Results 1 to 7 of 7

Thread: Most impactful improvements to the A.I

  
  1. #1

    Lightbulb Most impactful improvements to the A.I

    Hey guys! Thanks for putting all this hard work into this awesome mod! I finished the game 20 years ago and now doing another run with last 2 levels being on horizon for the New Game+ campaign using 2010 Alpha.

    It's a human trait and the game's nature to figure out the easiest and least unfavorable resolution for a situation which made me abuse the A.Is obvious weaknesses. I see the A.I abuse and deficiencies being the biggest flaws as it makes you play a very lazy, easy to win defensive game.
    The main gist of it should be tweaking the A.I to use bridges and destroy doors more effectively - I'll write only the possibly easy to implement ones.

    > A.I often does never try to get past a door.
    > Even if a door will be attacked then it's mostly by imps and a magic door can endure a lot of imp hits!
    > When A.I builds a bridge, they won't almost ever use it properly to attack unless a stray imp gets there - mostly the bridge just gets converted by my own idle imps and then sold which will give atleast another 10 minutes of peace.
    > Placing a lava trap guarantees your safety for that corridor indefinitely.
    > Levels with boulder traps are very easy to bait out and win with a single monster.
    > You can place 1 monster on the field which will be attacked and CTAd by all enemy monsters while you can possess someone with a strong auto attack and kill the heart without any interruptions.

    And of course even in 2010 Alpha, you can very easily bait an enemy monster to your ground, Disease it, wait a little for the disease to work and just ELECTRO DEUS them down easily due to CTA being often 1+ minute long. Strategies like these are what my lazy mind often ended up with as I was too lazy to have these long fights - rather just smash the gems until 100K gold and then bomb them down with no effort required.
    I believe following the philosophy that an average player typically dumps all monster when the heart is in vision is a good philosophy to apply when making improvements - maybe you can hold off A.I sending all troops at once like that?

    Other than all that I did feel challenged at some points and the game does feel marginally better than the original as my previous attempts for a rerun were cancelled due to poor QoL of the original version.

    Thanks again!
    Last edited by AggressumNatum; January 8th, 2020 at 18:41.

  2. #2

    Default Re: Most impactful improvements to the A.I

    Welcome, and thanks for the feedback.

    I'm fully aware by the many limitations of the A.I. Unfortunately, since Mefisto left there's nobody with both his skill and commitment working on the game. You mentioned playing r2010, so that makes you aware I already tried to do improvements where I can. He builds better rooms now, he uses some more spells. And he can actually cast 'destroy walls', although I think the situations where he would use it are so rare there's a good chance nobody will ever see it. But the AI behavior is sadly far from simple, and even obvious improvements might never make it into the game.

    What my strategy instead is, is to give mapmakers more tools to make better maps around the existing limitations. Have you seen this campaign, where Kix actually used many new features to create a very hard but (for what I've played and read) very good campaign. Give that a try if you are left wanting more after you completed your current campaign.

    And yes, that means that to make a good map which challenges a player you simply shouldn't give him the possibility to wall in with gems indefinitely, and don't rely on just enemy keepers to make the map interesting. And yes, lava traps are mostly out, as are bodies of lava/water between the player and his enemy. I did make a very short map with just enemy keepers, you could try that as well.

  3. #3

    Default Re: Most impactful improvements to the A.I

    Quote Originally Posted by YourMaster View Post
    Welcome, and thanks for the feedback.
    What a swift response!

    I was expecting an answer of such. I guess those problems could also be solved in a similar fashion for the New Game Plus then. I don't know how popular that campaign is but I assume that most old schoolers would like / prefer it; as it has a strong nostalgia element to it and normal mode is a bit too easy.
    I noticed that transfer creatures were removed / locked away in NG+ - I'd rather give transfers back as they're very fun and remove boulders / lava traps and make pathways to the player's base by dirt, gold or just plain land.
    I also believe naming it "Hard Main Campaign" would encourage the selection of that for first timers as well.

    I'll try the other campaigns as well (:

  4. #4

    Default Re: Most impactful improvements to the A.I

    My goals was just that, for it to be played by people wanting to play the original campaign again, but not have it feel like a tutorial. I wanted to stay close to the original campaign though.
    Most transfers I hid behind secret objectives, mostly because most maps if you start with a transferred creature you can just cheese the level straight away to win.

    I don't think the original campaign lends itself all that well to a defensive play style. Mostly, because you're never really attacked and you can't win by sitting in your dungeon. You have to go attack to actually win. And the maps where you can just sit and wait forever until all your creatures are level 10, you could have easily won when your creatures were level 6 as well. But which maps do you think are too obviously exploitable by a lava trap or something?
    Level 10 I think is actually pretty ok now, I think blue attacks pretty soon and you'd better be ready for it.

    The campaign is a few years old now, perhaps I'll revisit it sometimes to see where I can tweak it. I was considering changing up some of the still untouched secret levels, and perhaps make it so you can complete them only once for a permanent bonus.

  5. #5

    Default Re: Most impactful improvements to the A.I

    Quote Originally Posted by YourMaster View Post
    My goals was just that, for it to be played by people wanting to play the original campaign again, but not have it feel like a tutorial. I wanted to stay close to the original campaign though.
    Most transfers I hid behind secret objectives, mostly because most maps if you start with a transferred creature you can just cheese the level straight away to win.

    I don't think the original campaign lends itself all that well to a defensive play style. Mostly, because you're never really attacked and you can't win by sitting in your dungeon. You have to go attack to actually win. And the maps where you can just sit and wait forever until all your creatures are level 10, you could have easily won when your creatures were level 6 as well. But which maps do you think are too obviously exploitable by a lava trap or something?
    Level 10 I think is actually pretty ok now, I think blue attacks pretty soon and you'd better be ready for it.

    The campaign is a few years old now, perhaps I'll revisit it sometimes to see where I can tweak it. I was considering changing up some of the still untouched secret levels, and perhaps make it so you can complete them only once for a permanent bonus.
    I'd argue that transfer creature is a very unique feature of the game and removing that modifies the originality of the game far more than a bridge being replaced with dirt blocks. It also adds a macro dynamic to the game - I kept my strongest creature alive and/or tried to get at least one of them to level 10 before end of the game just in case there was an option to transfer. I also made sure that my imps had found every library and identified every magical box. I know that levels can be done in a speed run fashion with it but that's essentially cheating and you can already enable cheats anyways. If it still is a concern there could simply be magic doors or gold blocks which would stop you from mining to the enemy base to fast?

    On levels where there are gem stones there really isn't any challenge, even the last level was banalized by me through placing just 2 magic doors, letting the heroes pile up, and then gun them down with 3 lightning strikes - was unnecessary but I was lazy and it makes no difference with 20 vampires from which half were level 10. I think the most challenge was on the tower defence level and two others which had unexpected amount of hero waves.

    I read the patch notes of it right before I started playing it and I didn't feel almost any challenge, as it was the same as with the lava level before it - if a bridge appeared, you just convert it and sell it for 10 additional minutes of peace.

    I think it's worth to rebalance the New Game Plus mode and make it truly hard, not cheesy; so that not every obstacle would be met with the mindset of "Oh I guess I had to just sell Magic doors / mine gems longer and wait for my monsters to level up". The only level where I felt that I truly have to put in some strategy was the last level's keeper A.I as I was swarmed with Vampires who were too expensive with even gems to be knocked down. I resorted to isolating a few high level vampires, killed them with monsters, converted and placed them in Temple / Scavenger to make sure that the rest of them wouldn't be a problem either.

    Making secret levels have a bonus in campaigns is a good idea but that should also be communicated to the player some how then. I didn't play them as I didn't see a reason for it.

  6. #6

    Default Re: Most impactful improvements to the A.I

    You could play the two imp-slapping levels if you still have a save. They are quite different from the original and worth it just for the experience alone. The other secrets are kept original.
    But yes, the communicating it to players is the hard bit. The major advantage of the original and the NG+ campaign is that the texts are in all languages, which I don't all speak. Some of the hidden transfer level objectives I communicated by flashing the button as soon as you failed the objective, but many others are already very hard to know. I'm thinking instead of making the levels harder to complete each time you play them.

    But making it hard really wasn't my goal for this campaign. Some of the maps if you really want to make them hard, they would become almost unrecognizable. And in any case, the feel of the levels would be much different. I made it mostly for people who like the original campaign, and want to play it a second time without having their hand held all the time. So the most important changes for me are for example that you know have to research the barracks on the first level you get it, and that the early levels start counting as soon as the level start instead of when you build all the rooms. If people want to play hard levels, just play new levels. Like I said, that campaign Kix made is hard, the in included Undead keeper campaigns are also really good and really hard.

    That being said, I would love to be able to have the computer defend his bridges. Perhaps I can get that to work sometimes.

    On a side note, I never 'cheese' the levels with my transfer, but I don't really see how that's any more of a cheat then taking the opposite approach and do stuff like bridge selling or walling it with gems. The AI simply has so many limitations that it's almost impossible to make a map where it's impossible to exploit any of those weaknesses and if you do you end up cutting most rooms, spells and traps from the game. Stuff like Guard posts, Cave in, Lightning, Lava Trap, Prison, Graveyard, the list goes on and on.

  7. #7

    Default Re: Most impactful improvements to the A.I

    I second trying http://keeperklan.com/threads/7050-C...st-isle-(2019) this campaign its a very well thought out campaign. A little frustrating sometimes but really interesting. The campaign really makes you think and poses a real challenge for people who have mastered the original.
    Last edited by CSWinnall; January 10th, 2020 at 18:05.

Similar Threads

  1. KeeperFX bugs & improvements
    By Schizz in forum KeeperFX
    Replies: 1
    Last Post: February 17th, 2018, 13:19

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •