Advanced magnetic microscopies are key to advancing our understanding and application of novel magnetic phenomenon such as skyrmions, spinwaves, and domain walls. However, due to the diffraction-limit of light, achieving the 10 – 100 nanometer spatial resolution and 10 – 100 picosecond temporal resolution required to image these phenomena is beyond the reach of table-top techniques. My dissertation research has been to develop stroboscopic magnetic microscopy techniques that use picosecond thermal gradients to transduce magnetization into a voltage. In magnetic metals, this is accomplished via the anomalous Nernst effect and in ferromagnetic insulator/heavy metal bilayers the signal is due to the longitudinal spin Seebeck effect detected via the inverse spin Hall effect. Using focused, 3 ps laser pulses to heat cobalt and permalloy films, I demonstrate that the anomalous Nernst effect can image magnetization with 10-100 ps temporal resolution, sub-micron spatial resolution, and sensitivity to the in-plane moment of 0.1°/√Hz. I then show how this sensitivity and resolution can be applied for phase-sensitive ferromagnetic resonance imaging in ultrathin YIG/Pt bilayers in which we observe spatial variation of the resonance field, amplitude, phase, and linewidth. To conclude, I present the development of a near-field scanning optical microscope to create nanoscale thermal gradients and achieve spatial resolution of magnetic textures below the diffraction limit. The advent of these far- and near-field magneto-thermal imaging techniques will enable the table-top measurement of nanoscale magnetization dynamics in thin film devices.