Space photometry has revealed flaring activity on many types of stars across the Hertzsprung-Russell diagram. With the abundance of data provided by Kepler and TESS, it is now possible to study the characteristics of stellar flares in greater detail, beyond the simple parameterization by amplitude and duration. Here we use the FLATW’RM2 deep learning based tool to identify flares on all available Kepler and TESS short cadence light curves, then use principal component analysis to explore the temporal morphology of flares, looking for systematic differences in their shapes after scaling, and also trying to link these differences to the basic astrophysical properties of the stars.