About "x_0" and "x_next" in the diffusion.py

[shrxht]

                x0 = self.pred_x0_from_xt(x_t, noise_pred, C_pred, cur_time)
                x0.clamp_(-1., 1.)
                C_pred = -1 * x0
                x_next = self.pred_xtms_from_xt(x_t, noise_pred, C_pred, cur_time, s)

why traj[t-1] = x_next.detach(), why not traj[t-1] = x_0.detach()

[GuHuangAI]

@shrxht
You can seem x0 as an estimation in each denoising step, and x_next is the real denoised signal.