Abstract
This work presents a comprehensive analysis of the load stepping method in digital photoelasticity, examined through four complementary perspectives: the classical formulation of load stepping, its extension using polarized-camera technology, a computational hybrid strategy, and exploratory deep learning approaches. Motivation arises from the need to quantify stress fields with greater robustness in contexts where experiments produce multiple fringe map displacements, as occurs under increasing load conditions. This approach aims to overcome the limitations of conventional phase-shifting techniques and lookup-table-based strategies while paving the way for new methodologies capable of generating multiple fringe displacements within a single-shot acquisition framework. The methodology combines synthetic simulations, experimental acquisitions with polarized array and color cameras, and a recent high-sensitivity technique for generating reference fringe-order maps. Demodulated stress fields are compared against these references using standard performance metrics such as MSE, PSNR, SSIM, and the maximum reliably recovered fringe order. The results demonstrate that while the classical approach remains theoretically sound, the polarized-camera update simplifies acquisition, the hybrid strategy reduces optical requirements through computational reconstruction, and deep learning shows promising adaptability to complex and irregular fringe patterns.