Testing Infantry Modeling 1.03

[M1A1TC]

I created a test scenarion to check new 1.03 infantry AT use versus enemy tank.

Download this scenario, then order quick, hunt, move around the corner until enemy contact. Observe infantry(if they move to closest cover, deploy AT weapons, stay in place and die, etc.

http://www.mediafire.com/?cddmxhnxzy3

[stikkypixie]

Nice, we can use this to test 1.04 as well.

[M1A1TC]

I used 4 Stryker Infantry Squads with AT-4 versus 4 Syrian tanks. They each start out in different cover, and also have different cover (trench, grass, wall) available once they come in contact with enemy tank.

One thing I noticed that tank will still fire through the wall and destroy infantry with MG

[John Catsack]

It also seemed odd that, in broad daylight at short range, the T-55 spotted my infantry first. In the left-middle test my infantry didn't see the tank until a shell landed on them.

[Cpl Steiner]

My squads behaved like the greenest of new recruits, getting themselves wiped out every time. It's a crying shame as the graphics of CM:SF are without a doubt streets ahead of CMx1. If only the game-play was at the same level!

Unlike some who say this proves the game is fundamentally flawed, I wouldn't go that far. I just think infantry Tac-AI is very poor at the moment. I am still hoping it can be substantially improved. For BFC's sake, it has to be.

The main problems I saw in this mini-scenario:

1. Infantry walk right past walls without even attempting to spot anything around the other side.

2. Even when fired at by a tank at virtually point blank range, they take several seconds to spot it, by which time half of them are dead.

3. Even when they have AT-4s available, they don't seem to want to use them without a specific targeting order. Or else it takes considerably longer for them to do it - longer than any player is prepared to wait.

4. When 2 men were left and they decided to run for a ditch, they both hit the dirt as soon as the first man reached it, leaving the other one out in the open. It's as if they can't individually take cover. Not sure if this is the case but it looked like it in this limited test.

I will try the same scenario with 1.04 and see what the difference is.

[M1A1TC]

I had exactly the same results most of the time

I would expect:

As soon as contact made, i.e.first guy dies, rest of squad hits dirt, go for cover, fires ATs at the tank. Sometimes this happened, but it took a very long time. Why cant they spot a tank that is in the open, daylight, and about 20 feet infront of them, until it fires a shot first?

[ September 29, 2007, 04:36 PM: Message edited by: M1A1TankCommander ]

[KNac]

Yes it's strange... I see the defender ALLWAYS spots before the attacked does. Chek out the '1:1 abstractions workaround' thread the screenshoots I posted an you can see. Even before the advancing us rifle squad had stopped and they got positive conatct report, the enemy was shooting allready.

I can almost certainlly say this is niot due to LOS/LOF checking algorithm, as if enemy has LOS to you, you have it to them, but your advancing squad seems to percibe the enemy allways later, a faltal second later or so. What is a bit flawed is troops not checking corners before advancing, for example, but as I've seen squads assault building very nicelly when they want, I'm sure this can be coded somewhat and there are workaround to current LOS system limitations. Also there are soem tricks scenario designers can do, like placing buildings corners in the center of action spots, opposed to the corner of action spots.

Haven'ttried your test scenario maybe later.

[John Catsack]

I don't think it's unreasonable that a moving target would be spotted before a stationary one, but it probably shouldn't happen when a unit is hunting or otherwise looking carefully for the enemy and it certainly shouldn't happen when the stationary unit is a tank sitting in the open.