The Wrong Left Turn and the Uncanny Valley

[Guest Guest]

deleted per user request

[CommC]

Quote:

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

Michael:

A modern armored vehicle can easily travel at speeds of 60 km/hr or so ... thats only 20 minutes to cover the entire 20 km map.. on 6 km map, thats only 10 minutes to travel across the entire map. Where are you getting this 3 hour timeframe?

Game designers need to seriously think about what they want the player to do and what contributes to victory vs defeat. If you just line up two armies in front of each other and shoot it out, you will have missed much of the depth and essence of modern armored combat. Maneuver to flank and use of terrain is everything.

I hope whatever game emerges from CMSF development will consider these aspects of armored warfare.

[Michael Dorosh]

Originally posted by CommC:

Quote:

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

Michael:

A modern armored vehicle can easily travel at speeds of 60 km/hr or so ... thats only 20 minutes to cover the entire 20 km map.. on 6 km map, thats only 10 minutes to travel across the entire map. Where are you getting this 3 hour timeframe?

Game designers need to seriously think about what they want the player to do and what contributes to victory vs defeat. If you just line up two armies in front of each other and shoot it out, you will have missed much of the depth and essence of modern armored combat. Maneuver to flank and use of terrain is everything.

I hope whatever game emerges from CMSF development will consider these aspects of armored warfare.

Combat Mission is a combined arms game, not an armoured sim - you can spend a few hunny and get Steel Beasts for that, can't you?

I wouldn't be disappointed at all to see tanks in Combat Mission relegated back to their proper place in the scheme of things, and letting the infantry do more interesting things, as is the case historically. Tanks have always been over-represented in CM. Less so in CM:SF, given their prevalence in modern operations, but even so, infantry is dramatically uninteresting because they have so little capability.

Make the infantry more interesting and you don't need to advance to contact for 20 grid squares with the armour - that's a different game altogether.

Let's see infantry properly breaking down into fireteams, fight from (and assault) two-man trenches, scale cliffs, send out scouts and flankers, patrol for the enemy, snatch prisoners, and so all that stuff the infantry is supposed to do - instead of just being window dressing for the tanks.

[Pinetree]

Ssnake of Esim games(Steel Beasts) has made an interesting point in this thread. He believes that..

..many areas of the engine clearly are rooted in WW2 - maps are generally too small, there are no metrics for collateral damage etc.

(An excellent semi-review by him in that thread BTW)

This is my feeling too. CMSF,IMHO, does feel a bit like a WW2 game with a modern twist. As much fun as it is, I think there should be some sort of penalty for flattening half the neighbourhood...

[Michael Dorosh]

Originally posted by Pinetree:

This is my feeling too. CMSF,IMHO, does feel a bit like a WW2 game with a modern twist. As much fun as it is, I think there should be some sort of penalty for flattening half the neighbourhood...

This exists in the editor already - it is possible to penalize a player for destruction of property.

[Pandur]

instead of just being window dressing for the tanks.

hehe

as far as i understood, game whise and also partly to reality transferable;

you have the vehicles for overwatch or special fire missions likea area targets or generall support, and ovcourse primary as anti tank weapon.

some vehicles are also primary designed as taxis and support.

but that gets you nothing as attacker when all your inf is also around(or in) them. the enemy wont show up becouse there is nothing to shoot at for him.

so you need to get closer with the cheepest stuff you have->infantry.

THAT will make the enemy fire out of 1 simple reason id say.

he fears that you come to close and so negating his natural bonus he gains by his position compared to you wich have to move over mostly unvaforable ground. if you reach his cover its your cover too more or less.

that will make him open up if he can risk compleat destruction of the unit(s) by the overwatch, if not he can still hide or try to sneak away. if the overwatch put up enough psycological pressure(in muliplayer )to keep the enemy hidden you won more than you lost.

now as it was in TOW and to "some" extent also in CMSF you dont need the infantry that hard.

i KNOW we asume that vehicles have good or superior optics but that doesnt help that good when the enemy doesnt like to show himself and knows about "good optics".

in the end, right now you need nothing to "invite" the enemy into presenting himself. you can peper away from afar and thats it since we asume that superiour optics spott enemys out of nowhere. you can clear out houses to the last guy wihtout infantry.

but that was all said befor and will said again, i hope it get a bit tweaked...

[monkeezgob]

Originally posted by Pinetree:

Ssnake of Esim games(Steel Beasts) has made an interesting point in this thread. He believes that..

</font><blockquote>quote:</font><hr />..many areas of the engine clearly are rooted in WW2 - maps are generally too small, there are no metrics for collateral damage etc.

(An excellent semi-review by him in that thread BTW)

This is my feeling too. CMSF,IMHO, does feel a bit like a WW2 game with a modern twist. As much fun as it is, I think there should be some sort of penalty for flattening half the neighbourhood... </font>

[thewood]

I played the original steel beasts and thought infantry was way under modeled. SB was really built for a cols war type full front war. Get into urban combat and CMSF, with all its warts, really shows what it can do.

[JonS]

Originally posted by Michael Dorosh:

</font><blockquote>quote:</font><hr />Originally posted by Pinetree:

This is my feeling too. CMSF,IMHO, does feel a bit like a WW2 game with a modern twist. As much fun as it is, I think there should be some sort of penalty for flattening half the neighbourhood...

This exists in the editor already - it is possible to penalize a player for destruction of property. </font>

[Aacooper]

I think the scale is fine -- this isn't a long range armor shoot-out game, and assuming you were allowed to make larger maps, you'd probably end up with a poor frame rate. If you want an armor shoot out with manoeuver, create a Golan heights scenario where the typical range was 500m or less.

On the other hand, scenario designers should avoid making maps like in one or two of the campaign scenarios where you have M-1s on one end of the map destroying T-72's as soon as they are teleported onto the other end of the map as reinforcements. I don't know if any two-player playable scenarios are like that, but that's neither fun nor "realistic".

[moneymaxx]

Originally posted by GreenAsJade:

You don't have to have 1:1 representation to corretly model the response that you described of a squad in a building to MG fire on the building.

It would be interesting to find a concrete example where 1:1 representation gives you something you really can't have in abstract form, must have 1:1, and really do want to have in commanding a company of squads to do its job...

GaJ

Before I try to answer I'd like to give another CMx1 example. HQs are very good at killing tanks with just grenades from up to 30 meters. There were threads about this and I think to recall that the reasoning was something like this, the grenade isn't actually thrown but a member of the HQ approaches the tank and places the grenade, it's an abstraction. But that member is never killed if the HQ is in a position where it can't be seen by enemy units even though the approach path to the tank is covered by dozens of enemy units.

In a perfect 1:1 representation that wouldn't happen (again, at this stage CMSF is IMHO far away from that), one soldier steps out in the open 1 gets probably killed.

Yes, you could model that in abstract form, one example would be rules (if a unit is in a concrete building and draws unexpected fire from a MG gun then the probability of loosing X squad members in the first 30 seconds is......). But then you would end up with a large list of rules for specific situations in which the abstractions don't work as intended. Like in the above example where there is a mismatch between the LOS abstractions (the HQ in an unspotable location) and the abstraction of the grenade attack (1 member of the unit approaches the tank, should be sighted but retains it's invisibility).

To wrap this up I'd like to ask why would one want to stick with abstractions if there was perfect 1:1 representation. The combat results would be more accurate, the graphical representation more satisfying and you wouldn't, as a programmer, have to worry so much about the interaction of various types of abstraction.

I think that CMSF still is far away from the "holy grail" of 1:1 representation/calculation as many threads about TacAI, LOS/LOF etc. show and I really don't know if they can make it work in the future (I hope so), but for me it's the right direction.

[Michael Dorosh]

Originally posted by moneymaxx:

To wrap this up I'd like to ask why would one want to stick with abstractions if there was perfect 1:1 representation.

We'd have to agree on what "perfect" means. I doubt we could.

The combat results would be more accurate, the graphical representation more satisfying and you wouldn't, as a programmer, have to worry so much about the interaction of various types of abstraction.

It doesn't necessarily follow that the results would be more "accurate". Just the simple matter of getting squads into a building will be interesting when we get to WWII. There are specific drills for it now. Were there in 1944? What kind of animations will we see? I see that as an example where abstraction may be better than specific representation. Having 1944 soldiers doing 2006 battle drills would be unsatisfying.

EDIT to add, though, that I'm looking forward to the 1:1 debates concerning the use of the Bren Gun on the defensive using the sustained fire kit.

[ August 12, 2007, 09:39 PM: Message edited by: Michael Dorosh ]

[LongLeftFlank]

Originally posted by Michael Dorosh:

Just the simple matter of getting squads into a building will be interesting when we get to WWII. There are specific drills for it now. Were there in 1944? What kind of animations will we see? I see that as an example where abstraction may be better than specific representation. Having 1944 soldiers doing 2006 battle drills would be unsatisfying.

Great point, Michael and it really opens a can of worms regarding premodern battlefield psychology. SLA Marshall's Pareto heuristic (i.e. 1 in 5 ETO GIs uses his weapon) has taken a (not entirely undeserved) hammering on this board. So to what extent will we see Wignam's "3 or 4 gutful men" leading assaults while the bulk of the squad hangs back or cowers outright? Will huge performance gaps be assumed to exist between "green" and "elite" units that a player can bank on? Or will combat effectiveness be treated be more randomly, a la Squad Leader: "men fight like lions one moment and scared hares the next"? Will there be berserk type incidents, or Audie Murphy heroes?

[ August 12, 2007, 10:00 PM: Message edited by: LongLeftFlank ]

[Michael Dorosh]

I hadn't thought of that LLF - Strome Galloway says the same things as SLA Marshall - that most riflemen really didn't do anything in combat. So how is THAT going to look in 1:1?

And yeah - what about the outliers? Berserkers and heroes. We have fanaticism in CMX1.

I was going to say that it will get as dull to see a squad enter a building the exact same way over and over, just as it is dull to hear the same .wav go "MOVE OUT, MEN" over and over. But, of course, you can have more than one animation for everything and simply randomize them. However, your point about outliers is very apt. Will all riflemen simply blaze away heroically? Because the anecdotal information - if one ignores Marshall - is that it is unrealistic to depict them that way.

Will we see national characteristics crop up? Squad Leader handled the US troops very differently than other nationalities - Greenwood mentioned in The General that even so, the troops in SL still charged with an abandon "that would make Kelly's Heroes proud."

[moneymaxx]

Originally posted by Michael Dorosh:

It doesn't necessarily follow that the results would be more "accurate". Just the simple matter of getting squads into a building will be interesting when we get to WWII. There are specific drills for it now. Were there in 1944? What kind of animations will we see? I see that as an example where abstraction may be better than specific representation. Having 1944 soldiers doing 2006 battle drills would be unsatisfying.

With accurate I was only referring to the combat results not to the visuals. I understand that if we talk about visuals we enter grog country and I dare not to venture there .

As I said earlier, trying to kill an enemy squad in a heavy building isn't a viable tactic, in CMx1 it was because of limitations of the game engine. In a perfect 1:1 calculation only the soldier being in the LOF of the MG would be killed.

By saying perfect I mean such things like, no firing through walls, the squad members will, depending on the tactical situation, either engage the unit in full force or search cover inside the building, if ordered to leave the building the squad members will identify and use cover in relation to the perceived threat axis etc..

Obviously, to succeed in the creation of such a 1:1 system, one would have to program a terribly good TacAI otherwise, I think, we would be better of with abstractions.

[dalem]

Originally posted by moneymaxx:

To wrap this up I'd like to ask why would one want to stick with abstractions if there was perfect 1:1 representation.

Because unless you're the fire team or squad leader, you're commanding units, not individuals.

The only way it's gonna work is if you show me individuals but let me not "care" about them, and the only way I'm not going to "care" about them is if their actions and reactions are "reasonable" in the context of the game's output.

-dale

[dalem]

Originally posted by Michael Dorosh:

EDIT to add, though, that I'm looking forward to the 1:1 debates concerning the use of the Bren Gun on the defensive using the sustained fire kit.

One of my ASL Journals has a line or two about the Bren tripods found on the Universal Carriers. I'm waiting to pull that one out until (if?) the WWII version is made.

-dale

[moneymaxx]

Originally posted by dalem:

</font><blockquote>quote:</font><hr />Originally posted by moneymaxx:

To wrap this up I'd like to ask why would one want to stick with abstractions if there was perfect 1:1 representation.

Because unless you're the fire team or squad leader, you're commanding units, not individuals.

The only way it's gonna work is if you show me individuals but let me not "care" about them, and the only way I'm not going to "care" about them is if their actions and reactions are "reasonable" in the context of the game's output.

-dale </font>

[Battlefront.com]

I'm glad that most of you seem to understand that we really had no choice but to go 1:1, even if you don't fully understand the array of reasons for it (though some of you seem to grasp it quite well). It is a complicated and fundamentally difficult thing to debate because of that. The best way i can put it is there were many reasons to move forward with 1:1, there were very few to stay with gross abstractions. Since there really isn't anything inbetween, 1:1 it is.

The perfect form of 1:1 is everything visual looking exactly like it would in a real wartime environment such as the game is simulating. I think it is fair to say that we'll never see perfect 1:1 combat in our development lifetime. It isn't just the hardware as the massive amounts of AI programming that are needed. And the more AI that is executed, the more powerful the system has to be.

Well, then go "WeGo" and be done with needing to keep things calculated in RealTime, some have said. Well, it doesn't work that way either. Everything that happens must be stored and recorded. Everything. That means the Soldier Dude #124 that spent 1/4 second hesitating near a rock long enough to get his foot over it... guess what? That has to be stored. You guys have no idea how much data we're talking about for even a few seconds of play. PBEM games with only a company sized force on each side are already pretty big. Every change we make to synch up the 1:1 environment means more data stored which mens bigger files. So let's keep that in mind.

This is actually one of the advantages of RT. The information is streamed onto the screen and then promptly forgotten about. When you save the game it saves the current nano second only, nothing more.

Many of the complaints we see about this or that aspect of 1:1 simulation are quite fixable. In fact, today Charles just fixed one of the big ones people have been complaining about. It won't take us long to get most of things that are irking people fixed. There are others, however, that are either going to stay as they are now or will only be improved a little bit as time goes on. The only one I've seen mentioned thus far that falls into that sort of problem is the 1:1 reaction of soldiers to their environment. We can make improvements, and will, but some amount of visual abstractions will remain forever.

Steve

[Michael Dorosh]

Originally posted by dalem:

</font><blockquote>quote:</font><hr />Originally posted by Michael Dorosh:

EDIT to add, though, that I'm looking forward to the 1:1 debates concerning the use of the Bren Gun on the defensive using the sustained fire kit.

One of my ASL Journals has a line or two about the Bren tripods found on the Universal Carriers. I'm waiting to pull that one out until (if?) the WWII version is made.

-dale </font>

[Battlefront.com]

moneymaxx,

But only IF that position in space is "reasonable" and I think we agree that this hasn't been achieved in CMSF (so far??) with individual soldiers positioning themselves in absurd locations only to be butchered.

This is a bit hard to generalize, though, since there are basically two issues:

1. Is the TacAI putting the guys in the wrong Action Spot (which is what we call a single tile during gameplay)

2. Is the simulation doing something wrong while the unit is in that Action Spot

These are two entirely different issues. If I tell a unit to go along a street and take a right, but instead of following the street it goes into a building and out the other side... that is the first type of problem. The second type is a Team being in the exact Action Spot you want it to be in (or you find it a reasonable place, at least) and the guys aren't interacting with the terrain as you think the should. Of course, there is also a third type and that is where the terrain itself isn't working right, like the wall LOS/LOF problem we will get fixed soon. But let's not get that one confused in the 1:1 discussion because that sort of thing happened even in CMx1 without 1:1 simulation.

Getting the TacAI to *not* move into certain spots is generally something we can fix. As I said above, today Charles fixed a pretty annoying one and you'll see that in v1.03. Some things we can't fix, like people expecting they can put down 1 or 2 Waypoints 100m away in a complex environment and have the unit read the player's mind to ensure that the unit goes along the EXACT path that the player had in his mind. We can make it tend to move in more predictable ways, but there is no way to make it do exactly what the player has in mind without some help in the form of shorter paths and more waypoints. Well, not in a realistically laid out map.

Simulating the dudes doing little things you expect to see done is a lot harder. It involves increasing the fidelity of the terrain, animations, AI, etc. Due to the practical limitations I mentioned above, there are limits and that means no matter where we draw the line it won't be anywhere near what people perceive as perfect. Therefore, we have to include some abstractions or the game would never be able to run or we'd never have the time to finish it. Neither are viable options, so some amount of abstraction will always remain.

The trick is, just like CMx1, to make the abstractions balance out the problems. We've got more of that going on than anybody realized, and that's intentional. We never highlighted the abstractions in CMx1 What we can do is educate you as to where those abstractions are and what to expect from them. This took about 12 months to do for CMBO, so I hope nobody expects that we'll cover all these things in great detail within the next couple of days

As with previous CM games, it is probable that we don't have all the abstractions tweaked the right way for all situations. We can play around with stuff when attention is focused on something specific and we see the validity of those observations. We did this for nearly 4 years with CMx1 so again, let's not expect everything to get identified and tweaked by next Friday CMx1 games evolved after they were released, so to with CM:SF.

Steve

[Battlefront.com]

Oh, and I meant to add (again) that all wargames have abstractions and CM:SF is no exception. Simulating things 1:1 theoretically means no abstractions, but completely realistic 1:1 simulation of combat is simply not possible. Therefore, 1:1 offers CM a means of dramatically reducing the abstractions found in CMx1. By doing so the game is not only richer and more realistic, it is also easier for us to work with. And that means we can do more things with the CMx2 engine than we ever could with a CMx1 type engine based on non 1:1 elements. The full benefits of this aren't apparent to you now because you are only seeing the first iteration of the engine. Time will show why 1:1 was the way to go from a simulation point of view.

In case any of you think this all sounds like BS and we just did 1:1 to make more money... you're not completely wrong. We will make more money off of CMx2 than we ever could off of a CMx1 type engine thanks (in large part) of 1:1 being at the core of the game engine. That means that we are "guilt as charged" of caring how much compensation we receive for years of hard work. However, that doesn't mean that the rest of the reasons are BS. 1:1 took the shackles off and that's a great thing for us as creative game designers. We probably would have gone 1:1 even if it meant no potential increase in revenue. The fact that it will increase revenue is just icing on the cake as far as we are concerned.

Steve

[dalem]

Originally posted by Michael Dorosh:

</font><blockquote>quote:</font><hr />Originally posted by dalem:

</font><blockquote>quote:</font><hr />Originally posted by Michael Dorosh:

EDIT to add, though, that I'm looking forward to the 1:1 debates concerning the use of the Bren Gun on the defensive using the sustained fire kit.

One of my ASL Journals has a line or two about the Bren tripods found on the Universal Carriers. I'm waiting to pull that one out until (if?) the WWII version is made.

-dale </font>

[GreenAsJade]

Getting the TacAI to *not* move into certain spots is generally something we can fix. As I said above, today Charles fixed a pretty annoying one and you'll see that in v1.03. Some things we can't fix, like people expecting they can put down 1 or 2 Waypoints 100m away in a complex environment and have the unit read the player's mind to ensure that the unit goes along the EXACT path that the player had in his mind.

I'm curious about this one.

It seems to me that there are lots of cases where the unit should easily go along the path I had in mind: it should follow the line that I drew on the ground!

Only if I asked it to go somewhere where it can't... then it has an excuse to deviate.

And when it deviates, it would be most playable (it seems to me) and comprehendable if it stayed as close as possible to the line I drew on the ground. That's not "reading my mind" it's a pretty well bounded problem to solve.

Right now (1.01 demo) I have a unit reproducable tripping of for _miles_ from my line for absolutely no apparent reason.

GaJ

[Michael Dorosh]

Originally posted by Battlefront.com:

What we can do is educate you as to where those abstractions are and what to expect from them. This took about 12 months to do for CMBO, so I hope nobody expects that we'll cover all these things in great detail within the next couple of days

As with previous CM games, it is probable that we don't have all the abstractions tweaked the right way for all situations. We can play around with stuff when attention is focused on something specific and we see the validity of those observations. We did this for nearly 4 years with CMx1 so again, let's not expect everything to get identified and tweaked by next Friday

Yep. Perception's the killer. Like we've all said before, the thing about computer games is that no one knows what goes on under the hood. Instead of just a good gamer's guide the way CM:BB had, I think the game would benefit from a good set of Designer's Notes, just like the old boardgames had - either included with the game, or done in articles in the hobby press - in which issues like that could be fully explained. It makes the game that much more comprehensible, and frankly, rich for those that desire that level of understanding.

And it would give dalem something to do in the crapper.