Transformer-based architectures have recently advanced the image reconstruction quality of super-resolution (SR) models. Yet, their scalability remains limited by quadratic attention costs and coarse patch embeddings that weaken pixel-level fidelity. We propose TaylorIR, a plug-and-play framework that enforces 1x1 patch embeddings for true pixel-wise reasoning and replaces conventional self-attention with TaylorShift, a Taylor-series-based attention mechanism enabling full token interactions with near-linear complexity. Across multiple SR benchmarks, TaylorIR delivers state-of-the-art performance while reducing memory consumption by up to 60%, effectively bridging the gap between fine-grained detail restoration and efficient transformer scaling.
@article{arxiv.2411.10231,
title = {A Low-Resolution Image is Worth 1x1 Words: Enabling Fine Image Super-Resolution with Transformers and TaylorShift},
author = {Sanath Budakegowdanadoddi Nagaraju and Brian Bernhard Moser and Tobias Christian Nauen and Stanislav Frolov and Federico Raue and Andreas Dengel},
journal= {arXiv preprint arXiv:2411.10231},
year = {2025}
}