This is a call for opening up discussions regard creating a unified solution for Cloud Anchors.
We're so saddened by the fact that there's still no open standard for anchoring content to the real world. We want to build platform-agnostic co-located experiences for any MR/AR capable devices (AR-kit, AR Core, HL2, ML2, Quest Pro, etc.).
Azure Spatial Anchors has come a long way, and would be our go-to solution for space tracking - except it doesn't work on the MagicLeap.
If it worked, we would love to integrate it with Microsofts World Locking Tools in order to easily create co-located multiplayer experiences, but also in general for having a much more robust solution to anchoring bigger objects and working with physics between anchors (these issues and solutions are described in the documentation).
World Locking Tools promises a superior approach to anchoring content compared to traditional anchors while building on top of traditional "spatial anchors" and staying platform agnostic.
An alternative is using Area Targets from vuforia. However, these targets can only be generated and used in the same runtime on Lidar enabled iOS devices, requiring customers to generate co-located experiences using other third party hardware than what the experience is being run on.
Lastly - and this is our current solution - is using visual markers like QR or Aruco markers, or 3D model targets for aligning a virtual origin to the physical space. When only using one marker, this introduces drifting the further away from the origin you get (or just general bad tracking in some devices). This could again be mitigated by using multiple anchors and World Locking Tools to lock the virtual space to multiple physical anchors, but wouldn't it be much nicer if these where all just cross-platform anchors?
That's what this post is about. Let's push for a common cloud-anchor/point-cloud/spatial-map/whatever-you-wanna-call-it standard.
Here's my small contribution - A ML2 Spongy anchor data provider for WLT;
MagicLeap2 - Spongy Anchor data provider.zip (6.0 KB)
Let me know what everyone else thinks!