We need this in live experiences! I have many plans to do server-authoritative stuff, and many cool game mechanics that I need this for.
This is a great first step towards improving the physics engines capability, much appreciated! However as others have said being able to use this in live servers would be beyond useful.
A few months ago with the announcement of the new Humanoid movement controllers you guys said you were working on generalised server authority, can we expect to see this anytime soon? An ETA would be greatly appreciated.
I second everyone else that this would be great to have in live games. There’s a feature request for this by @MrChickenRocket with 165 likes. [Anti-Exploit / Security / Rollback] - Allow us to manually step the physics simulation
Another use case for the Studio API is automated testing.
Really can u tell for what u can use this?
really what game changing? People just writing those comments for hearts. ‘omg game changer’ , 'omg so good"
WOAH YEAH!
Now we can do accelerated training for self learning AIs’ that uses reinforcement learning libraries like DataPredict!
Thanks Roblox Team!
It would’ve been incredibly handy if these new physics functions were introduced into native experiences! I would then have the possibility to completely pause the simulation without having to anchor every single part, and I would be able to speed the simulation up which I’ve been trying to do to no avail for years at this point…
Essentially, the current network model for character movement is entirely client authoritative, allowing exploiters to easily teleport and fly around just by modifying where their character is.
Of course, this can be very game breaking, and the only real fix is to manually move character authority to the server. However, this makes the responsiveness of movement very sluggish to the user, especially for people with constant lag spikes, so there needs to be some movement extrapolation and rollback to make it feel smooth.
The extrapolation and rollback parts are what we’re missing. We already have spatial queries and unreliable remotes, but we have no way of manually stepping the humanoid physics for validation. That’s why having this API in live games would be game changing.
Please
Dear god please
(Per instance too)
OK, this is actually enormous. I wonder how much of a performance impact would this have though.
Please make this available for in-game scripts as well.
This would be extremely useful for live Roblox games as well, not only for debugging but new possibilities that isn’t currently possible.
Hmm
Is there anything stopping you from having this for live sessions?
It would be awesome to be able to stop and slow-motion the physics simulation
I don’t know much about Roblox physics and or more then basic coding but this is cool and looks like it would be a good thing to have for live games/builds etc.
Does that mean… we will get physics in viewport frames soon?
SLOW MOTION!!! Thank you roblox devs!!!
This is amazing, lots of use case here. I played around with it a bit and I found some inconsistencies when running StepPhysics on Humanoid models in solo studio sessions versus team create sessions, the difference can be seen by these examples:
Solo Studio Session:
Team Create Studio Sessions:
I tilted the character deliberately to kind of get the point across a bit more clearly but basically Humanoid models behave as expected in a solo sessions but in a team create sessions they behave like regular BaseParts with no influence by the Humanoid at all.
@m0bsterlobster Anything you can add to this and any info on whether or not this will/can be fixed? Would love to know because I kind of need this to work for something I am working on
Hi OKevinO,
Thanks for highlighting this issue! We’re actively working on fixing it
Best,
M0bsterLobster
Would really be nice if we could Pause physics and control the Physics timestep manually in Live experiences. Just being able to pause it would open the possibility for alternative physics engines, for example in my case I don’t need actual ‘phyisics’, only Collisions using simple and efficient AABB for scalability.