A deep convolutional neural network is trained using Hinode data that has been degraded to match the resolution and sensitivity of HMI observations. Once trained, the network can enhance HMI intensity and vector magnetic field data to the resolution and quality of Hinode.