Roblox replication isn't so good. Fighting games aren't always compatible with server authority.
As an alternative, we use delay based lerping, which yields a pretty good result beyond 200ms ping.
#RobloxDev#RobloxStudio#Roblox
[Video from April 2026]
@TheLegPikachu@AvaluateOrijua no in that the principles applied are identical (lerp, extrapolation, using buffers)
yes in that i don't use chrono, what i have is more rudimentary compared to chrono
performance wise what I have is probably similar to chrono for pvp, but worse for pve
Development update for Rise of Corruption!
Some work on optimisation @ 60 fps:
- Client: improvement from 510/s max ➔ 140/s max, a 73% improvement
- Server: improvement from 60/s max ➔ 40/s max, a 33% improvement
(worked on it while sick...)
#RobloxDev#RobloxStudio#Roblox
@aominoa_roc@bytejon@AvaluateOrijua Wrong. Its advantage is securing most character and physics behavior which accounts for most cheats. You can’t prevent these without server authority unless you make your own(ie chickynoid). Roblox just rolls it into a useable system for you already with decent prediction
While what you said is valid, there are still issues with SA that stop me from using it. I also have some pragmatic reasons why I don't use SA, which you may or may not agree with
1. Main issue I have with SA is IAS. IAS has questionable support for multiple inputs on controller (I've actually experimented with this) and will pose mobile control issues (sorry, this is a UI limitation). It would be pragmatic to keep drop controller and mobile support and not use SA
If ROC were a battleground game I wouldn't have a good excuse to not use SA, but given the amount of inputs we have (if you've played BM2, SCR on Roblox), I've decided SA wasn't suitable. This is what I was implying by SA isn't so suitable for fighting games
(technically IAS does support double inputs on controller, and I've played around with it. it's still inconsistent)
2. The reason why I mentioned remotes over physics cheats, despite the first having no relation to SA, is because many exploits for fighting games aren't physics based. The ones I've seen either abuse poor remote protection, poor hitbox sanity checks, or scripts akin to macros. In a pragmatic sense, serious exploiters wouldn't want to be detected in the first place (though obviously you still have those who find it funny to increase their walkspeed)
Even if we talk about cheats for other games, say FPS games, they tend to either hide themselves in CoreGui or are "externals" (i.e. they just inject ImGui). Neither can be detected by game developers, only mitigated
I guess the cat and mouse game you mention are the exploiters who get out of their way to tackle client sided anti-cheats (like the Roblox devs who really don't like Dex Explorer, or people using networking libraries to make it harder to use remote spy, or even just StreamingEnabled to deal with saveinstance)
3. I have to admit I won't be doing any checks for physics based exploits, aside from basic magnitude checks on the client side (someone with a decompiler can easily defeat this). This is an obvious downside of not using SA, but given the IAS issue, I don't have a choice
You're also right an exploiter can certainly find a margin where they can execute physics based exploits (another cat and mouse game), including playing their own animations, but from a pragmatic perspective I would treat the advantage as marginal
(note: I have to disclaim I only have a year of experience of Roblox development)