animation synch... approach/design

[chiblue]

I have been implementing Tnet into my project,  and after a shaky start and lots of help from this forum things are coming together.  The major break through came when I moved the collider to a child object of the remote avatar object.  But now my question is relating to animation.  I have started building the remove client animation control and I want to ask if my approach is the correct one.  I have read much on the unity forums relating to this, both for and against my approach but would like to get some thoughts here given the network technology...

My approach to the animation is that I am sending state changes to the clients... these state changes are move state, weapon state and stance state...

Move state:

Idle, walk, run, jump etc... then based on the move direction I calculate if strafe, forward or backward animation is needed...  or a blend of those..

Weapon state:

Idle, reload, start fire, end fire, etc... this blends animations with the move state animations...

Stance State:

Go crouch, Go prone, Go stand....  these are self explanatory...

On the unity forums I have seen were these state approach is met with some scorn,  in that some people are saying this is a bag approach and all animations for move primarily should be done using the movement vectors on the position and rotation.  Personally I tried this and found the animations are very glitchy... many also talked about prediction models to predict the future location of the remote client... I also tried this but it got rather complex and from my perspective I gained little benefit...

As I said this is really a question about ideas and recommendations... the approach I have works well in my testing,  but before I develop my network code further I wanted to see if I am potentially going to hit a problem resulting in having to rip code out which would be complex later on...

Thanks for any thoughts on this..

[ArenMook]

State changes should be fine in this case, I'd say. Sure, it's generally a better idea to sync the actual vectors that get changed as the result of the actions (sync the effect rather than the cause), but in your case I don't think it matters much.

[chiblue]

Thanks,  that was my thought given that I am not being distinct in the state,  i.e. walk covers a bunch of situations based on vectors..  but it is good to have some input...

[Etarnalazure]

The way I sync my animations is by an int.

I use the TNAutoSync to sync up what number the player
is currently on (0 = idle, 1 = walking, 2 = running ect.)

and the other players just check to see what number the
player currently is on and runs though a switch case.

[ArenMook]

Yeah that's not the best way to go about it. You should either sync the action, or the action's effect. When you do that, you only sync when something changes, and it occurs immediately.

You instead are watching a variable and synchronizing it periodically. Doing it like that means that there is a delay between your action and when you synchronize the effect.

It's quite simple to do it with the first approach. Instead of calling a function such as void SetMyState(int stateID), do an RFC to it like so: tno.Send("SetMyState", Target.AllSaved, stateID). Don't forget to add the [RFC] prefix to it.

This way you don't need the AutoSync script at all.