Please wait !

[Glubokii Boy]

Hello.

I'm very much looking forward to the V4.0 update with it's new features but the more i think about some of these features that have been mentioned i get a bit concerned about how to be able to use them effectivly.

I'm primarely thinking about the:

- AI withdraw order

and

- AI area target order

What i feel will be very much needed to use these new AI options in a good way are an update to the timing options when it comes to the AI programing. What is needed i belive is something like a WAIT-option.

The EXIT BETWEEN option that we have now will be difficult to use when combined with triggers. This option will work just fine i guess when the AI movement uses simply a set time table for its orders but when

combined with triggered movements it will be impossible to know at what time those orders will be activated.

This will make it very difficult to program a nice shoot-and-scoot move for example.

If you want a vehichle to move into possition on a crest or come around a building and then remain in place for a set period of time (ex. 2 minutes) before withdrawing again this will be very difficult to do if the move into

possition is triggered by a terrain objective. The designer will not know when this will happen.

This problem will also exist if the designer would like an AI group to areafire at a specific location for a set number of minutes when combined with triggered movements. No way of knowing when this move will happen.

Having a choise to kick off the next order either with the exit between option or a new WAIT order would be a very nice thing to have. I feel that a WAIT option would simplyfy things when it comes to AI-programing not

only when trying to do the two examples that i have mentioned above but also in some other circumstances.

When using the WAIT option the AI group will remain in place at the order location doing what it is ordered to do until the set number of minutes have expired and then move on to the next one.

[Bulletpoint]

I agree. The withdrawal feature will be a great addition, but area fire will be difficult to use well with the current order options.

Never understood why the AI is based on "exit between" orders anyway. It gives of course a way to build in some unpredictability in the orders, so you don't know if the right flank attack will begin at T+5, T+10, or T+15 minutes, but it's extremely cumbersome to program the AI this way, compared to a simple "Wait here for x minutes" order. The whole AI programming interface could do with a reworking to make it more flexible and powerful, while still keeping things simple.

Aside from big reworkings, this game has so many "low hanging design fruits" that could be picked to improve the game with hardly any effort. Just a little thing like being able to activate a reinforcement by a trigger would open up so many possibilities for scenario design, and it would be fast and easy to implement.

 

[Pak40]

I could be wrong but I don't think EXIT BETWEEN means that there is randomness built into the order. Rather I think it means that the order will be executed within the time frame specified as soon as the unit has satisfied the previous order.

[nik mond]

 

1 hour ago, Pak40 said:

I could be wrong but I don't think EXIT BETWEEN means that there is randomness built into the order. Rather I think it means that the order will be executed within the time frame specified as soon as the unit has satisfied the previous order.

That is correct. It is a no exit before x:xx(min time) providing the conditions of the trigger are met. Once the condition is triggered, and x:xx(min time) has been reached then the unit moves. If the conditions of the present trigger is still not met by y:yy(max time) then the script is activated anyways because the max time condition is reached, meaning the unit will execute its move at that point. Triggers are linear, the previous script must be executed before the next trigger is available in all cases. 

[Glubokii Boy]

Thanks for describing how this works guys. The problem with the timing of the AI moves still remains though as far as i can understand.Take this example...You as the scenariodesigner wants to do this:

The player will be commanding the allies. The german AI have a pair of StuGs located behind a hill in ambush possition.

You make a terrain objective with a trigger function - trigger player armour - that will have the AI group containing the two StuGs advance up the hill to a hull-down possition against the triggerzone.

All good and well...The StuGs will do that ones the player moves any armoured unit into the terrain objective (or when the specified exit before time is met) . Now comes the problem though...

As the desiger i want those StuGs to remain in place on that hill only for a short period of time (lets say 2 minutes) before backing down again behind cover. I have set the initial AI order to move into the hulldown possition like this:

 

- The move order is activated by the trigger (terrain objective).

- The exit between is set to 00.05 and 00.15 (between the fifth  and fifteenth minute).

 

This means that the StuGs will advance into possition any time after the fifth minute as soon as the player moves some kind of armour into the terrain objective. If the player decides to not advance any armour into this location the StuGs will still move into possition on the hill and minute 15.

What kind of timing do i set for the withdraw order to make sure the StuGs only remain in place at the hull-down possition for my intended 2 minutes.

I have to set the exit after to 00.08 minutes i guess if the move into possition takes one minute to allow for the fact that the player may move tanks into the triggerzone at the earliest possible time (5th minute).

How about the exit before time ?

I can not set this to exit before 00.11 for example. It would work fine if the player in fact did move tanks into the terrain objective at the first possible time (5th minute) but what if he did not do that ?

What if the player decides to not move his tanks into that possition until the 12th minute. In this case the AI StuGs will not be in place at the hull-down possition until the 13th minute.

Having the timing set for the withdraw order to exit between 00.08 and 00.11 will not work in this case if i want the 2 AI StuGs to remain in possition for 2 minutes before withdrawing since they are not in place until minute 13.

This is where i feel a WAIT-option would come in handy. Simply set the withdraw order to WAIT 2 minutes. With such an option the StuGs will remain in place for my desired 2 minutes regardless of when they advance into possition on the hill.

[Bulletpoint]

12 hours ago, Pak40 said:

I could be wrong but I don't think EXIT BETWEEN means that there is randomness built into the order. Rather I think it means that the order will be executed within the time frame specified as soon as the unit has satisfied the previous order.

This is how the manual describes it, but I doubt it's correct.. at least the behaviour I saw when I played around with various settings was that "Exit between 00:00 and 10:00" simply means "move out at random time between 00:00 and 10:00".

Maybe there's a hidden logic behind it. But giving a  unit more time to complete an order doesn't seem to make it move differently. It simply seems to chose a random time to start running at full speed towards the target zone.

Would be nice if giving a unit 30 minutes to cover a small distance would mean that it made more breaks along the way, stopping where there is cover. But I don't see it working that way currently.

[Bulletpoint]

Double post - the forum software gave an error message when I posted it and I found two copies had been posted.

Edited November 8, 2016 by Bulletpoint

[Glubokii Boy]

46 minutes ago, Bulletpoint said:

simply means "move out at random time between 00:00 and 10:00".

 

 

 

I think  this is correct...

As long as you do not include a trigger in the order thats the way it works...That have been my experience also atleast.

But If you do include a trigger then the AI will indeed wait until the exit after time until the group moves out unless the trigger-conditions have been met earlier.

 

 

 

[Glubokii Boy]

1 hour ago, Bulletpoint said:

 

Would be nice if giving a unit 30 minutes to cover a small distance would mean that it made more breaks along the way, stopping where there is cover. But I don't see it working that way currently.

To a certain degree i guess you can tweak this by using the other settings...Changing between DASH and ASSULT for example and also between CAUSIUS and ACTIVE will atleast have some effect on the way the troops move about on the map...

Perhaps not exactelly what you are looking for but it should make some difference atleast..

[Bulletpoint]

18 minutes ago, RepsolCBR said:

To a certain degree i guess you can tweak this by using the other settings...Changing between DASH and ASSULT for example and also between CAUSIUS and ACTIVE will atleast have some effect on the way the troops move about on the map...

Perhaps not exactelly what you are looking for but it should make some difference atleast..

I have been experimenting with this too, and it seems these settings only determine the pattern of when sub-units will move, not how each unit actually moves. For example, setting move type to "assault" for a platoon will make individual squads conduct something a bit like bounding overwatch, but each team when it moves will still just plot a long "quick" movement to the next spot.

Each leap will still be way too long and teams won't stop at natural points of cover. They'll happily run 10 metres past a low wall for example, to stop in open ground.

It would be cool if each team had basic AI to think "Are we getting close to the end of our movement path now? Are we currently in a spot that offers cover or concealment? If so, let's just stop a bit early instead of running the last few metres".

This would not need a big complicated overall battle plan AI, just some simple rules to represent common sense for the individual team.

Edited November 8, 2016 by Bulletpoint

[Glubokii Boy]

That would no doubt be a very nice thing to have but i think that programing something like that is way more complicated then one might think.Having each individual AI team perform more human-like would require massive calculations i'm sure.

Taking into acount the elevations and terrain types both at a distance and in their immidiate vicinity to determin the LOS/LOF situation in all directions.The pixel troops would also need to realice what kind of threat they are facing...long range fire power ? short range firepower ? HE ? and be able to act accordingly.

Such a simple thing for example as deciding wich is better... to take cover behind a wall or move forward in a ditch will most likely depend on many, many things.

I think that everyone would like to see a smarter AI but i don't think that it will be a very easy thing to get there.IMO the best thing we can do is to help the AI perform better. Very high up on my wishlist for future updates to the game-engine comes

Additional AI groups

and perhaps also

Additional orderslots for each AI group

IIRC we now have 16 AI groups each with 16 orderslots. For smaller scenarios this is most often enough but when the scenario-size begins to be something like battalion minus these 16 AI-groups could easily become to few.

If these AI-groups could be increased to something like 32 i think this will go a long way in allowing a scenario desiger to 'help' the AI perform way better.

With 16 AI-groups and a battalion sized force the AI groups needs to be rather large...more or less platoon-sized. Although it is possible to get things to work quite well even with larger AI-groups they are a bit tricky to handle.

For example if you set an AI platoon (with 3 squads + HQ and 1 or two supporting HMGs/LMGs) to advance down a street in a village/town using maybe 2 or 3 waypoints (orders) a few things usually happens (atleast when i try to do this...)

- If i want my platoon to advance on both sides of the street they will usually end up doing a very stupid thing. At complete random the squads will change wich side of the street they adcance on and decide to cross the open street in full wiew of enemy fire.

This is not what i want. The teams that are set to advance down the left side of the street should advance down the left side of the street and the teams on the right should advance down the right side.

- I never really know where the HQ will end up. That team picks one of the avaliable spots at the next order location the same way as the other teams do. At complete random it seems.

- My supporting Machineguns acts pretty much like any other team. It does not neccesarily pick the best supporting option of the avaliable orderspots. Neither does it remain behind the advancing squads for any longer period of time to support them. They are treated like a regular squad and moves forward in the same way as the others.

With more AI groups i as a designer could split up this platoon sized AI group into smaller sections and thereby significantelly help the AI perform better.

example.

- two squads are set to only move down the left side of the street.

- one squad is set to only move down the right side of the street.

No more criss-crossing of the street right infront of enemy firepower...

- My HQ team is a seperate unit and could be set to advance where and when i want it.

- Same with my machineguns...

Many might say that an idea like this will make scenariodesigning even more complicated and time consuming. I don't agree.

I think that with more AI-groups avaliable the designer will have a far greater chans of getting the AI to perform the way he wants with his first try. Significantelly cutting down on the tweaking and testing over and over that is usally neede now to get the larger AI-groups to perform as intended.

Tweaking the AI, testing, tweaking some more, testing again....more tweaking etc, etc is what is taking the most time. My oppinion atleast.

I think that more AI groups is the best way BF can move forward when it comes to improving the AI

the next best thing would be more trigger options...

[Bulletpoint]

1 hour ago, RepsolCBR said:

Having each individual AI team perform more human-like would require massive calculations i'm sure.

Taking into acount the elevations and terrain types both at a distance and in their immidiate vicinity to determin the LOS/LOF situation in all directions.The pixel troops would also need to realice what kind of threat they are facing...long range fire power ? short range firepower ? HE ? and be able to act accordingly.

The whole point was that it would not need massive calculations, because the underlying logic would be simple: If movement line intersects a place of cover, stop moving (until next movement leg).

It wouldn't give perfect results in all cases, but in the majority of cases it would be much better than just stopping at random points.

[Bulletpoint]

1 hour ago, RepsolCBR said:

- I never really know where the HQ will end up. That team picks one of the avaliable spots at the next order location the same way as the other teams do. At complete random it seems.

Yes, this is a major problem in the way the game works currently. I get around it when I design scenarios by making the HQ unit his own AI group. That allows me to place the HQ in a specific placs (or small area), and then paint a much bigger area around it for the rest of the platoon.

Another related issue is that AT guns often have a separate team of ammo carriers. So if I want to make it random what spot the gun deploys to, I end up with the gun choosing one of the available spots and the ammo carriers choose another place far away from there. The only way to solve it is to make a whole separate AI plan for each spot I want the gun to be able to deploy to. Which limits how flexible I can be with the rest of the AI.