Keeping an history of the rendered frames

[W1lly]

Hello Babyloniens,

 

I am currently working on latency-sensitive events such as shooting at another player in the context of a real-time multiplayer game. I would like to be able to check the past positions and animation states of my players.  It would be very handy if I could access my scene object as it was a few frames ago. This way I wouldn't have to reconstruct the scene each time I need to access my player's past positions, and I could directly cast rays on the scene to determine the aiming of the player who is shooting. 

To do that, I was thinking that the best solution would be to keep an history of the past rendered frames.

Is there a way of doing that in babylon.js ?

[GameMonetize]

Not really but you can  register a function on scene.afterRender and save meshes positions to an array for instance

[W1lly]

Ok I see, thank you for your answer Deltakosh

One more question though: how could I save the exact bones position/rotation for a given frame? Also I read that making instances of a mesh would result in all instances sharing the same armature, is that correct? If so, is there a way of "copying" the armature rather than creating a new armature manually?

[GameMonetize]

You can clone the armature but it could be really expensive. As I understand your ask you just want players positions right? so why do you want to save armature as well?

[W1lly]

Actually, ideally I would also want to save the bones positions. When a player shoot at another player, depending if the targeted player is running/jumping or whatever, the hitboxes won't be in the same position since they follow the armature transformations, which depends on the animation running at this moment.

 

I seem to have solved this, not sure if this is the best solution but still, if someone is interested to know how I solved this:

 - On each rendering frame I just make a clone of my player's mesh and modify its position to reflect the player's position. I also clone the skeleton of the mesh

 - Then, knowing the current frame of the animation being played, I make a scene.beginAnimation between the frame and frame+1, at a speed of 0 (so I end up with the cloned mesh being stuck in the position of the current frame of animation)

 - I store this clone in an array with the length fixed to the number of past positions I want to save, lets say 10 for instance (so there is a mechanism that makes sure the array is always of length 10, deleting the index 0 and adding at the end when the array is full). I also store the exact timestamp of the rendering frame.

 - Know, when casting a ray I can easily choose the appropriate clone by modifying the predicate of the function.

 

Not sure if I am very clear here...

Anyway, I am still open to suggestions 

[GameMonetize]

Glad that you found a way