Need Advice From Experienced Combat Scripter

[sage jacinth]

Hi everyone,

I’m working on a combat system and I’m trying to understand the core layers that make it feel solid, dynamic, and responsive. I’d like some guidance from people who have experience with action-heavy combat systems.

Specifically, I want to figure out how to implement action canceling. For example, if a player is blocking and takes a critical hit that breaks the block, the block action should be immediately canceled. I want all interactions to feel smooth and reactive.

My goal is to create a combat system similar in feel to Ghoul Re, with fast and engaging mechanics. Video Example: Press

I’m also thinking whether to structure the system using priority-based states in a state machine or to rely on allowed transitions, and I’d love advice on which approach is better for handling these kinds of interactions efficiently.

Any tips, examples, or best practices would be greatly appreciated!

I've already created a combat system, but it's not exactly what I want, that's why I'm asking

[vale coral]

just first make the core machanics

like block and than that it stops the block

after that u can always polish it

[night wyvern]

you, and thousands, tens of thousands of other kids who took one look at tsb and thought "I could do that!" ... 😮‍💨

[sage jacinth]

I don’t like the TSB combat system. Even though it feels smooth and reactive, I’m still trying to figure out a better way to script combat, and I’ve rewritten the combat system from scratch many times to reach goal I want

But I still think my current scripts aren’t perfect, and I’m sure there’s always a better way to implement some mechanics than the way I did

I have been playing these types of games a lot, like Deepwoken, Ghoul Re, Type Soul, and I was even high rank in one of the best clans, but I found Ghoul Re’s combat system way more fun than the others

[night wyvern]

playing games does not directly translate to ability to develop games

[night wyvern]

as if you, and the thousands, sorry, tens of thousands, actually i'm probably still underestimating, let's just round down and say millions of other kids who look at tsb, deepwoken, and others like them, and then think "yeah i could make that"... 😮‍💨

[umbral rain]

yeah i could make that

[sage jacinth]

You’re always trying to bring up TSB and “I could make that.” I never said that I can make the same combat and make it perfect. I’ve spent a lot of time trying to find better ways, and that’s why I made this post - to get suggestions from people who can actually make good systems. I mentioned Ghoul Re because I feel like its combat is one of the best I’ve ever seen.

[night wyvern]

is that not exactly what you are asking?

[sage jacinth]

No ? I’m not claiming I can make the perfect system like Ghoul Re. I just want advice on improving my own combat system and learning from people who have experience with these kinds of mechanics.

[night wyvern]

well if you want advice to improve your combat system, you'd need to start with some of the problems you're having. you didn't do that. you seem to just be asking 'how to make Ghoul Re combat? 🥺🥺🥺'

i mean, in fair you have this

I want to figure out how to implement action canceling. For example, if a player is blocking and takes a critical hit that breaks the block, the block action should be immediately canceled.
but this is just an if statement lol

(inside on_crit()) - if isblocking(plr) then cancel_block(plr) end ?

[sage jacinth]

“How to make similar in feel to ghoul re?”

It doesn’t mean I want to make exact clone of their combat

[night wyvern]

you need help catching edge cases -- what advice more did you need than just simply catch the edge cases?

I’m also thinking whether to structure the system using priority-based states in a state machine or to rely on allowed transitions
again this is like asking how to write an if statement. you realize a state machine isn't a fancy construct. it's literally just an if statement. a debounce is a state machine for example, just a very small one.

you've surely written debounces before? yeah? then you know everything you need about state machines.

[sage jacinth]

Yeah, this stuff is easy to make

[night wyvern]

I’d love advice on which approach is better for handling these kinds of interactions efficiently.
maybe stop restarting from scratch and finish what you have

or maybe you want to keep restarting over and over and never finishing it, idk 🤷 people want different things

[sage jacinth]

But you should understand, I’m sure that state machine isn’t the only thing they use to make great combat. I’m certain they’re using a lot more techniques and systems as well

[night wyvern]

nope

there's no advanced techniques, no tricky tricks, no hooplah no cranberry sauce. it's just what happens when you put productive code to work, and maintain it over time.

the only exception to this is rollback netcode and frankly tsb doesn't have it, i doubt these other games have it either

they just assume you have low ping and the rest is done as some combination of on the server and on the client where network ownership permits

or maybe even all of it is done on the client and the server has no authority making it really easy for exploiters to cheat

hit(everyone,999999dmg)

[night wyvern]

but this is more likely

which is like the most naive approach you can take

[night wyvern]

and just briefly skimming through that video you linked, i see no reason not to believe this is just a basic combat system. nothing fancy going on.

just low-ping interactions. nothing else

you can tell this because there is:

1. no knockback at all.
2. very little to no physics interactions at all whatsoever
3. i don't think i saw a single stun that was not either telegraphed with a lot of time for clients to predict, or just outright not connected to the player at all.
4. when players do touch your spike and get stunned, there is a delay after they lose stun before their character starts moving again and they just hang in the air for a bit.

so i wouldn't call this smooth at all

i'd call it primitive more than anything.

it's just animations and vfx, nothing else.

yep there it is, finally found proof

"So Smooooooooooooooooooth"

[sage jacinth]

There’s knockback on certain skills or last hits.

so I wouldn’t call this smooth at all

-> Script-wise, it might not be perfectly smooth, but if you’ve actually played these games (which I doubt you’ve done much of), you could say the combat is actually fun and engaging

[night wyvern]

notice how they jump past the spike then teleport back into it, and then they jitter again after the spike ends. this is telltale sign of high latency

[night wyvern]

yes and i'm sure every single one of them locks your movement and holds you in place for about 0.3 seconds

or the other player just stands in front for a moment after being hit before actually being knocked back. probably both ngl

that's why there's little knockback, because if there was a lot of it this would be extremely noticeable - more than it already is

[sage jacinth]

Why would it need to hold you for few seconds after just being knocked back after last hit ?

[pliant moss]

** You are now Level 13! **

[night wyvern]

because you cannot predict player movement

holding you in place lets other clients predict exactly when and where they would get knocked back, allows for smoother experience instead of walking past a hitbox then being teleported into the knockback, or being knocked back in place.

this plays into the network ownership model that roblox uses

no knockback verification happens on the server, the client decides for itself because all control is, rightfully, delegated to the client.

i'm sure cheating is rampant in this game

[night wyvern]

so ya just back to my first advice; just finish what you actually have and stop restarting from scratch 'bEcAuSe eFfiCiEnCy '

performance is not going to be a problem for code like this

nor does efficiency actually matter

heck, particle emitters and debris instances are probably going to take up more cpu, memory, and bandwidth than anything your actual combat code does.

like just to reiterate, the 'efficiency and optimization' question starts and stops with simply asking: "does your game stay above 60fps under the highest load situations?" if the answer is yes, any more optimization you do is premature.

[sage jacinth]

I’ve only seen one that could deal a thousand hits per second with certain weapon to instantly kill a player (which was fixed really quickly). Other than that, I haven’t seen anything besides the fact that they can fly, and I’ve only seen a couple of people like that the entire time

[night wyvern]

haha yeah i bet cheers to the game taking forever to level up, and an active discord where the game devs ban people if proof of cheaters is shown

there's much more subtle ways to cheat than just teleporting behind players and throwing 1000 hits at once

like... if you detect you're in a hitbox, automatically roll out of it.

or... a bot detects extremely high probability of a special attack landing that deals a lot of damage; automatically use the attack.

[sage jacinth]

Exactly why, still haven't they come up with anything other than flight?

[night wyvern]

i just gave you 2 examples that you probably encounter regularly but have no idea because it's indistinguishable from what a human player can theoretically do.

like just as an example, in fps games there are trigger bots that don't automatically aim for you, but they do automatically press the shoot button whenever your cursor is over an enemy.

that makes the cheat look more believable, and thus, more likely to get away with it

cheats like this would be absolutely rampant in a game like this

specially if there's an established route for moderating cheating players who make it painfully obvious they're cheating.

if someone has spent hundreds of hours in this game with cheats, do you really think they're going to risk losing all that by being an obvious cheater and getting reported to the devs? of course not.

so they stick to more subtle cheats

it's not an outright instant win, but it's an edge that other players will not have.

like automatically pressing shoot whenever you're aiming at a player - i mean if you never actually aim at anyone, cheating or not you still won't win.

but it does tip the odds in the cheater's favor

another example in fps games is wallhacks - again, it's not always strictly obvious, you do have some hints that the player has more information than they should but it's very hard to prove a wallhack. this game is guaranteed to have an equivalent roster of subtle cheats exactly like this.

AI query: common hacks in roblox ghoul://re
Overview of Exploits
The general term "hacks" or "exploits" in the context of Ghoul://Re refers to third-party software that injects scripts into the game environment, modifying gameplay mechanics beyond their intended use.
The primary types of exploits mentioned in search results include:
Auto-Farming: This allows a player's character to automatically kill bosses or NPCs repeatedly without manual input, often overnight, to quickly accumulate in-game currency (yen), RC cells, experience points, and rare item drops like fragments and blueprints.
Auto-Grip/Devour: This involves scripts that automatically "grip" or "devour" other players or NPCs (if the player is a ghoul) to quickly gain RC cells, which are necessary for regeneration and powerful kagune abilities.
Automated Progression: General use of scripts to bypass normal gameplay loops, allowing users to rapidly level up or acquire high-tier items (like certain Kakuja or Arata items) that would normally require significant time and effort.

and this is just the top ones that the ai is aware of

and hey would you look at that

"auto-grip/devour"...
me 5 minutes ago:

like... if you detect you're in a hitbox, automatically roll out of it.
or... a bot detects extremely high probability of a special attack landing that deals a lot of damage; automatically use the attack.

and go figure would you fukken look at that, auto-m1 https://www.tiktok.com/@aestal117/video/7486093482998746375

auto-use-skill

enemy lock-on mod

auto charge-attack

auto uppercut

this one is interesting

Instant-Kill Threshold

ESP (wallhacks)

and that's just at a quick glance of the cheater's menu

[sage jacinth]

Auto farming exists in every game, and you can't really avoid it. Auto grip-devour doesn’t allow hackers to instantly grip you, and I don't think instant kill exists at the moment. Have you seen when this video was out? Even with auto farming, you still can’t progress quickly and reach the top in 5 minutes. ESP doesn't help in this game, unlike in shooter games

[night wyvern]

ya i wouldnt think the esp is very useful here

but this doesn't change my point

there may not be obvious cheats because the devs take action on moderating their game

but there sure as hell are subtle cheats

[night wyvern]

boy howdy i sure do love just saying things just for the sake of saying things

[night wyvern]

so ya the tl;dr of all this is just finish what you have and it will look like theirs with a bit of fine-tuning

[sage jacinth]

And yeah, the game is being moderated, and they even hired exploiters to help create a good anti-cheat when they closed the game. They were paying a really good amount of money for it.

You don’t lose much after dying to an exploiter, so you can just join a new server. As for permanent death (where your progression resets), you can always restore it using robux or by opening a ticket with proof on their Discord server. (WAIT, PROOF FOR REAL? You can just press F12 to record that you were actually gripped by a hacker)

Roblox is literally filled with exploiters, so you can't really avoid them entirely

[night wyvern]

haha yeah, now you're starting to get it

this aint just about an m1

it's everything else around it

like this is just purely the combat and how the game is full of cheaters

this isn't everything else you can do in that game.

[sage jacinth]

But there's another thing - although Roblox is filled with exploiters, the only thing that changes is how well different games can be protected from them. Some are protected better, others less

[night wyvern]

you don't even have a basic combat down apparently, yet you want to be like them.

here's a tip, get your basic combat down, and then worry about fine-tuning it to feel how you want.

after you got a bunch of other content to work with it

like npcs

[sage jacinth]

.

Creating basic combat isn't hard

[night wyvern]

ye its just tuning

every system has to be tuned

how far does the dash go; how big is the hitbox; how much damage; how long is the stun; etc etc etc

[sage jacinth]

Just a few functions and handlers - that's all you need for M1, block, and perfect block

[night wyvern]

ye and if you stopped there, you didn't tune it. you made one attempt and went "welp this isn't perfect. DELETE IT ALL AND START AGAIN".

it's not perfect because you didn't tune it to perfection.

it's not because it isn't perfectly efficient. it's because you didn't simply finish it.

and then you had nothing to tune it against.

like i can pretty much guarantee you every single combat game started out like yours until they added more content to the game over time and had plenty of time, and tonnes of feedback from real players, to tune the feel of it with.

you have none of that.

you spent a day, didn't like the output, and now you're asking what you're doing wrong.

[sage jacinth]

Alright, I hear you. My post might have been stupid, but "welp, this isn't perfect. DELETE IT ALL AND START AGAIN" - I rewrote it from scratch because the code was becoming messy. I always tried to start over by looking at my last script and modifying it to make it clearer and more expandable, but I still think there's a better way than I'm using now

[night wyvern]

you are doing all the right things, you just aren't sticking to it

[night wyvern]

because the code was becoming messy
that is exactly 💯 what i am talking about.

it became a little messy so you gave up. the messy part is extremely normal

the giving up is on you

no code that does actual functional work is going to be clean

and perfect

it never will be

time to accept reality

i actually have a video that i think would be perfect for you right now

[sage jacinth]

Anyway, I appreciate our dialogue and I got you, and I’ll take everything you said into account (at least I'll try). But as for the Ghoul Re combat system, I’ll still stick to my opinion. It might feel bad to you, but for me, as a user who’s played it a lot, it’s fun

[umbral rain]

u cant? so how will you solve responsiveness?

i never made combat system before

not fking interested

so normalize me even if im s2 :>

[night wyvern]

i highly recommend watching this; https://www.youtube.com/watch?v=tKbV6BpH-C8

[umbral rain]

1. Measure
2. Try Something
3. Measure again.

Very wise

[night wyvern]

i'm suggesting it's okay to sacrifice performance and adaptability for velocity to get your mechanics and features working the way you want. and often once you got the mechanical body of your code in place, it is exceptionally easy to wrap it in a new pattern with all the same behavior and feeling, just different ways of passing data around

[umbral rain]

i mean, just target players that have high end pc

:>

and no need for optimization

(jk)

[umbral rain]

by the way pyro, what do u use for networking

is it blink

is it

is it..?

[night wyvern]

remote events

or attributes

[umbral rain]

pftt everybody knows that

be more

specific

theres no way u use raw remote events and dont wrap them

right?

right..?

[night wyvern]

but ya no problem, it's probably the best advice you're going to get that you pretty much cannot pay for. up to you if you choose to follow it or disregard it, or whatever combination you'd be happy with it. i say it before and i say it again, i'm not your mom and you are perfectly capable making your own decisions good sir i hope your game gets off the ground soon

[umbral rain]

respect

[sage jacinth]

I'm not making a game because I understand that creating a good game requires a lot of experience, which I don't have at the moment. So, I'm just working on systems that could be useful in the future

[night wyvern]

speaking from experience this is just plain and simply not true; you can absolutely make good games with low experience, in fact s1 is enough

the trick is to pick something that isn't too technically demanding for its core gameplay loop

like even just something as simple as making the game pve instead of pvp would immediately 'throw out' a whole bunch of issues you're having and gives you an immediate path to make a game loop out of what you already have

you can then add pvp in later when players ask for it

[sage jacinth]

Well, rank doesn't mean much, because I'm pretty sure there are a lot of people who got their ranks using AI scripts. If we're talking about good games, it definitely requires much more experience

[night wyvern]

(not if, when)

[night wyvern]

nah ai cannot maintain a coherent script longer than 1 or 2 requirements

if you ask it to do 3 things at once it just breaks down and cannot handle it or maintain a coherent thought pattern past 2 things

and all those scripts get rejected because they don't do enough

i mean grow a garden is a great example

that game is incredibly simple, anyone could make it

[sage jacinth]

agreed

[night wyvern]

the trick is no one does because you have to stay on task over weeks, months

on one project

that takes dedication. not experience.

[sage jacinth]

Making a lot of AI prompts will eventually lead you to script something that works (even if it's not perfectly clear, but it will function). People can just prompt 3 different systems that could help them get to S1 (because it doesn't require much)

[night wyvern]

i cannot stress this enough.

I rewrote it from scratch because the code was becoming messy.
learning to deal with the technical debt you create for yourself without starting from scratch every time is easily one of the most important things you can get experience on

because there will always be times, in any game, where you need to deal with the technical debt.

[sage jacinth]

I think the only difficult part in GAG is the values, growing time, and other modifiers, because you'll usually struggle with a lot of values

[night wyvern]

nope that is fairly easy

it's just a deterministic loop

instead of calling plant:grow() once every minute or so, it calls like 1000 of them at once.

[umbral rain]

probably the branches

[night wyvern]

more likely it's a deterministic algorithm so you can call plant:grow(duration) instead but it's the same as calling each grow step 1000 times at once.

[umbral rain]

theres a tutorial for making gag

[umbral rain]

WHAT

[night wyvern]

so no, the tricky part of making grow a garden is the volume of work and content you need to put in it

[umbral rain]

i guess s1 is enough for making game

[night wyvern]

it's more than likely every plant has a custom grow method

and there are variations within them

[umbral rain]

im actually interested in how they do this

[night wyvern]

so if you then add something like ... pets that can eat straight off the plant ... yeah, that's going to cause some problems.

[sage jacinth]

Well, even though GAG is simple (in some ways), the devs actually did a lot and definitely spent a lot of time getting it ready for release, so it succeeded, and they made Roblox history (of kids and bots playing their game 😂)

[umbral rain]

https://devforum.roblox.com/t/how-to-make-roblox-grow-a-garden-game-tutorial-series/3769116

LMAO

[sage jacinth]

There's a lot of leaks, so you can just take a look of their scripts that were obfuscated

[night wyvern]

that's the other thing too - polishing and releasing a game, then maintaining it over time is far more valuable experience than making any one particular feature

it's harder to get to, but it is also better to get there as early as you possibly can

[umbral rain]

do u actually do comms pyro

i think u've told me the answer

it's a no

[night wyvern]

actually to be completely frank with you, the moment you hit 50+ natural 7dau players, your game is officialy alive, and at that point, the entire development cycle changes

and it changes because those players, who like the game, will give you feedback and suggestions

[umbral rain]

r u fr?

hm

[night wyvern]

experience working with real time feedback and suggestions is way more valuable than ability to make any one individual game feature or content

because it tells you what is or isn't working, what does or doesn't feel right

what parts are fun and what is boring. what you could add.

like that's probably why im a senior ranker, i have tonnes of experience with live dev cycles

.... outside of roblox tho

inside roblox i've released 2 bad games, neither have ccu but they were just stepping stones to make my next one

[umbral rain]

maybe that's why you're s3

[night wyvern]

ya s3 because outside roblox doesn't count, and i haven't made anything particularly special on roblox yet

[umbral rain]

maybe u'd get s4 soon

[night wyvern]

when my game is done it'll be s4 worthy

[umbral rain]

and it'd take me years to get on your level

im envy you.

fk man

[night wyvern]

don't envy hard work

because that's all it is.

hard work, and maybe a little bit of luck, but mostly just hard work and effort.

i didn't get to this level by watching youtube and playing grow a garden all day. i've been making stuff for years

[umbral rain]

you're right

[night wyvern]

and yeah, no, i don't bother because it's a massive waste of time and i really don't need the robux

i did before but i've got enough now

[umbral rain]

experience?

hm

[night wyvern]

yep learned that from experience, and i happen to be lucky enough to not need it

[umbral rain]

ngl im jst doing revenue and coms game bcs i seek experience

pretty much got s2 bcs i do comms

[night wyvern]

finishing and packaging products is releasing a game - lite version

[night wyvern]

yep, that's how it goes. just keep making stuff

[umbral rain]

i've made 3 game, sadly none of them got ccu

atleast they got me s2

and tons of experience

it's worth

i guess

[night wyvern]

that's where you got s2 from

[umbral rain]

im actually grateful now

lmao

lucky i made those 3 games

would hv stucked s1

if i didnt

[night wyvern]

like, i feel like i should impress upon you that i'm not joking when i say make bad games, just so you have something that you actually finished. it makes such a big difference actually finishing a full game, even if it is bad, compared to not even half-finishing one feature, or spending months restarting it because 'what if my game needs to do X' and then your game never actually does X.

this is one single gigafunction in the second game i released about 8 months ago, so this is my most recent released work, it's 150 lines for one function and it is some of the worst code i have written in quite a while. Why? Because i couldn't be remotely bothered to do it properly. Does it actually matter to the game? well, it works, and that's all the game needs from it.

and for reference the file this is in is a bit over 1000 lines long

[umbral rain]

ngl im making 2 games right now but sadly the leader is so.. bad.

i wish i could lead the game.

he doesnt see the vision like I do, maybe that what makes me always excited

when making game

i see the potential

thats why i accept revenue

[night wyvern]

just briefly skimming through the code i'd say there is in total about 3k-ish lines

maybe 4k being suuper generous

[umbral rain]

and god they're using free models

[night wyvern]

i use free models too

[umbral rain]

the difference is there are tons of backdoors

xd.

[night wyvern]

i curate all scripts

[sage jacinth]

btw, is luau the first language you learned?

[night wyvern]

now you might be wondering... why would i accept such low quality code as this? well, for 2 very simple reasons:

1. I wanted this game out ASAP. in total all of this took me a month of real-time including sleep, working like 8 hours a day, 7 days a week.
2. I had so much other work to do, not just the code. like building levels, picking out npcs and other assets and models, the ui, picking out sounds, setting up and configuring gamepasses and their icons, and configuring and balancing levels in the game, getting art and marketing material together etc.

So naturally with so much other work to do, something had to give and the easiest thing that i could give was resorting to code like this.

[night wyvern]

and the thing is, even with code this bad, it would be very easy to refactor this without changing the fundamental behavior or timing of anything

thing is i just don't really care to; it works good enough and isn't causing problems

[night wyvern]

no, my first 'language' was html. my first actual language was php, with a bit of javascript afterwards

+beginner level sql+css y'know all the fun webdev stuff

i made most of my own forum website out of php

you could login, post messages etc

wasn't the only website i made.

but it was the most 'complete' one i made

everything after is just tinkering with stuff like wordpress

[sage jacinth]

I'll keep developing systems until I feel ready to make a good game, not just a simple game "Kill to buy new weapon"

[night wyvern]

i even made a php/javascript combo chatbox 🙂

[sage jacinth]

Is luau knowledge can be helpful in learning java?

[night wyvern]

i've made all sorts of stuff since then but ya, i'm not kidding when i say just make something and finish it

[night wyvern]

javascript is not java. but yes, all programming languages are transferrable to some degree, often to a large degree.

e.g the concepts you learn in luau can apply to java, python, c++, c#, javascript, vbscript, php, etc, even assembly. it's turtles all the way down.

i don't think it's worth going out of your way to learn these other languages just for the sake of making things in roblox studio tho

it'd be better to stick to luau.

but hey i mean if you ever wanted to make a website you gotta start somewhere

[sage jacinth]

That's what I'm doing right now, luau is one of the easiest languages ig and once I master luau maybe start learning new language to use it outside roblox

[night wyvern]

ya that's fine. when making a game you should pick your engine first, not the language. like it would be unreasonable to try and shove lua into a game made with unity, you might add it for modders but core gameplay behavior would probably be unity's native language C#

[pine meteor]

i think we've lost the plot

@sage jacinth vfx plays a major role in combat

without it no matter how great your code is it will feel slow and sloppy

animation especially

[night wyvern]

if you got advice you think is relevant feel free to take over, i got my own combat game to work on good luck~

[umbral rain]

dang sir

u obviously will use your data storing for that game for sure

[sage jacinth]

yeah, I know, especially for techniques like Rollcasting, cuz players usually focus on anim and sounds

[visual creek]

if kick and target block then dont damage

easy

[bleak cairn]

is this ragebait

[raw lodge]

you clearly don’t understand