In the meantime, if I use an image anchor in a fixed position from a 3D object would that be a possible work around for the time being? Could i for example have an A4 paper sized image anchor on the ground in front of a real sofa and then use that as a reference to add AR objects to the sofa at a fixed location and scale?

Thanks for the clarity! Do you know if there is an API for object detection / object tracking coming out for visionOS anytime in the near future?