Include the following details (edit as applicable):
Issue category: Developer Tools
Device type & OS version: Android
Host machine & OS version: Windows
Issue Environment : On Device
ARDK version: 1.2
Unity version: 2020.3.32f1
I’m getting some pretty extreme drifting and tracking loss of my anchors which is becoming apparent is part and parcel of tracking in AR. It’s pretty hard to tell if the IARAnchors are working correctly in my build or if it is just the tracking quality. I’ll have a look at some of these suggestions:
In the meantime, can I expect arbitrarily placed anchors added to the ARSession to stay locked to a world position?
Hi Dudley HK, it would help to know your use case here? Also, please read the documentation on Tracking AR Anchors.
Furthermore, the farther away a virtual object is from the phone, the more drifty it will be relative to the real world. By creating ARAnchors you can create tons of stability where virtual objects that have their position computed relative to those ARAnchors will stay stable.
Also, if the device position tracking fails, often it will begin to drift off in one direction. If the device tracking incorrectly believes the phone is drifting to the right, the virtual camera will go to the right as well. This will cause all the IARAnchors to appear to drift to the left on the screen.
I hope this helps. Thanks.
Hi, thanks for the reply. I got it to work using the Placement Example. I think one of the main issues was trying to place big objects in front of me . The only issue now is that basic Anchors are a nightmare to find - probs best to use Plane or Image Tracking.
For a bit of background, I’m working towards creating/ displaying objects at geo-locations in the world using GPS data/ LatLon. I just got stuck on placing the object in the world once the user was close enough.
Is there a way to use a custom method for relocalizing when camera tracking fails? I found that while Lightship’s default system is quite good and probably sufficient for most use cases, my game is very fast paced and requires lots of quick relocalizations and I’d like to use a custom ML model for it.