Closed lizhiqi49 closed 8 months ago
Hi, thanks for your interest :) You would lose some of the fine details. This becomes more evident when optimizing the 2D latent space of SD, where the absence of fine details in the background becomes more noticeable.
thanks for your reply :)
Sure, closing this issue for now
What if the approximation is applied for all timesteps (including t<200)?