Thank you for sharing the code!
I have some problem understanding the deterministic generative process backward (the reverse of Equation 1). It seems like a deterministic process without any randomness, and once x_0 and zsem is determined, x_T is determined with the forward iteration.
So how can I get the results in Figure 3 (Reconstruction results and the variations induced by changing the stochastic subcode xT)? Do the results in figure 3 use random code as x_T, or just the regular add_noise process of DDIM?