There seems to be a case where the combination of Model Target and Human Occlusion (AROcclusionManager on ARFoundation) does not work well.
As shown in this attached video, the hand itself seems to be recognized, but it seems to fail to determine the depth map.
This seems to occur frequently, especially on the iPad.
Does anyone know how to solve this problem?
Note that when combined with Object Target (Vuforia 9.8), it works as expected.
Object Target has been removed in recent versions, but we would like it back, as it has advantages in terms of load and stability.
Unity 2021.3.8 / ARFoundation 4.2
iPad Pro / iOS 15.5