A number of issues should be getting fixed and some QoL features introduced in the upcoming patch last I heard, which should improve situations like AT Guns having improbable consecutive close range misses against vehicles. The devs posted a response to Tim Stone’s review on RPS outlining the mechanics or reasoning behind why some things are happening as they currently are, and in a lot of cases what they are doing to try and remedy it.
Though based on some tests community members over on the Matrix forums have been doing, there might be some odd stuff going on with the formula for spotting that might need either fixing or improving despite the developer assurance that spotting is not as simple as just a 200m bubble around the unit.
Developer Response to Tim Stone Review
Thanks for posting the link, I think some of the points made in that review are fair while others are mistaken or exaggerated.
On the ‘brief pause that may occur during battle’ – this is over-enthusiastic CPU usage caused by the AI doing an extensive search for the best place to move a unit so it can fire at one of your units that it wants to destroy. A fix for this pause is in the works.
There are some issues with vehicle targeting at close range. There is an update up (or very soon to be up) on Steam that includes a tweak to vehicle accuracy to reduce streaks of missed shots at close range. Each ‘shot’ in TBF is aimed by creating a sort of mathematical aiming ‘cone’ in which the exact direction of the shot is distributed randomly. For many weapons the size of this cone narrows as the gunner corrects for previous misses. This is similar to how the targeting reticle shrinks and blooms in some first person shooters. There is a maximum accuracy a soldier can have based on their experience and stress, however, and this maximum needed to be allowed to narrow further for tank/AT gunners or the random distribution was still allowing streaks of misses to occur at improbably close range.
The “Steep Angle” issue when vehicles are at extremely close range is a different problem but also something I plan to address. This has to do with the gunner failing to differentiate between the parts of the enemy vehicle’s 3D model he can see versus those parts he can actually depress the gun low enough to hit. The gunner is aiming for a point on the hull front that is a good target at normal ranges – it provides the most wiggle room to still hit the target even with aiming errors and shot-to-shot variation from the weapon. But this spot can become problematic when they are muzzle to muzzle because the game engine puts real-world constraints on how far each vehicle can elevate and depress its gun. When they’re that close the target can end up too low to point the gun at and the code needs to account for this special case.
Vehicle path finding through very tight spaces is a bit of a double edged sword. The path finder works on a 1x1m grid (twice as detailed as the old CC engine) and vehicle models vary quite a bit in size. Currently vehicles are allowed to pass within a certain minimum distance to impassable terrain even if the 3D models for larger vehicles may clip into the impassable terrain a little. This is a trade-off to reduce the frustration of ‘Can’t go there’ order failures with vehicles in tight terrain. The alternative would introduce opposite cases where smaller vehicles would refuse to take paths they should be able to. That said, tanks should not end up inside buildings and that is an issue I’ll look into.
Spotting is based on a formula where range is a factor, but so is the action of each man in the unit, how many men have LOS to the unit, and how much obscuring foliage or smoke is in the way. There may be cases where the formula can be improved but there is no hard coded 200m detection bubble. Vehicles in the open are spotted well beyond 200m.
The AI does not know exactly what units you have or where they are. It does know the rough odds between the forces on the field – 3:1, 1:1, 1:2, etc. It uses this rough measure to make decisions about when to attack or defend. This information is available to the player via the force balance bar on the pre-battle briefing screen and you can gauge how much it has changed during a battle via observed losses and the force cohesion shown at the top of the screen. The AI will never take less units than it can so this ratio will never be far wrong from the player’s point of view. Yes, the player can deliberately take less units into a battle – there’s no reason to do this, but you can. In this case the AI will know its force is stronger than yours. This is an edge case that causes the AI to have info the player wouldn’t, but it can’t actually occur the other way around because the AI won’t ever take less units. The alternative would be to have the AI fail to react when a player takes an action that tells the game ‘I am deliberately putting myself at a disadvantage.’ From the player’s point of view the AI would look pretty foolish if the player took a single rifle unit into a battle against an entire enemy company and the AI just sat back and waited for the player to attack.
So in summary, I think this review levels some fair criticism but also jumps to some unwarranted conclusions. Ultimately whether you find the game to be fun or not is up to you.
Regardless of whether the AI could overall be better at offensive operations. One thing is for sure, it will still gladly give out a hammering to the American forces in Kasserine Pass if you aren’t careful. Even with some practice outside of the overarching grand campaign, it can still be one of the toughest scenarios.