Accurate single-shot full-Stokes detection enabled by heterogeneous grain orientations in polycrystalline films

Decoding arbitrary polarization information in a cost-effective way is a key target for next-generation optical sensing. However, the design of full-Stokes detectors capable of resolving polarization states in a single shot remains challenging. Here we introduce GOStokes, an approach that leverages heterogeneous grain orientation in solution-processed metal halide semiconductors to extract Stokes parameters in a single measurement. By developing polycrystalline films exhibiting strong inherent circular and linear dichroism, we harness randomly oriented grains to produce varied polarization selectivity across the spatial domain. Integrating these films as multi-channel optical filters with commercial cameras enables real-time polarimetric detection and imaging, each generating a transmission map from a single exposure. Using deep learning, GOStokes precisely determines arbitrary polarization states with an averaged mean absolute error below 1%. Our demonstration underscores the potential of combining low-cost, scalable, polycrystalline films with reconstruction algorithms for advanced polarimetric applications.