AVSR-Diff: Scale-Agnostic Diffusion Priors for Temporally Consistent Arbitrary-Scale Video Super-Resolution
2026-07-01 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors propose AVSR-Diff, a new method for improving video super-resolution that works for any zoom level. They solve problems like blurry results at large zooms and flickering between video frames by separating how the model handles video details and scale adjustments. Their design uses a special module to keep video frames consistent over time and another to adjust detail frequencies based on the zoom level. Tests show their approach keeps sharp details and smooth videos better than previous flexible or fixed-scale methods.
Diffusion modelsVideo super-resolutionUpsampling scaleCoordinate-based methodsTemporal consistencyLatent denoisingVAE decoderFourier refinementGenerative priorsTemporal flickering
Authors
Geunhyuk Youk, Jeonghyeok Do, Dayeon Kim, Jihyong Oh, Munchurl Kim
Abstract
Diffusion models have significantly advanced video super-resolution (VSR) but remain largely constrained to fixed upsampling scales. Conversely, while coordinate-based arbitrary-scale VSR methods offer scale flexibility, they inherently suffer from severe over-smoothing at large scaling factors. Integrating generative priors with continuous decoding is promising but currently hindered by severe temporal flickering caused by the stochasticity of diffusion sampling. To address this, we propose AVSR-Diff (Arbitrary-scale Video Super-Resolution with Diffusion), a novel decoupled framework that separates scale-agnostic latent denoising from continuous coordinate rendering, effectively avoiding computationally heavy resolution-specific sampling. Our approach introduces a Temporally-Gated Feature Recurrence (TGFR) module to extract strictly aligned, temporally consistent latent priors. Furthermore, we design a continuous video VAE decoder incorporating a Scale-Aware Fourier Refinement (SAFR) module to dynamically adapt frequency components to any target scale. Extensive experiments demonstrate that AVSR-Diff consistently preserves high-frequency details and strong temporal stability across various scales, surpassing state-of-the-art arbitrary-scale baselines. Remarkably, our framework outperforms recent fixed-scale generative models even on their native resolution.