Hi, I'm wondering whether your crazytalk software for generating the original 'model' of a face could be improved if it could more accurately recognize and identify the 'background'. Suppose you take the picture with a uniform background, e.g. posterboard. The posterboard could have printed on it four black circles well outside the face, but inside the picture. then, with the camera in the same position, you take another picture of the background only, i.e. no face. The purpose of the four black circles is to define a reference, so that the system can take account of small changes in camera position between the original picture and the background-only picture (ike the system already in place, you could ask the user to define the locations of the four black dots). In principle, it should be possible to combine the actual picture and the background-only picture to get a pretty good idea of what the background is. Of course, only you would know whether this would improve the remainder of the process of generating the facial model.