This chapter presents methods for building graph Fourier transforms (GFTs) for image and video compression. A key insight is that classical transforms, such as the discrete sine/cosine transform (DCT) or the Karhunen–Loeve transform (KLT), can be interpreted from a graph perspective. The chapter considers two sets of techniques for designing graphs, from which the associated GFTs are derived: Graph learning oriented GFT (GL‐GFT), and Block‐adaptive GFT. The graph spectral approaches aim to find graph Laplacian matrices, which denote the inverse covariances for the models of interest. The chapter discusses more specific 1D line models, with rigorous derivations of two separate Gaussian Markov random fields for intra‐ and inter‐predicted blocks. The experimental results demonstrated that GL‐GFTs can provide considerable coding gains with respect to standard transform coding schemes using/DCT. In comparison with the KLTs obtained from sample covariances, GL‐GFTs are more robust and provide better generalization.