The differences between having an AR Anchor or not having the anchor

When I studies the AR Foundation document, I found there is a feature called "AR Anchor Manager"

This is the document: https://docs.unity3d.com/Packages/com.unity.xr.arfoundation@3.0/manual/anchor-manager.html

When I develop using ARFoundation, I place virtual objects under AR Session Origin and the system already keeps them align with the reality.

So what's the scenario that need to use ARAnchor? or I should use them when I add virtual objects ?

The document haven't give real world cases for me to understand.

 

Did you find an answer for that? I'm wondering the same thing

 

Typically, using anchors is a more precise way to place your AR content because anchors will update their positions when more environment data is collected.
Here are anchor use cases I know of:

1. You can use anchors if you're saving the ARKit ARWorldMap or Google Cloud Anchors. Then, these placed anchors can be saved and applied to future AR sessions.
2. You can attach anchors to planes. Then, when underling AR framework will update the plane's position, your anchor's position will be also adjusted.

 

I have tried placing three objects: Without anchor, with anchor, with anchor attached to the plane.
Objects are placed using AR raycast manager.

However, they shows the same behavior:
drifting away when tracking is bad, snapping back when the tracking is stabilized.

What do the AR anchors do and when do we need it?

 

They basically 'tether' an object to a feature-point in the environment. If the feature-point is updated (i.e. its position is determined to a more accurate degree), so is the anchor.

 

how can i spawn and delete objects in ARfoundation