AI Behavior Controller vs regular controller

[crashfly]

So does anyone {maybe say the developer} know what the differences are between the "AI Behavior Controller" and the regular "Behavior Controller"?

It seems at this current point that they function the same without any functional differences.

[Eruannon]

AI Behavior Controller does not work like regular Behavior Controller.

Specifically - AI Behavior Controller has access to additional instruction category, called 'autobase'.

[Green Cat]

Adding the "for devs": please create a series of tutorial videos:) Heck, take it next level and add them in game as a puzzle in the challenge or maybe a new section.

Also, this will further help with people not only understanding this awsome game that MANY don't realise how awesome it is because they don't touch the behaviours, but also will help people to point out how to increase QoL or other elements, since we can simple just say stuff like "tutorial 3, phase 4, the bots (so on)" and like this, others can easily check out the said moment, to improve communication and suggestions.

PS, of topic: add trains <- station are just storage. Rails are just fixed tracks. Wagon to storage or reverse are just item jump. Aka only rails and train models are really needed. Otherwise, it's just bots who follow each other, with main "train" being a strong, non storage, bot who pulls the "wagon" bots but is still way faster then having normal bots running around. And the "wagon bots" can only move via following the train or the wagon in front, and have massive storage size (since no I slot)

[crashfly]

@Green Cat - go hijack your own topic

@Eruannon - and what does the "autobase" category do? details or it did not happen.

[Phoenix]

"Also, this will further help with people not only understanding this awsome game that MANY don't realise how awesome it is because they don't touch the behaviours"

I do use behaviours, but as little as possible because of the performance hit ...
If I can use "Internal" to get the same results, then it's my first option ...

My first playthrough I was thrilled because of behaviours, but when my plant got bigger, the game became slugish, because of me using behaviors as much as possible to manage the flow's ...

And my rig isn't a potatoe box:

24 cores @4.5GHZ, 128MB RAM, 3060ti and 6TB of Samsung NVME Storage.

[Eruannon]

Originally posted by crashfly:

@Green Cat - go hijack your own topic

@Eruannon - and what does the "autobase" category do? details or it did not happen.

Usually You'd expect 'autobase' category of instructions to contain instructions that are used for AUTOmatic BASE construction and use.

If You have access to the AI Behavior Controller, then the description and list of instructions will be shown in behawior editor, with autobase instructions being listed as available only on AIController.
https://steamcommunity.com/sharedfiles/filedetails/?id=3427733779

If You wish to reject that reality just because some people don't care about explaining in detail how it works and how it is used then frankly that's Your problem.

An no, I will not explain how those instructions work or how they are used, as frankly - I never cared enough to bother learning them. Personally I prefer to make my own autobase solution.

[crashfly]

whether you understand those extra instructions or not, there is no documentation anywhere on how to use them.

to me, the reality is that few people will use those "extra" instructions if only because people prefer to "roll their own" and the most important is that those instructions come too late in the game to be of decent usage. too late in the game and no one knowing how to use them, makes them mostly useless for the game.

[Eruannon]

there is a difference in skill required between playing around with dedicated instruction for a task and creating a full codebase from ground up with signal encoding and multi-group coordination.

The fact that I don't know how to use autobase category instructions efficiently doesn't mean no one knows how to use it, and for many players, especially casual players those instructions can be much more useful, and convenient, than to someone whose idea of 'simple deconstructor' includes dynamic damage control, intel coordination with military units, automatic hardware replacement and upgrades all at faction-global range.


To me encoding and decoding any data I need into single signal is issue of few subroutines I prepared beforehand. To me creating a module that will upgrade to preferred weapon is literally 2 instructions that I prepared beforehand.

To others it may be hours of implementation and testing, and that is but one of elements of my automatic faction codebase, and that is if they even know how to begin working on such code. Learning curve difference between my implementation and use of AI Controllers' instructions is frankly staggering.



While You are right that documentation would be very welcome, calling them 'useless' is a bit overstatement. Useless for You and Your use case, or useless for me and my use case - that I agree. Useless in general - not really.

[crashfly]

might be an overstatement, but that does not make it false.

you do not use those functions because as you said you "roll your own".
additionally, you cannot explain how to use them as there is no documentation {please point it out if there is}
i am not going to waste my time with programming that has no documentation.
whose to say those functions won't change in the next update.

so until there is documentation from anyone {or even a rudimentary example}, the use case is non-existent and most certainly useless.

[Eruannon]

Overstatement is a statement which in this case - goes beyond reality, making it false. Just because two people who gave enough of a damn to take part of discussion here are not using it, or don't know how to use it (in my case - because I don't need/want to learn how to ride a bicycle when working on motorcycle, mostly because when bicycle was introduced I was half-way through working on motorcycle already; You - because You don't give a damn to learn it) doesn't mean it has no use case. It only means that there is no use case FOR OUR PURPOSES (or if we want to be a bit more correct - for MY purposes. You are unable to define whether or not those instructions have any use case for You, as You simply decided to go with 'no documentation = it must not have a use case').

You're of course free to not 'waste time with programming that has no documentation', by all means, but calling its use case 'non-existent' or 'useless' just because You didn't bother playing with it, is fallacious statement [Specifically - argument from personal incredulity).

In your case it is issue of YOU don't know how to use it (fine), You refuse to learn it, as not to waste time (fine), and because of that You call it 'most certainly useless and without use case' (false).

[crashfly]

you are just arguing the point for me. you have no valid use for the instructions. therefore they are useless.

give me a single use case. i dare you.

[Eruannon]

Use case:
Creation of automatic base components, to support manual player input.

Your argument is equivalent to:
"I and one whom I talk to have no single use for Python, I refuse to learn it, he doesn't know it because it is not used in what he is working on, as he's working in assembler, therefore Python is useless"

It would be ridiculous to humour it as anything but argument from personal incredulity, which is why I originally corrected Your overgeneralisation in statement that was equivalent to 'Python is useless for US (i.e. You and me), not in general'. You however decided to double down.

You originally wanted to know the difference between AI Controller and Normal Controller.
The difference lies in AI controllers' additional instructions.

Difference exists, and to some players the difference may be crucial in their implementation of automatic base solutions. To put it in other terms - I don't need trigonometric function instructions to implement convertion between radial and cartesian coordinates (I can use approximations using e.g. Taylor series to get accuracy I need/want), it doesn't mean that sin(x) instruction in programming language is useless. Some people might find use in it.


TL:DR Frankly if You can't comprehend that some people may find different instructions useful in their code, and it doesn't mean that instructions You don't use are 'useless', then I don't think further exchange from my side is warranted here.

[Green Cat]

Wtf is with this discussion?

The devs made it clear over and over again, you don't need them to finish the game and many people showed this.

The whole behaviour is QoL stuff. If you know what you are doing, it improves the automation of the base, but in no way is it a necessity. On the contrary, it will come down to preference, like I love programming the bots and buildings, but in the same time I did all this, someone else progressed the story and finished the game... except I don't care about that, because I love programming the bots so me playing is that.

So this is the advantage and disadvantage: you want to program them because you love that or you just want to finish the story


So, examples of what makes them so time consuming:

1: Create and army of bots, who, when shield is lower then X, said bot will go back, and the 2nd row will take over. So once the bot's shied is 100%, it goes back in line and then this repeats. And once all enemies are gone, it will continue patrolling the area until found new targets. If new target is to strong, will come back to base and give notice so I need to make adjusters, aka manual input for the 1st time. And like this, 0 damage <- someone not using behaviour: simple crafts more and more fighting bots without caring, and just manually move them.

2: Automatic constructions. I will give 4 situation.

A. Walls. Construct on x and y coordinates. Then add +1 x coordinators , repeat. And of course, for the other site, and similarly but for the Y. And a braking when hitting corner. Like this, if I press start: all walls are build without me doing anything. If enemy destroyed a wall, this will once again rebuild it.

B. On the same idea of increase and calculated placement (yes, a lot of fiddling around), once I activate, it will automatically create the building with network power range, correctly places once to each other. Like this, I can enjoy a symmetrically correct base that I can increase at any time, because the walls will serve as a "stop". Meaning if I ever extend my base, build a new wall, remove the old one, I can restart this behaviour, and will expand my network without me doing anything.

C. Same ideas for turrets, and I do mean multiple of type of them. As long as I create a good one, I can fill out all the walls without me needing to place any of them

D. Infinite resource. At one point you can regenerate resources, so using behaviour will ensure you are mining until the value drops, stop, wait until fully regenerated. Or simple create a Mining ore needs to be more then X amount.

3: Battle units, each turret against the correct enemy. I think this is self explained, and sure enough, you can argue this is possible without use of behaviour, and you are correct. So what I am talking about is how the units move. Instead of everyone going together because of radar..... even if they will be surrounded by non target enemies because the target enemy is behind them, using the behaviour control, you can set it so that the bots go in group, and reorganise and change format based on what target are closes to them. And yes, it's more time consuming then if you just where to send a legion of bots and don't care if they get destroyed. So again, it's each one's preference

If none of this matter for you, it's clear that you have 0 interest in the behaviours. So it doesn't matter what advantages it has because your question is not about benefits but about preferences (A good example is someone putting music for relax, they say it makes them feel relax, but you don't feel relaxed (don't get benefit) and dismiss the other person exactly because the music they like, is not something that you like = it will never make you feel relax)

[Eruannon]

Greencat, while I agree with the main argument of Yours (one highlighted especially in the last paragraph), I believe crashfly's core argument does not apply to behaviors as whole (or at least I never got a feeling it does), but specifically a subset of behavior instructions, specifically those introduced by 'AI Behavior Controller', and listed in 'autobase' category.

Part of his argument is that as neither I nor he knows how to make them work [why we didn't learn them is for slightly different reason], and neither of us have a use for them - therefore they are useless.

Whether the same argument could be used for other standard behavior controller instructions ['mine' comes to mind as one of best examples] is beside the point ;)

I'd go with 'every instruction has a use case, but not every instruction is useful for everyone. It doesn't mean the instruction is useless' ;)