Three different capturing setups are realized, each with 5 Kinects in a circular configuration. The third capturing setup is the one used in the “Huawei/3DLife ACM Multimedia Grand Challenge for 2013” (http://mmv.eecs.qmul.ac.uk/mmgc2013/).   For each dataset, two reconstruction methods are considered: A variant of the explicit fusion method described in [1], as well as the volumetric Poisson reconstruction method [2]. The first method runs in real-time (~10 fps). The second one is slower, but produces smooth and watertight reconstructions.   The datasets are given as ASCII-PLY files (one for each frame, including vertex positions, normals and colors per vertex). All frames are compressed into a .rar file.  

[1] D. Alexiadis, D. Zarpalas, P. Daras, "Real-time, full 3-D reconstruction of moving foreground objects from multiple consumer depth cameras", IEEE Transactions on Multimedia, VOL. 15, NO. 2, pp. 339-358, February 2013  

[2] M. Kazhdan, M. Bolitho, and H. Hoppe, “Poisson surface reconstruction,” in Proc. of the 4th Eurographics symposium on Geometry processing, SGP ’06, Switzerland, 2006, pp. 61–70.