Images representing the amplitude of the Fourier transform of the ground truth image (FT – GT), and of the predictions obtained with our latent diffusion model (FT – LDM), with the progressive model (FT – PROG), and with the enhance model (FT – ENHANCE). The red circles indicate the Fourier frequencies corresponding to a spatial resolution of 2^′′. This visualization highlights how each model predicts or blurs high-frequency features, which are crucial for capturing fine details.