Inspired by the recent advances in implicitly representing signals with\ntrained neural networks, we aim to learn a continuous representation for\nnarrow-baseline 4D light fields. We propose an implicit representation model\nfor 4D light fields which is conditioned on a sparse set of input views. Our\nmodel is trained to output the light field values for a continuous range of\nquery spatio-angular coordinates. Given a sparse set of input views, our scheme\ncan super-resolve the input in both spatial and angular domains by flexible\nfactors. consists of a feature extractor and a decoder which are trained on a\ndataset of light field patches. The feature extractor captures per-pixel\nfeatures from the input views. These features can be resized to a desired\nspatial resolution and fed to the decoder along with the query coordinates.\nThis formulation enables us to reconstruct light field views at any desired\nspatial and angular resolution. Additionally, our network can handle scenarios\nin which input views are either of low-resolution or with missing pixels.\nExperiments show that our method achieves state-of-the-art performance for the\ntask of view synthesis while being computationally fast.\n