An entertainment device generates a composite image with a combiner that combines camera-captured images with a computer-generated image of an object resting on a virtual surface. The device also includes a detector that detects image movement in the captured images in one or more contact point regions corresponding to image positions at which the object contacts the virtual surface. The device further comprises an initiator for initiating movement of the object to a new position with respect to the virtual surface in response to detected motion in the contact point regions. The detector detects whether a first image area corresponding to a captured image feature is greater than a predetermined proportion of a second image area corresponding to a full field of view of the camera. If the first image area is greater than the predetermined proportion, the initiator initiates movement of the object to an avoidance position.