Any good resources on turnA based AI? I think most difficult would be 4X AI, since it would have to manage a lot of different aspects. Also competent wargame AI. Also /vst/ gamedev thread.
Gamedev thread
Any good resources on turnA based AI? I think most difficult would be 4X AI, since it would have to manage a lot of different aspects. Also competent wargame AI. Also /vst/ gamedev thread.
Here's analysis of MoO1 AI
https://docs.google.com/document/d/1zDVpOyQmr5AdIlZIXBY8Y5Ij8oOrDYzVmyA2jB7vswk/edit
very nice, ty anon
might want to take all the stuff here with a grain of salt since that game was meant to run on 1996 hardware. you can put together an AI using "greedy" solutions that, though heavier on the performance, tend to be better overall as well as much faster to put together and easier to maintain
What happned to that one vst poster who was making a mesoamerican city builder? He even had a demo you could play but then no more threads on it
i've done game dev professionally and as a hobby (in fact i'm making like a 1v1 or 2v2 combat-focused 4x-styled game right now)
i don't really have any resources to offer but i'd recommend not relying on behavior trees, because they will quickly grow into incomprehensible messes, especially when you start adding complexity. instead, for the most part, keep your behavior trees short and use bunch of score-incentive styled approaches. essentially: evaluating a bunch of actions, giving each of them a score based on how urgent/useful/cheap they are, and then executing the actions with the highest scores. for something like a 4x, you'd create a bunch of these systems for various different purposes like diplomacy, combat, acquiring resources, building improvements, researching and so on, and have them either interface directly, or preferably, let each of these components guide the AI without direct scripting. for example
>warmonger AI in 4x
>is given a hardcoded incentive to build extra units, and will thus build extra units
>thus, the AI will have more units in general, which increases its incentive for starting wars (possess bigger military than enemy -> add score to war action)
>the AI will suffer from high upkeep from its units, starts to value money more (low money/income -> add score incentive to acquiring money)
>thus, it is easier to bribe the warmonger AI into peace, and the warmonger AI will start to look for different targets
that's basically how it'd go.
This sounds good. Right now I'm gathering any potential algorithms that could assist in this. The main difficulty is of course making an AI that's fun to play against, not necessarily an extremely competent one. Your description is pretty good at describing that.
Also some info can be gathered from open sourced Star Ruler 2 code
https://github.com/BlindMindStudios/StarRuler2-Source
And Remnants of the Precursors code
https://github.com/rayfowler/rotp-public
https://github.com/Xilmi/rotp-coder
Has anyone been able to solve the dilemma of static production?
E.g. in CK3 you can build a "Market Village" that will produce 0.7 gold every month, but it will cost you 75 gold. The game itself last 7032, months, therefore, there is no longer point in building them after mid-way of the game, because half of their potential gold has been lost.
I think there are two solutions, either make the cost of buildings decline over time or get rid of static value products that can either become prosperous and produce 7 gold per month to shit and produce 0 gold a month.
This is probably a bizarre dilemma, but it is something I keep coming back to for some reason.
Don't make buildings do only "static production".
I don't understand why that's a problem. Whether or not its a good design is a big "it depends", but the basic choice of whether to make an investment that takes time to pay for itself is pretty much strategy 101.
If you haven't written a minimax tree to play connect 4 or something that's a good starting point. No modern game would actually use one, but it'll give you all the theoretical context on how to think about gamestate in a way that's useful to AI. Specifically the heuristic function, which
is describing.
You could just scale the value with time.
Or research
My advice is to stop looking up to shittily designed strategy in name only games like parashit. If a campaign is short then it's a feature, not a problem, i.e. there's a strategy to rush market to make as much money out of it as possible. If the campaign is long then just make investment return time significantly shorter, which is pretty easy when you actually care and think about the game you're making. If you are worried about market becoming a mindless choice, make it an opprtunity cost like in Civ4: the player decides if it's better to get extra money now by building the market, or to build a unit for deterrence of rival civs.
>My advice is to stop looking up to shittily designed strategy in name only games like parashit.
Good to hear that I'm not the only one who sees how awful they are designed. It is so tiresome how the shills will defend every half-assed mechanic to the teeth.
75g with 0.7g per month income means that payback time is just 100 months. What's the problem?
Games where building every possible building everywhere doesn't makes sense are actually better than the ones where you just spam all buildings
>75g with 0.7g per month income means that payback time is just 100 months. What's the problem?
I once read the economist theory which stated that buy price should be at least twenty times the annual net value of the enterprise. Which would be twenty years or 240 months.
Issue as I laid, is the time the building is constructed, if it is constructed mid-game it can only produce half of its potential profit, thus only time to build it is during the early game.
>if it is constructed mid-game it can only produce half of its potential profit, thus only time to build it is during the early game.
What? Even if you construct building when only 10% of game time remains, it will still be profitable
yes but maximum potential has been lost?
Why care about that?
I because I think it is boring
how do you get out of bed in the morning knowing you did not make optimal decisions in life 5 years ago?
sounds interesting. Where does faction 1 get their gear from? Depending how your action economy goes (eg troops can only shoot once per turn) spamming zombies seems like it'd always win if they can drag the game out long enough.
>sounds interesting. Where does faction 1 get their gear from?
it depends. right now if you buy your faction's own stuff then it costs the "Gear" resource, which is a resource every faction has but each one uses it a bit differently. there's also a "Money" resource that is used to buy non-faction weapons and units in the marketplace, but those tend to be pretty expensive, so you can only buy a few of them every now and then (basically useful for situational combos, or to alleviate a current tactical issue)
the faction i was talking about has a special gimmick where they excel at amassing the Gear resource, so every one of their cheap, spammable rifleman unit can potentially turn into a powerhouse if you stack them with enough high-grade gear.
>Depending how your action economy goes (eg troops can only shoot once per turn) spamming zombies seems like it'd always win if they can drag the game out long enough.
right now i'm planning on the zombies mostly being a nuisance, like they are slow and will slowly wither away on their own, but they will also spawn at a steady stream from their spawning pits and and force the other player to split their forces to hold their territory. just having a unit or two to defend should be enough to keep them at bay, but they are cheap for the zombie faction to spam. it's basically like an attrition/division tool (until the mid-lategame, at which point they can be made more formidable)
I would recommend you add something like a Carnage meter for non important missions.
So if the player is able to kill a large amount of enemies quickly enough he can just press a button to accept a win instead of dragging it out.
one of the chief reasons i moved away from the tactics stuff is because i hate xcom and how that game drags on because of shit like that. i have never finished a campaign of xcom because there's always like 2 busywork missions for every hard one, and even the hard ones mostly feel like busywork.
this game's going to be heavily focused on multiplayer because that's something i can understand a lot better as a designer
>So I guess the zones you control is what gives you the resources? Having difference resources that all the factions have access to but use differently sounds like a good setup, since there's then different strategic importance to different zones for each faction.
i was hoping for asymmetry to be a big deal overall. some factions will have completely unique resources, while others may not have access to a certain resource at all. i really liked how endless legend approached its faction design so i was hoping to lift some of that stuff
>Having like a xcom-like overwatch might help
overwatch is 100% going to be in the game, hopefully to dissuade constant, wide swarming, thus shortening turns and condensing the action to more concentrated pushes and such. basically you should have 2x the units the enemy has for a specific area if you intend to assault it, so a lot of the strategic gameplay comes from guessing/scouting where your enemy is deploying his main force, and where his defenses are weak, and so on.
i'm a little worried about turn times since it's not going to be simultaneous turns like a lot of turn-based games. as silly as it might sound, i think it'd be better for the health of the game if half of your 20 unit force is on constant overwatch, just to keep turns short. the marketplace is supposed to help a bit with this too, since at least you can browse the shop while your opponent is making his plays.
>how that game drags on because of shit like that
i should say, how that game drags on because it doesn't have any time-saving features like that carnage meter thing you mentioned
So I guess the zones you control is what gives you the resources? Having difference resources that all the factions have access to but use differently sounds like a good setup, since there's then different strategic importance to different zones for each faction.
yeah considering you would still need to order your territory guards to shoot the zombies it seems like there's a danger of it becoming tedious. Having like a xcom-like overwatch might help, since then they'd just shoot the Zs without you needing to order them to.
you build buildings not to get return on investment, you build them to increase velocity of your economy, to have higher income and being able to support more expenses.
testing out some scuffed territory mechanics
That actually looks nice anon. I'm sure it is the fruit of many months of sweat.
it's based on an old game project i posted about here a while ago. the original game project was a fairly typical xcom styled tactics, but with a lot of silly weapons and abilities.
this new game uses the same visuals and some old logic as a base, but i redesigned basically everything else and added netcode. it's supposed to be more like a turn-based strategy game with tactics elements, like an unholy mix of elements lifted from 4x, RTS and tactics. i'm not sure it'll be any good but at least i'm excited enough to work on the game again. some of the old vestigial design elements from the other games, like equippable weapons and items/abilities will hopefully make it kinda unique as a strategy game. one faction will rely heavily on gearing their basic units with different types of weapons and armor in order to deal with various different threats. another faction can build spawning pits and spam the opponent with disposable zombies from all sides, with scientists speeding up tech tree progress as well as doing mind control and resurrection as a support mechanic. a third faction concept will heavily rely on stealing from the enemy and using the in-game marketplace to build powerful unit/weapon/ability combos while occasionally scamming the opponent with NFTs or some shit.
but anyway this is all a long ways ahead. hopefully it won't suck
>some of the old vestigial design elements from the other games
from the other game i mean, the one i made originally
Seems familiar
Have you guys had any luck in bumming Patreon money? I guess you need a steam page to make that work.
Any thoughts on avoiding the mid/lategame issue where you've obtained a dominating advantage and have won in all but name but haven't ended the game yet so the rest of the game becomes a bunch of tedious busywork to finish it out?
My other question is, it seems like a common recommendation for making strategy games is to go the enginedev route, but what issues would occur with making a strategy game in premade engines (such as Unreal, Unity, or Godot)?
i assume you mean mostly free-for-all 4x games? more competitive 1v1 or 2v2 RTS games tend to end in surrender when winning becomes impossible for one party.
off the top of my head, in something like a 4x game, domination victory could work by holding 50% of the player capitals for a number of turns. that way you only need to show that you are capable of winning the game and just chill on defense for some time to close it out
>My other question is, it seems like a common recommendation for making strategy games is to go the enginedev route, but what issues would occur with making a strategy game in premade engines (such as Unreal, Unity, or Godot)?
not a huge expert on unreal, but it seemed like the engine really wants you to make shooter, action or roleplaying game with it. everything else seems to demand a lot of extra work. it can feasibly work, but it's also pretty slow to develop on because c++ is kinda fricky. theoretically it can perform better than unity, but blueprints aren't performant at all and it's easier to write optimized C# than optimized C++, so that probably won't realize easily.
unity is a lot more receptive to weirder ideas and has a frickton of quality of life features for development that unreal doesn't, like the editor integration, easy drag and drop, easier debugging, error handling and other shit like coroutines. everything is built around gameobjects and components, which makes development easy and fast, though sometimes messy and not so performant. even though unity is built around slow as frick C#, i don't think performance is necessarily an issue. this small RTS game called Line War has like 300 units on screen and it runs fine, though it is pretty simplistic. i think you'd mostly notice the performance if you start stacking a lot of expensive AI scripting or put in a billion tiny gameobjects or something like that (hence why endless legend chugs like hell in the lategame)
>My other question is...
Optimization and code control, basically. By their nature, premade engines just aren't as optimized as a dedicated engine can be. Usually not much of a problem, but for example, if the premade doesn't handle entities in the right way, giving them a lot of unnecessary variables or whatever, it can lead to massive headaches if optimization does become an issue. Then of course, because it's a prebuilt engine, you might not even be able to fix these issues on the "correct" layer, instead relying on hacks and workarounds, if it gets addressed at all.
We can get more specific, like the way Godot handles lighting (a max of 8 sources or so, built in to the engine) might be completely horrible for what you're doing, but that's the short of it.
A premade engine trades ease-of-use and (ideally) development speed for... basically everything on the technical level.
snowballing in long strategy games is a bit of a cursed problem. There's a contradiction in design goals, you want the player decisions to matter and give advantages and you want to have this long epic game with interesting decisions to make the whole way through. But early game advantages results in this compounding advantage effect where they end up giving a player a runaway lead.
The solution suggested here is to essentially shorten the game by making the victory conditions more achievable. Its fine, but it takes away from the "epic long game" feeling. But that's why its a cursed problem, you can't really do much about it without compromising one or the other conflicting design goals.
As an example of what another solution might look like, you could introduce some catchup mechanics. If a player is in a losing position later in the game perhaps you get some special unit or resource bonuses. While this could make lategames more competitive, it also means that your early game decisions matter less because even if you play badly the game bails you out later.
Only solution is End Game crises. Since reducing player power arbitrarily is gay your only course of action is to make shit more difficult. If you don't want any Sci-Fi BS you could just do something like Real Divided and just unify the AI against the player.
Make people win before that happens.
See what Emperor of the Fading Suns did, for example.
Personally, I like to think that the difficulty should scale according to empire size. In shit PDX games the games are designed to become easier and easier the bigger you blob.
If you look at empires like Achmenids and Ottomans you see their expansion quickly stagnates.
Imperator came close of depicting this, e.g. in order to prevent revolt risks from rising you had to assign troops to governors, naturally, they botched this by not making it impactful enough and punishing you for it by making the governor immediately rebel, and they ended up removing that feature with 2.0.
PDX games in general textbooks how to not design anything.
>My other question is, it seems like a common recommendation for making strategy games is to go the enginedev route, but what issues would occur with making a strategy game in premade engines (such as Unreal, Unity, or Godot)?
Two issues with them.
1. They're made for action paradigm games. For an RTS, or worse yet, a GS, you're gonna be doing a lot of work adapting the engine to the game you're making anyways - so why not just make your own and not be stuck working with their ideas of how C++ "should be" written?
2. They're often not well optimized for CPU-heavy games. This has been improving somewhat in recent years, but honestly they still fricking suck at it. This ties back to the whole action game thing. In those games, CPU work is pretty minimal and it's mostly all about rendering - so the engines don't prioritize doing heavy CPU number crunching.
The one thing that premade engines do bring to the table that's pretty useful is rendering. Writing your own renderer can be a lot of work, and they save you from that. But for every other aspect of a game you're basically going to be doing all the work yourself anyways.
Assuming a Civ-ish game like other anons have, I have a wacky, probably unworkable idea.
Basically, you'd have a bunch of fairly powerful bonuses that you can get, but only through forming a breakaway civilisation from your original civilisation.
So say you start as Babylon. You play the standard 4x early-game and dominate your local area. Then, if a rebellion happens in one of your cities, and you get the option to play as the rebellion. This turns you into (say) Assyria, which gives you the Assyrian bonuses on top of the Babylonian bonuses, but you have to fight off your old Babylonian Empire (now controlled by the AI). Or a barbarian tribe on your borders starts to organise and you can then play as them(so Babylon>Persia)
The idea is that choosing to "start over" as a smaller power should provide powerful enough bonuses to offset the loss of territory, and should also provide a rubberbanding effect where a powerful and successful empire would have less opportunity to reform into something new.
Idk, it just tickles my history autism that there could be a 4x game where a Rome Player could choose to undergo the Triumvirate Wars because it would make their civilisation stronger (if none of the other players manage to intervene)
i don't think that's necessarily a bad idea but it would probably be a little difficult to implement in a typical 4x game where minmaxing your civ's numbers is basically 90% of the gameplay
imo that sounds like it'd would work better in a game that was ground-up to be more of a history roleplaying game, like a game where you're competing for having the most interesting civilization in history (with points gained by having wars, revolutions, wonders, great people and advances in technology, like humankind but not shitty)
>a little difficult to implement in a typical 4x game where minmaxing your civ's numbers is basically 90% of the gameplay
The idea is that you have to weaken yourself in the short-term in order to get the Good Numbers
But yeah, it's a bit out there
>imo that sounds like it'd would work better in a game that was ground-up to be more of a history roleplaying game, like a game where you're competing for having the most interesting civilization in history (with points gained by having wars, revolutions, wonders, great people and advances in technology, like humankind but not shitty)
I'd actually say that's the opposite of a roleplaying game, because you're taking a teleological view, rather than the view of any person that would have lived during that civilisation.
However, the idea of a game where you compete to create the most interesting history possible is actually quite a good one!
Though it's the sort of thing, how "interesting" something is would have to be defined by consensus among the players.
There are quite a few pen and paper games that follow the premise of "collectively generate a history", stuff like Microscope, or A Quiet Year
How would one quantify "interesting" in replicable computer terms?
> Microscope, or A Quiet Year
A man of culture, I see. Tabletop games are really good to study for game design, since the medium demands things be kept simple.
> How would one quantify "interesting" in replicable computer terms?
That's almost begging the question of "what makes a game fun?". People don't like games because the numbers go up, they like them because they're interesting.
That's a good foundational book. For more practical game AI though I'd check out the AI Game Programming Wisdom series, they're collections of articles by industry vets on specific solutions they created in games.
You're basically describing Small World. It is indeed good design.
>how to deal with runaway players?
AI is programmed to gang up on number 1 or rising stars or anyone who feels more competent than them (kinda like players are)
AI is secretly programmed to feed cities and resources to the strongest contender against the player
end game crisis (one thing I'd want to include in a Master of Magic remake is a magical empire, cast forward in time, that traumatically inserts itself on top of an existing empire and starts blitzkriegin'. It doesn't even have to be the player's to work)
something that uses the game's rules against you (so if a highly successful city is densely populated, it would be a plague. Or if success is interconnected trade routes, it's a lockdown/embargo. Things in that theme.)
>early game buildings can't keep up with late game demands
Advanced buildings that give you a 50 percent bonus on all science created in this town, like CiV had in buildings like the Observatory and Bank
Relatedly, buildings that work in concert and give you bonuses
Secondary buildings that give you a 50 percent bonus to all science created in this settlement, like CiV
Oh wait, one more: assuming fantasy, a tower of Babylon sort of project where you spend all your empire's resources to do something interesting but ultimately really dumb- such as, for example, tearing open a portal to a new world that contains an empire even bigger than yours.
I really want to make a game like capitalism 2 but with some sort of population mechanic with different types of people similar to Victoria 2. I saw someone post that using percentage values to represent pops is better than using pops, but I thought that Victoria 2 was a hybrid of this system? Doesn't each group generate as a pop with dynamic population? I really want to make this but theres absolutely no guides or documentation for this stuff.
I want to make a game using JavaScript's 2d canvas, but I'm beginning to think its performance will a lot of work to convert it to webgl canvas?
>'m beginning to think its performance
*its performance will be insufficient, an wonder if it would require a lot of work to convert it to webgl canvas?
If you decouple your game logic from the graphics swapping frameworks shouldn't be much trouble.
yeah its an easy problem in singleplayer games
Testing palette coloring for the coat of arms kinda likes their look. JavaScript made the process kinda painful, but here we are.
i wanted to keep the thread alive so heres some hastily created new UI elements and a construction panel kind of thing. not sure how to represent the different weapons/abilities/items a unit has yet.
this small UI task has allowed me to avoid rebuilding my project for a newer networking API for yet another few days
i will probably have to redesign the way equippable items are represented on units because this looks scuffed lmao
look decent. All of the text needs more of a margin imo, runs too close to its bounding boxes.
I hate UI design, its so unintuitive and finnicky. The only thing I've found that makes it easier for me is to do a lot of pencil and vector art sketches before trying to actually implement it.
the interface here reminds me of Rimworld for some reason. The simplistic cartoon artstyle with fat black outlines maybe,
>velocity of your economy
what is that if not the rate of return on investment?
>what is that if not the rate of return on investment?
in paradox games it is total income per tick and yes, you choose buildings with best rate of roi.
but building with 0.7 income is better than the same building at the end not because it will return you thousands of currency. It support ~1 unit and at the beginning of a game 1 unit is usually more important than at the end. How much money it will generate during remaining game is irrelevant. You always want to increase your total income.
>the interface here reminds me of Rimworld for some reason
it's programmer art, plus i'm trying to spend as little time as possible on it. probably the same reasons why rimworld looks the way it does
I think it looks good
well since we're having an aggy daggy crossover might as well "shill" my free first game
https://gorridev.itch.io/55days
It's a tower-defence game where you have both static "towers" and movable units
It started out as one of those "attrition/meatgrinder" games like warfare 1917 but I wasn't happy with it so It eventually became a tower-defence that's kind of trying to emulate PvE turtling in an RTS
It's not amazing but it took me six months to make mostly because I also spent that time learning my tools and stuff like pixelart and music
Currently working on something else but it's still early I don't really want to show or promise too much
How do you guys push forward?
I have this obsession that I have to set deadlines to myself otherwise I just get carried away
there's not a lot of successful gamedevs on Ganker, so not really the right place to ask.
I'm strictly a hobbydev, but for what its worth; you need to keep your scope small. Ideaguying is well and good but realistically only 10% or less of what you come up with should you actually try to implement. Boil your concept down to its core, then boil that down into concrete, achievable goals.
If I took it more seriously I'd draw up kanban tickets or something, but I just have a markdown document with todo items. Having bite-sized pieces that have a defined 'done' state is really good for motivation. I like to keep a separate doc of all the completed items as well, not sure why but it just makes me feel better.
>get carried away
tells me you're not defining exactly what you're trying to achieve before you start working on it. Experimenting is well and good, but you need to have a restricted scope or it'll never end. Define a hypothesis, test it, record the result and move onto something else. If more questions come up as a result make it a different unit of work, you probably don't need to find out right now.
tl;dr
Having a well-maintained list of todo items and picking what you feel like at the moment and being able to move things from "todo" to "done" is how I push forward
for me is:
>make gamedevving routine
>have a to-do list with manageable items
>spend 80% of my free time on devving
>take it one day at the time and try to take pride in each day's progress
Remind myself of all the homosexuals on /agdg/ who proudly espouse moronic opinions. The desire to one day be able to just meet their idiotic drivel with a screenshot of a good game I've made and then ask them to post theirs and call them a nodev when they can't is what pushes me forward and enables me to yesdev.
a noble pursuit
have a bunch of difficult, crucial tasks required for finishing the game that you dread doing, but won't, so that you will feel guilty and procrastinate by building the rest of the game. and then sunk cost fallacy will hit and you will do those original difficult tasks
I have so many ideas and I know how to execute them, but I'm afraid I can't deliver any serious graphics... Does anyone have the same problem?
it kinda depends. you can get away with simple/placeholder graphics for a very long time if you're making a type of game that doesn't rely on visuals too much
(good) game companies tend not to work on graphics until they've tested the gameplay anyway. ideas aren't worth diddly until they go through testing
JUST DO IT!!! DON'T HOLD BACK!!! I'LL BE PROUD OF YOU NO MATTER WHAT!!!
Why can't I even draw something simple half-decently?
imo the one thing i've learnt from churning out a lot of programmer art is that it's best to keep things as simple as possible. rimworld is a good example of this. the whole game looks like it's built out of programmer art, but it works because it looks clean and the use of color is good. rimworld also uses a lot of thick outlines and simple shapes, which makes it easy to look at the game gameplay-wise
personally i still don't know how to draw a human being but i managed to assemble crude cartoon fellas out of a bunch of simple body part sprites and animate them with a bunch of formulas and stuff, here
. it's not a great visual style or anything but it's at least serviceable (and has saved me a lot of time)
practice, gaygit
not really related to what you originally posted about here's a good example of bad vs good visual design in a strategy game.
in Iron Harvest, the terrain and surrounding buildings have so much detail that the units just drown in the visual clutter. the units are also really small and hard to make out. the end result is that it's just kind of visually straining to look at the game
in comparison, the CoH2 terrain looks a lot less busy, with less intense colors and smoother shapes, almost like it was painted on with a brush. that allows for the more jagged and visually complicated player units and structures to "pop out"
read this, return to your game in 10 years. you might want to make a phd in the meantime so that you don't feel like you wasted all that time on nothing
time to resurrect the thread again, perhaps for the last time
i finally got around to rewriting the core combat logic to allow for a couple of new things
>units can now have armor, which directly reduces the amount of damage taken (taking 0 damage looks a bit sillyright now)
>armor usually comes from equippable items, such as the riot gear here
>units that take cover reduce damage equal to their speed stat, like armor, only it works directionally and only against ranged attacks
>melee attacking can be done "at range", as in you click a guy to run up to them and smack them (also doable at sprint range)
i also redid some visuals, like more clearer/better animations for taking cover, weapon recoil, dying, and so on. offhands now exist (like the riot gear)
wrong webm oops, i hadn't made the unit's innate power stat affect attacks in that one
as it turns out it doesn't apply here either
welp i'm off to bed