This paper proposes a modified algorithm that predicts filament usage in humanmanikin three-dimensional (3D) printing. In 3D printing, the use of support structuresowing to overhangs leads to an increased printing time, filament length, and final surfaceruggedness, which can be minimized by selection correct choice regarding object printingorientation. To predict the amount of filament needed in 3D printing, we use the shadowprojectionidea employed in previous works. We modify the original idea so that trianglesare classified into alpha and beta groups with respect to their normal vector. Then, thegroups are converted into alpha and beta voxels, respectively. Finally, the voxels that liehighest in the Z-direction are copied to top-covering voxels. In this way, the final volume ofthe filament can be easily calculated using simple linear algebra. For ease of modeling, weassumed that the internal volume and the support structure were 100% filled. The proposedalgorithm is applied to the famous Stanford Bunny data and some manikin parts.