Time curves visualise patterns of temporal evolution in data by projecting high-dimensional time-dependent data to chronologically connected points in a two-dimensional space [1]. They are applicable to a variety of data types and for this reason, this thesis focuses on the classes video data and flow data. In this thesis, time curves are examined by implementing a flexible, interactive and explorative framework. This includes the implementation of time curves in a novel way, i.e., with an adapted stress-based formulation (loss function) that is solved numerically via gradient descent. The framework consists of a barycentric triangle, which is used to determine a custom loss function for creating time curves. This custom loss function comprises the sum of three variations of a loss function, to which the user assigns individual weights. For the evaluation, the time curves for seven edge cases are arranged according to the position of their custom loss function in the barycentric triangle. This facilitates comparing the time curves and visually emphasises the influences of the corresponding stress functions. After evaluating each visualisation individually, the best time curve is selected based on the expected patterns and then compared to the conventional time curve of the same data set. The evaluations shows that this method leads to improved time curves, which visualise desired properties, such as emphasising the temporal component and avoiding visual clutter.