Context. Modern space telescopes are currently providing high-precision light curves for a large fraction of the sky, such that many new variable stars are being discovered. However, some stars have periodic variability with periods on the order of minutes and require high-cadence photometry to probe the physical mechanisms responsible. A cadence of less than a minute is often required to remove Nyquist ambiguities and confirm rapid variability, which forces observers to obtain high-cadence ground-based photometry. Aims. We aim to provide a modern software package to reduce ground-based photometric time series data and deliver optimised (differential) light curves. To produce high-quality light curves, which maximise the amplitude signal-to-noise ratio of short-period variability in a Fourier spectrum, we require adaptive elliptical aperture photometry as this represents a significant advantage compared to aperture photometry using circular apertures of fixed radii. Methods. The methodology of our code and its advantages are demonstrated using high-cadence ground-based photometry from the South African Astronomical Observatory (SAAO) of a confirmed rapidly oscillating Ap (roAp) star. Furthermore, we employed our software package to search for rapid oscillations in three candidate roAp stars. Results. We demonstrate that our pipeline represents a significant improvement in the quality of light curves, and we make it available to the community for use with different instruments and observatories. We search for and demonstrate the lack of high-frequency roAp pulsations to a limit of ∼1 mmag using B data in the three Ap stars HD 158596, HD 166542, and HD 181810. Conclusions. We demonstrate the significant improvement in the extraction of short-period variability caused by high-frequency pulsation modes, and discuss the implication of null detections in three Ap stars.