Root motion in multiplayer update?

Does anyone know if there are still problems with using root motion in a multiplayer game?
I read this and I’m concerned because it seems like the only way to get fluid smooth character movement / abilities

I’m working on a game that maximizes the player limit so I’m not sure if root motion is possible for me?

Or should I

Searching about this topic leaves unanswered questions that we really need!

I’m still hoping for some sort of official comment on this, but it’s been a while with no response at all from anyone at Epic.

Thanks for bringing this up again CyberDev.

Here’s my original thread and post:

Well there are multiplayer games using root motion no? Such as GTA, any smooth style with directional changes and flipping around fluidly, so how are they doing it. Different more complex form of networking built specifically around RM?

Ohh men, been looking for a clear answer for a while too. I’ve had a few meetings about this at the job and it be really nice to understand what are the drawbacks of this system in multiplayer (our game is both single player and multiplayer).

Well root motion is the future but coders like in place using the movement component while animators love root motion as it keeps the player grounded. If you wish you could use RM and can replicate but the movement component handles a lot more that you would have to more or less build from scratch using RM.

Until there is a networked “root motion” component as there is a in place movement component it’s a risk at least our/my coders are not willing to take.

P.S. Epic has yet to state “anything” as to networking with root motion.

just to give my humble feedback. I’ve spent few hours on setting up root motion for a single player game. With rootmotion from montage only, a careful replication of all parameters involved in montage control over network should work fairly well. But I think you may need to mix rootmotion with inplace motion at some point to have a more precise control.
This mix will raise a problem : the properties of the capsule component after a rootmotion are partially updated (basically, you exit rootmotion sequence with a 0 linear speed most of the time), so you may have to manually compute things at this point (I am still investigating this transition to make it smooth).
Basically, using rootmotion for complex attack or hit sequences or jump start and such is likely to work well, while idle/walk/run (including in arch), and jump/fall may be better handled in-place over network. Basically, network desync is especially noticeable in high-speed translation, which are the most frequent action taken by a player (who is more often walking and jumping that kicking, even in a hack&slash).

This mix may help coder and animators to love each other again…