Fig. 1: The forward diffusion q(zt∣zt−1) (dashed arrows) progressively degrades the GroundTruth mural image z0.
From: Diffusion-based heterogeneous network for ancient mural restoration
Guided by the latent space learned from the distribution of GroundTruth images, the conditional reverse diffusion pθ(zt−1∣zt, x) (solid arrows) iteratively estimates and removes the injected noise, restoring the degraded mural image x.
