Jump to content

Recommended Posts

Posted

Hey All,

Just to say I am an A-10 driver here so my missions focus on ground attack mostly vs Russian designed equipment.

I have building missions exclusively using the highest AI setting for ground vehicles.  But, I notice that they seem just a bit "too unreal" at being "really really good". 

In the real world, what do I really know about the actual skill level of the people that operate these weapons or the weapon system effectiveness? 

I know just about nothing zip nada.

For example, the ZSU-57-2 is a optically sighted with a mechanical firing computer that uses hand and foot cranks to position the gun and turret.  From what I have read it has a great gun, but in general considered not a very successful weapon system.  This is true also for the ZU-23-2 truck mounted AAA that works the same way.  When using Excellent AI both of these weapons become absolutely deadly if I fly into gun range to a level of efficiency approaching 100% during the day.  To the AI's credit they don't shoot well at night.

I can certainly experiment with different AI levels on my own, but I was hoping to speed up the process with a bit of information. 

Is there some information out there that describes how the AI actually performs?  For example response time, firing accuracy, etc...

Caldera

 

  • Like 1
Posted (edited)

IIRC (and I can't find the documentation, so the following may all be bull), in DCS most projectile-based ground UI fire with a predictive bullet trajectory based on distance and speed of the target. The difference between AI levels merely being a randomized deviation from the perfect trajectory: the higher the AI, the smaller the deviation. That's why even rookie enemy AI foot soldiers routinely kill your sorry a** sitting in the cockpit at 1 km distance (a.k.a. 'sniper rookie' - ok, not in a Hog's bathtub, but surely in a Huey). To me it looks as if there is very little modelling of real-world ability going on, with the possible exception of maximum tracking speed and reload time. But there seems to be little else influencing enemy abilities. If they see you, they always aim true minus some minor ability random, and have the uncanny ability to always accurately gauge your distance (position) and speed (velocity vector).  

Edited by cfrag
Posted

Thanks CFrag,

I did a little testing today, just to see.   I made 10 passes at the same speed over a ZSU-57-2 at 5000 AGL for each AI level.  As usual for this game the results surprised me a bit.

Number of times hit in 10 passes:

  • 1 - Average
  • 6 - Good
  • 6 - High
  • 3 - Excellent

Well, go figure this one out... 

I have no clue at this point and I did not feel like doing 100 passes to get more accurate results.

Caldera

Posted

Well, those numbers do track with an RNG-based ability, with perhaps 'Good' a tad high - but we are looking at a low sample base, and highly uncontrolled conditions. RNG seem disingenuous to humans. Ever played XCOM? Your sniper with 99% to-hit probability missing twice? That's RNG. For the time being, it seems safe to assume that there isn't a lot to learn about ground-based AI except that there is very little of it. 

Posted (edited)

CFrag,

"Well, those numbers do track with an RNG-based ability, with perhaps 'Good' a tad high"

Yes, and the Excellent number seems to me too low.  One would expect more of a linear or curved increase of some sort from Average to Excellent.  Not the bell curve result that I got from my small amount of testing.

" Ever played XCOM?"

LOL

I did play X-COM.  Allot...     Basically, I cheated because I saved very often before I moved.  I reloaded, if the result made zero sense to me.  I did not want to fail with a good strategy to a game that basically cheats just to increase the players frustration factor.

RNG you say, that has to be  it.  With much repetition sooner or later even a straight up RNG can be proven to be statistically predictable.  But with only a few repetitions, it is all over the map and often makes zero sense for an expected outcome.  My opinion, DCS may cheat just like X-COM and that the RNG might not be straight up.

 

For my part, I think that I will change my AAA strategy and use the Average AI more often.  Just seems like more fun.

Caldera

Edited by Caldera
  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...