SimpleTexture is one of the image-based methods for representing a 3D object. In this paper, a visual hull-based method for compressing a 3D object geometry is proposed when the object is represented by several SimpleTextures. To this end, a visual hull of an object is generated from depth maps of SimpleTextures. The generated visual hull is then utilized for the prediction of the depth maps and hence, spatial redundancy between depth maps is removed. Experimental results show that the proposed methods shows better compression ratio than solely utilizing 2D image encoders.