A hypernetwork estimates parameters $\{\mathbf{b}_1, \mathbf{W}_2\}^{(i,j)}$ of pixel-wise, local neural heat fields. The phase shifts $\mathbf{b}_1$ operate on globally learned components, before thermal activations scale each component depending on their frequency and the desired scaling factor. The components are then linearly combined using coefficients $\mathbf{W}_2$, resulting in an appropriately-blurred, continuous local neural field. This field is then rasterized at the appropriate sampling rate (resolution) to yield a part of the final output image (red square). Unlike previous methods, correct anti-aliasing is guaranteed by design!
Due to its principled observation model, our method achieves state-of-the-art performance on a variety of super-resolution benchmarks.
If you found our work helpful, consider citing our paper 😊:
@article{becker2025thera,
title={Thera: Aliasing-Free Arbitrary-Scale Super-Resolution with Neural Heat Fields},
author={Becker, Alexander and Daudt, Rodrigo Caye and Narnhofer, Dominik and Peters, Torben and Metzger, Nando and Wegner, Jan Dirk and Schindler, Konrad},
journal={arXiv preprint arXiv:2311.17643},
year={2025}
}