It is possible to unstack parts from multiple locations using the Wrist camera and only 1 cam locate node.
Here is a program template to make it :
Then teach your object in the cam locate node, then select the ignore snapshot position option from the cam locate setting menu.
Robot will then start from position one until it can't detect any part then it will move to point 2 and so on.
Once at position 4, if no part is being detected, the layer is changed, the snapshot position is offsetted, and the current pos is set to 1 so the robot will retake a picture from pos one on the next layer. A popup will appear asking to remove the interlayer.
The camera can tolerate approximately +-10% scaling when detecting an object. Therefore if the thickness of your object is relatively small, it is possible that the camera detects the part of the layer underneath if there is no part on the layer it is looking for and no interlayer to hide these parts. That's why I added interlayers. It might not be necessary if the thickness of your part is sufficient to avoid being within the +-10% scaling.