Spectral Prefiltering of Neural Fields
This addresses the resolution limitation in neural fields for visual signal representation, though it is incremental as it builds on existing neural field techniques.
The paper tackles the problem of neural fields operating at fixed resolutions by introducing a method for prefiltering them in a single forward pass, achieving faster training and inference with quantitative and qualitative improvements over existing methods.
Neural fields excel at representing continuous visual signals but typically operate at a single, fixed resolution. We present a simple yet powerful method to optimize neural fields that can be prefiltered in a single forward pass. Key innovations and features include: (1) We perform convolutional filtering in the input domain by analytically scaling Fourier feature embeddings with the filter's frequency response. (2) This closed-form modulation generalizes beyond Gaussian filtering and supports other parametric filters (Box and Lanczos) that are unseen at training time. (3) We train the neural field using single-sample Monte Carlo estimates of the filtered signal. Our method is fast during both training and inference, and imposes no additional constraints on the network architecture. We show quantitative and qualitative improvements over existing methods for neural-field filtering.