The paper discusses a method for non-invasive imaging through opaque scattering layers, focusing on retrieving the shape of an object hidden behind a diffusing screen. The authors highlight the challenges posed by light scattering, which often results in blurred images, making it difficult to image through thick clouds or deep into biological tissues. They introduce a technique that leverages the optical memory effect, where the speckle pattern generated by coherent light scattered by a scattering layer is correlated with the pattern generated by the same layer at different angles of incidence. This correlation allows for the measurement of the autocorrelation of the hidden object's fluorescence, which can be inverted to reconstruct the object's shape using an iterative algorithm like the Error Reduction method. The method is demonstrated through synthetic data and shown to be effective even with limited information about the scattering layer. The paper also addresses potential limitations and pitfalls, such as the loss of absolute position information and the need for ensemble averaging to improve accuracy.The paper discusses a method for non-invasive imaging through opaque scattering layers, focusing on retrieving the shape of an object hidden behind a diffusing screen. The authors highlight the challenges posed by light scattering, which often results in blurred images, making it difficult to image through thick clouds or deep into biological tissues. They introduce a technique that leverages the optical memory effect, where the speckle pattern generated by coherent light scattered by a scattering layer is correlated with the pattern generated by the same layer at different angles of incidence. This correlation allows for the measurement of the autocorrelation of the hidden object's fluorescence, which can be inverted to reconstruct the object's shape using an iterative algorithm like the Error Reduction method. The method is demonstrated through synthetic data and shown to be effective even with limited information about the scattering layer. The paper also addresses potential limitations and pitfalls, such as the loss of absolute position information and the need for ensemble averaging to improve accuracy.