abstraction at the squad level is for old boardgames.

They key is to make 1:1 work better and more realistically as we progress into the future. Personally, I think 1:1 is a little bit crappy in CMSF 1.01, but I can imagine how it could work so much better with a series of adjustments.

[LongLeftFlank]

Originally posted by JasonC:

At the risk of recommending something hopeless in implementation terms, I point out that many AIs use "influence maps" for things besides pathfinding.

Example - I'm infantry. So I take every spotted enemy on the map and I estimate his visible areas, weighted perhaps for range or some estimate of soft firepower. Every spot now has an enemy fire value, and I should take it into account in pathfinding. I look for minima of it, near me (especially if my morale state is poor etc).

I can do the same with my own visibility. Suppose I have a hide order. Then I want small LOS footprint. Suppose instead I am in good order and ready to fire. Then I perhaps want LOS to a known enemy, and if there are none, wide LOS or LOS that creates a "threat map" that overlaps with a "close to known enemies" map.

Any comments on this idea, Steve? I'm still willing to be convinced that it's just my lousy playing, but I'm losing a lot of squaddies swanning about in a crossfire and no amount of micro seems to help much. Their sense of self-preservation seems distinctly zombie-like.

[molotov_billy]

Originally posted by moneymaxx:

In CMx1 you could kill an entire platoon in a heavy building (e.g. a church) by MG fire, because the probability to be hit is shared by all members of the squad. But that's not very realistic, most squad members should find perfect cover and while some unlucky guys standing at a window might be killed at the beginning, the rest should be 100% save for the rest of the engagement.

Strange that you should mention this, I'm seeing the very same behavior in Shock Force - in fact relying on it to finish missions. Either walls are not being modeled correctly, or all small arms penetrate everything they hit, or there actually is an amount of abstraction going on - because it's the norm for entire squads to be wiped out by small arms alone, from inside structures. Even when firing indirectly at a building section - where every bullet is visually hitting the center of the wall, not dispersed - entire squads are wiped out.

This goes back to what JasonC said about only modeling half of the story - so what if you simulate the arc of every bullet, if on the flipside your 1:1 troops suicidally cling to the windows and walls in all situations, never taking cover - and when they do make their cowering animation, bullets are passing through walls to kill them, anyway.

I can clear out any urban terrain in shock force with small arms alone, without ever using high explosive or entering an enemy occupied building - and to the last man. Review your final map situation at the end of these scenarios - their's nothing left. No broken half-squads hiding in the middle of a house, or avoiding contact all-together. Each soldier is up and firing nearly the entire time, during every contact with the enemy, until they are dead, or until there are no more targets. Doesn't work that way in reality.

Sure, 1;1 works in theory, if you model all of the relevant systems. But because of time constraints or processor speeds, only weapon systems are accurately modeled - oh, and the vision obscuring effects of dust, of course. The result, in SF, is a broken one.

Who's saying that we should go back to three man point representations of squads? Nobody, I hope. But it certainly is true that SF will not be a valid simulation until these issues are worked out.

[Sergei]

The real test will be in simulating the Finnish 'sisu' 1:1 in CMBBx2. And pinecones. It better have pinecones!

[Runyan99]

Yes, and we'll finally get to see how the Finns could fend off an entire Soviet armored division with some rope and logs with 1:1 accuracy.

[PFMM]

For the most part, all I see are people producing perfect answers and then fabricating questions to fit that answer. If any of those people ever design something, you will quickly dicover this is a certain path to failure and disaster.

[patrik pass]

1:1 was great step forward. for me it changet the whole tactics. because i care more for individula soldiers and im motivated keeping looses at zero(or near zero)

i dont get why michael dorosch thinks, taht problems(bugs) in shock force are all rooted in 1:1 vizualization.

ai can be improved/pathfinding too/multicore optimalization/. and sure it will be.

shock force is for me great game(im enjoyig it like modern steel panthers in 3d).

for me is sohock force much better as previous combat mission games. michael dorosh said, that shock force is somewhere in the middle and now noone is really happy with that. then - im the middle man! shock force is now great and will be greater for me!

im playing in real time. totaly forgot heavy time consuming/and after some time boring wego.

on the other side - im playing against computer. didnt tried yet realtime play against human(so here have michael dorosh maybe point) but again - when i should have problems managing units in real time...my oponent will be human too.so he will be fighting with same problems and same time limit. thats sounds fair duel to me...

[patrik pass]

i forgot to add:

somewhere in the middle is for me theatre of war. impossibility to use buildings is so absurd, that it deform whole tactics. do you think that shock force have such a problems?

[Alpha Mike]

Originally posted by rlg85:

</font><blockquote>quote:</font><hr />Originally posted by Alpha Mike:

Too many words.... Head Exploding....

Battlefront, please note: CM:SF pleases *me*

Ok asking this nicely...

If you had bothered to read the posts preceding yours, you would see that your post contributes nothing other than spam.

So would people stop making drive by posts in every random thread with "This game sucks" or "This game rocks" generalizations?

MD's great post and some of the resulting discussion however, provides real feedback for BFC to consider (whether they do anything with it is up to them, as CM is their baby and they can do what they will)

If you had bothered to read (instead of complaining about words) you would have seen there was a legitimate discussion about the merits of 1:1 in this scale.

Fanbois posting how great and completely flawless the game is over and over followed by people set on hating the game posting how bad it is over and over pretty much accomplishes nothing other than adding posts to the board. </font>

[GreenAsJade]

I think that what this boils down to is that Steve reckons the way to go is to get 1:1 working properly, so although I don't care if Private Jones is picking his nose, at least if it matters then that info will be there.

Fair enough... bring it on!

GaJ

[thewood]

Whether its the 1:1 scale or the premature release, its a major departure for BFC from its roots. The 1:1 scale bothers me much less than the premature release. I really hope they go back to the old business model of being done when its done. I have wasted enough of my time trying to figure out what is a bug and what is a feature. I think they have done a big disservice to thier arguement on scale because most of the things people a are complaining about with the RT and 1:1 scale would seem to be bugs and have nothing to do with change in game mechanics.

[Molinator]

The problem is not the 1:1 representation, the true problem is that it does not work properly so far. The question is: When?, how long we have to wait for the TRUE 1.0 of CMSF?

[DavidFields]

But I think that MD's original point, as I understand it, is that a game element can sometimes be more realistic, and doable by computer, but still should not be done for game design reasons.

An excellent design element in CM, as a contrast, is the element of supply. Tanks have load-outs, infantry has finite supplies. This had so many brilliant and subtle game effects that it is difficult to then go back to any "unlimited ammo" game simulations. For example, sub-machine gun units then have marked differences from HMG. And the length of the scenarios was subtley affected by the decision.

If CM had been a board game, me and my friends likely would have had whole modules involving resupply during the battles (there are truck units after all, and carriers). We would have made rules on how long a unit would have had to remain stationary with a resupply vehicle, or preset "dump".

But, with some thought, my friends and I would have been wrong. Because playing a 150 turn scenario sounds exciting, but usually isn't. For one thing, it is hard to keep play-balance that long, and once play turns one-sided, better to start fresh with a new battle/challenge.

(Theoretically, in long battles players would use something like half their units, and then rotate those units out and put in fresh/supplied units half way through. But how many players have the self-restraint to do that? Most scenario builders know they have to dribble in reinforcements to keep the battles going--this is understanding gamer psychology)

On reflection, it would seem to me that 1:1 Tac AI would not just require as good of Tac AI as CM1...but something like 10 times as good. It is always breath-taking when possibilities go up exponentially--and that is even ignoring possible interaction effects which could push the exponent up further. The TacAI is likely what can make this element work.

It also would appear that Battlefront has decided to bet the company on 1:1 working. [Or, Tow hitting...I guess there are two bets]

One certainly can't fault them for lack of boldness--sort of like charging all your T34s over the hill.

And I would bet that an overwhelming percentage of posters to this forum, whether they currently like or don't like CMSF, want Battlefront to succeed.

[SgtMuhammed]

I think we should be more careful about what we name our threads. I had no idea what this thread was about and only looked at it because it was mentioned in another thread.

Anyway; So far most people here have made very valid points. Some, I feel, are oversimplifying a real problem while others are being overly pessimistic in their outlook.

Is 1:1 the future of computer gaming in any form? Of course it is. The history of computer gaming is a quest for increased realism and fidelity. One day I believe it will be possible to have wargames that allow you to plan grand strategy and command squads in the same battle. Eventually even farther down than that. Say a FPS that gives the player control of every aspect of WWII. Of course I fully expect that future desktops will have the computing power of CRAY super-computers. So the future of 1:1 isn't really a subject of debate. Like Steve said, it is here to stay.

Is it suitable for a company level wargame or higher? Yes provided the engine can handle it, and there's the rub. In the future CMSF will be an elementary school project, but we don't play in the future. It is a lot like the Northrop Flying Wing. There was nothing fundamentally wrong with the design but the systems at the time couldn't handle it. 40 years of developement later and the B2 Spirit is testament to the soundness of the basic design. Like that early Flying Wing CMSF will fly and gives us a great representation of what is to come but some things are still beyond what even a high end PC can handle.

Right now there is only one thing that 1:1 soldiers do like their RL counterparts; get shot. Actually they do this better than real soldiers by several magnitudes. This is the easy part. Simple equations to track polygons allow our digital troops to get shot.

The trickier part is getting them to act like the soldiers they are supposed to represent. I don't care if private Snuffy Smith has a fever or a full canteen but I do care if he can't take cover or climb through a window or enter the door he is told to rather than run out on the street to enter another door in the same building. These are the things I expect the AI to be able to do and if you are going to show me a 150 man company I expect each man in that company to do these things rather than just be able to be shot. I don't expect to be required or even able to give every private his specific orders and to see them carried out robotically but I expect that when I give a squad an order that they will do so with some attempt to do so realisticly. Unfortunately I feel that the current state of computing technology simply doesn't allow this.

There is a quantum leap between programing AI for 20 units per company to 150 units per company. Especially when those 150 are acting on orders given to only a few and attempting to interpret those commands and evaluate their environment and react to it in an intelligent fashion. I have faith that BFC will make every effort to resolve this problem as best they can but I also feel that in this area computing power will defeat them.

In conclusion (finally I hear you saying), I don't feel that CMSF is fundamentally flawed. I applaud their desire to be on the cutting edge rather than just churning out another cookie cutter like so many other companies do. While I think the path they have chosen is the correct one I think they may have taken that step too soon.

[KNac]

sgtgoody, the problem is not computing technology. For the CMSF scale I don't think there is a problem even now with current technology and proccesing power (and I'm not talking about CRAY, just current computers, maybe topend, but current computers), there are projects out there which are much more harder on hardware and have no problems. The problem is for the programmer who have to program the AI.

Hardware is not the limit anylonger, as its developed exponentially, computing power increases a lot (and more it will in the future). But programming AI is quite difficult, the hardest programming area that exist, and making it right even more. Charles have demostrated allready he is a superhuman so maybe he is up to the task haha, but we may never see it as detailed as we want to get it (not with the current BF manpower, as I said earlier in the thread), with a perfect 1:1 rep & a matched 1:1 calculation of physical events as well as 1:1 integrated interaction with the enviorement per the AI.

I'm optismist though, and I think that with some patches we will have a good enough TacAI that will make the game enjoyable as well as natural feeling that things happen as should. With years and if BF continues with this aproach (which it seems) and developes the engine even further, things will be added gradually and a lot of stuff will be really scaled to 1:1 with less abstraction; but for this, the AI must evolve at the same time.

[SgtMuhammed]

Hey, I sincerly hope they prove me wrong.

Regardless, I like the game and I know I will like it even more later. I guess seeing these guys looking like the men I had the honor to lead makes me a little prickly about watching them act like idiots. (Not that I didn't have a few of those.)

[KAding]

1:1 worked well enough in the Close Combat games, no? So why wouldn't it be possible with CM? Is it really that much more complex for the TacAI? If so, why?

Sure the scale of CC was smaller, but I don't see how that would be relevant here?

[ShiftZ]

This an example of a canned cycle or macro that is used in CNC EIA programming for turning centers.

G0X1.15Z.1

G71U.08R.02

G71P100Q101U.01W.002F.012

N100G0G42X-.065

G1Z0F.008

X.7336F.007

G3X.7618W-.0059R.02F.003

G1X.9882W-.1132F.007

G3X1.0W-.0141R.02F.003

G1Z-5.0F.009

X1.15W-.05

N101G0G40X1.15

The G71 lines are instructing the computer to machine a profile with an .08 depth of cut per pass with a feed rate of .012 per revolution while leaving .01 on diameter and .002 on the face for a finishing application. The profile is indicated by everything between lines P100 and Q101. To use this cycle for finishing it would look like this;

G0X1.5Z.1

G70P100Q101

Very simple and abstracted and more then one tool gould use this cycle if it was needed but it creates time problems if there is any internal work like drilling and or boring and all tools are limited to what is in the cycle.

A more complex version but using more code and offering greater flexibility and in the long run faster turn around time would look like this;

G0X1.15Z.1

G1Z.002F.045

X-.065F.012

G0X1.15Z.1

G71U.08R.02

G71P100Q101U.01W.002F.012

N100G0G42X.7336

G1Z0

G3X.7618W-.0059R.02

G1X.9882W-.1132

G3X1.0W-.0141R.02

G1Z-5.0

X1.15W-.05

N101G0G40X1.15

The finishing pass;

G0X1.05Z.1

G0G42X-.065

G1Z0F.008

X.7336F.007

G3X.7618W-.0059R.02F.003

G1X.9882W-.1132F.007

G3X1.0W-.0141R.02F.003

G1Z-4.95F.009

X1.15W-.05F.025

G0G40X1.2W.2

This is a very basic example but I think it's more or less the idea of 1:1 gameplay being expressed by BF.C. The abstractions are perhaps macros of some type and everytime the same situation is encontered the same cycle is used regardless of enviromental changes. To have the flexibilty needed for real time each and every situation requires it's own code rather then a one macro fits all type of programming. Perhaps all of the various situations have not been factored in just yet as the more people play the more problems come out with the program and it's just a matter of editing. I know my example is waaaaay off of whats being discussed but it's how I understand it.

I could be wrong.

[ August 12, 2007, 12:49 PM: Message edited by: ShiftZ ]

[thewood]

A thread can't digress any more than this...I never thought CNC programming would ever be mentioned in a BFC thread. Its a sign that end times are upon us.

[SgtMuhammed]

ShiftZ,

Uh... thanks. I think.

[Pandur]

iam at the point where iam often pretty turned off to play CMSF.

the TacAi is, well,...

ploted waypoints have often nothing in common with the pathes vehicles will take.

the infanty is stuck for regrouping at each waypoint(like TOW), wich leaves many waypoints out of question, also for vehicles to some extent.

you can see the enemys "unit symbols" instead of a "?" as soon as you spotted them...

also hididng isnt possible anymore...i dont understand how ambushes work in RL.

i nearly go to say that everything wich isnt played in urban terrain is sensless. i did a nice open map in editor...its unplayable... .

when there are no houses where you can hide the enemy in, no matter if blue or red, they get spotted as if they wont hide, than you peper em away from afar while they seem to like that, they dont duck or hide in their "trench" or so. mabe thats just 40cm deep trenches!?

since this patch we also have the "vehicles rotate to somewhere" behaviour!? but not as bad as in TOW it seems.

or was that just bad luck on my side!? vehicles rotated in silly direction without my ability to stop it...

all that and more can be tweaked, not so bad in the end but iam also turned off by the "semi" representation of things. getting shot at through entire houses. or through meters of ground,...all that feels like a big step back.

well, sorry for the whining but i feel CMSF has a long way to go. there is an exellent game in the basics but right now i mostly quit scenarios, or the rarely working QB´s, with Alt+Q becouse some of the things wich get daily reported happens every 1 or 2 turns.

i even found myself trieing RT again so i dont see 90% of this strange behaviour but no...no...and no RT.

lets see what the patches will bring...we allready got the nice option to map absolute keys wich is awsome.

CMx2 will shine some day but not now in my opinnion.

[CommC]

Many of us have dreamed of a combined arms company level sim for years and I applaud the efforts of BFC towards this end in CMSF.

I think 1:1 and RT are essential for such a sim. We just need to persevere to get the kinks out. The feature to be able to give orders while paused may seem simple, but is a major breakthough in the thinking on the topic, in my opinion.

The challenge is to get the tac AI, armor/gunnery model, 3D terrain model and operational AI, all working and accurate within the computing capacity of today's PCs. I think this can be done, but it will take some skilled programming and some compromises. I, for one, am more willing to compromise on terrain detail in favor of a more accurate combat model.

But I'm a little confused about the apparent satisfaction in the community about the scale. To me, the scale in CMSF is one of the biggest negatives. The maximum map size is only 6 km, which is only 1.5x the range of the tank weaponry. This doesn't allow room for maneuver, tactics, etc. with armored vehicles. We need a 3D engine that doesn't bog a computer down and can represent both larger maps, at least 20 x 20 km, as well as sufficient detail in urban areas to accurately model infantry operations in buildings, etc. Other programs have done this on the PC, so I know it is possible.

I think the current way CMSF handles infantry in buildings is a great step forward, and just needs some tweaking regarding LOF/LOS issues, etc., to be perfect.

I am thrilled with Battlefront's efforts so far with CMSF, and I fully support their efforts to push forward with improvements and enhancements. I have, and will be, voting with my pocketbook. Battlefront: Carry on... please and don't be discouraged by the whiners...

[ August 12, 2007, 04:40 PM: Message edited by: CommC ]

[Michael Dorosh]

Originally posted by CommC:

Many of us have dreamed of a combined arms company level sim...

I think 1:1 and RT are essential for such a sim...

We need a 3D engine that doesn't bog a computer down and can represent both larger maps, at least 20 x 20 km...

Carry on... please and don't be discouraged by the whiners...

LOL. I have no idea what game you're trying to describe. Something with a 3 hour advance to contact perhaps?