In phase mask-based lensless cameras, the mask placed in front of the image sensor determines the transfer function of the imaging system. The microscopic structures in the phase masks create a two-dimensional point-spread function (PSF) which is shift-invariant, and the imaging system’s forward model can be modeled as a convolution between the scene and the 2D PSF like below. With known PSF, the blurry raw measurement from the image sensor can be computationally deconvolved to reconstruct the image of the

scene. Thus, in assembling a lensless camera, it is essential to place the phase mask at the proper distance from the image sensor such that the mask creates sharp and high-contrast PSF, which is crucial for the reconstruction of high-quality images.

    To compute the optimal thickness of the replicated mask layer, we built a simple ray-tracing model using ray transfer matrix analysis. By replicating the master phase mask’s surface structure directly on the image sensor, our method further simplifies the fabrication of lensless cameras and delivers a rigid and durable device with a small form factor.

    Our prototype device has an open-faced design without any apertures and generates high-quality photographic reconstructions with high light collection efficiency. We analyze the performance of our prototype device and demonstrate various imaging applications, including the digital refocusing capabilities using depth-dependent PSFs.