A method to create realistic, controllable, human body models is discussed in this paper. Coordinate data is acquired using a range scanner, and is presented to a modeling synthesizer that generates the whole body geometry using size-related control parameters. The synthesizer essentially manipulates the body geometry of an existing template model, to create a realistic model of the acquired data.

The underlying principle of the technique is shape interpolation using radial basis functions (RBFs). However, the authors did not discuss any literature on shape interpolation or morphing using RBFs [1], or other techniques of shape interpolation [2,3,4]. References related to anthropometric measurements [5] were not included.

The method discussed is similar to that proposed by Popovic et al. [6]. In section 1.1 (page 4), it may be easier to read about the existing method [6] first, followed by its comparison to the proposed method. The difference appears to be in the sizing parameters, or anthropometric measurements, that assist in shape interpolation between different models. Eight sizing parameters (page 9) and 33 joints (page 6) were selected. However, it is unclear how this translates to an R8->R25 (page 11) mapping in the joint synthesizer. Issues concerning feature matching for correspondence and registration, prior to shape interpolation, are not discussed. Nonetheless, the concept of using anthropometric measures, for human body manipulation, is very interesting, and does have potential in the human body modeling domain.