Updated RealSense Avatar Multiplayer Test

[MartyG]

Hi everyone,

Earlier in the year, we tried a multiplayer test with RealSense, in which we placed two copies of our full-body avatar in a scene and had them hug and kiss using a single RealSense camera.  The results were mixed: although they were able to position their bodies and hands to express intimate actions, collision detection was not yet working properly and so the couple's arms were very vulnerable to passing through each other.

Now that we have a collision system in our latest version of the avatar, we decided to repeat the test so that we could compare progress in enabling complex full-body multiplayer with RealSense.

Note: the video below is not naughty but probably best not to view at work if anyone is likely to object to seeing hugging and kissing squirrel-people!

https://youtu.be/Hq9Nkyux1nI

We have mentioned in other posts about our collision system that it can make the avatar laggy if it is over-used in the avatar parts.  So whilst one avatar on its own moves well, two doubles the amount of collision physics that Unity needs to process and so movement is awkward, but solid arm-to-body collisions are at least taking place.

We tried setting one of the avatars to only use the default weaker collision settings instead of the enhanced ones.  The two avatars moved faster but the one with weaker physics settings tended to pass through the one that was fully physics-enabled.  Further testing found that this wasn't the fault of Unity's physics but the size of the collder fields used in our collision system, which are quite large in order to increase the reliability of collisions.  So there is certainly a possibility for effective complex multiplayer (or at least complex physical interaction with NPC avatars) if those fields can be fine-tuned further.

[Efreet1]

Hahaha very cool Marty :)

Keep up the good work!

[MartyG]

Thanks so much, Efreet!  I appreciate the feedback and encouragement very much indeed.   ^_^