The field of depth estimation and 3D perception is witnessing significant advancements with the integration of large-scale pretrained models and innovative architectures. Researchers are exploring the potential of pretrained depth representations to improve image dehazing and depth completion tasks. Furthermore, there is a growing interest in developing unified frameworks that can bridge the gap between monocular and stereo depth estimation, enabling more robust and accurate 3D perception. Noteworthy papers in this area include: OmniDepth, which introduces a cross-attentive alignment mechanism to harmonize multi-view geometry with monocular context, achieving state-of-the-art results on several benchmarks. DualMat, which presents a novel dual-path diffusion framework for estimating Physically Based Rendering materials from single images, achieving significant improvements in albedo estimation and metallic-roughness prediction errors.