How does scan localization work with multiple VPS private scans

I’m using several scans of the same location as private VPS points. On some devices, if I’m only initializing the localization with a single scan ID I get a quick result and the point localizes succesfully. If however I initialize the scanner with 5 scan IDs, on some devices it fails to localize at all. Is it a coincidence, or does the localization backend not return a result if it has multiple scans with the same probability match?

Hello! Could you please tell us more about your project? When you created the scans, did you create them as private or test scans? If you are uploading Test scans, you just upload one scan, not multiple. Additionally, are any of your scans failing to upload activate and/or do some devices simply fail to localize to the VPS locations?

Private scans, we’re only testing with scans that succesfully uploaded and activated. On one or two devices it resolves even if we list all of them in the localizer initialization, but on others it only resolves if we request a single one of them in the localizer initialization (but it resolves eventually regardless of which exact scan we use). All scans are in the same location, of the same object, in several different times of day.

If this is the expected behavior of private scans, it’s fine, we can work around it, but I’d just like a confirmation if it really is expected behavior, and if there’s a way to get the localizer service to return several partially matching scans and let our app sort the results to determine what to attach the AR content to.

Do you mind sharing your scans or any screenshots of your scans. Additionally, how are you loading the scans into your project?

I can’t share the exact scans as the client won’t permit it. They are stored as prefabs and loaded into the system via LocationManager.SetARLocations(array) + StartTracking methods.
And my question isn’t about these specific scans, I was just wondering about the underlying logic of matching the current user position to several possible scans. It seems that we get a more consistent location activation when providing a single ARLocation, then if we actually set it to a list with several ARLocations (even though it’s the same physical location, just several different private scans). So I’m wondering if the localization backend actually does wait until it can differentiate between two similar scans with similar current matching score, rather than just returning the highest matching one as soon as at least one pases some match percentage threshold.