A formula question for you another paper "Synthetic Training for Monocular Human Mesh Recovery."

[littlejiumi]

Recently, I read your other paper "Synthetic Training for Monocular Human Mesh Recovery." , which is also an excellent work. I am very interested in D2S projection. I agree that the formula (12),(13) make sense, however, the formula (8)-(11) maybe something wrong. Firstly, (10) is the \delta s, but why (13) is s_i? Secondly, the second "=" in both (10) and (11) are wrong. An I wonder why \delta s = f_x/d - f_x*z_i/ d+z_i, I think it should be \delta s = f_x/d - f_x/ d+z_i. I would very much like to get the correct version of these formulas. Thank you so much!