InstaVSR: Taming Diffusion for Efficient and Temporally Consistent Video Super-Resolution

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hu, Jintong, Chen, Bin, Hu, Zhenyu, Liu, Jiayue, Wang, Guo, Qi, Lu
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866914426492289024
author Hu, Jintong
Chen, Bin
Hu, Zhenyu
Liu, Jiayue
Wang, Guo
Qi, Lu
author_facet Hu, Jintong
Chen, Bin
Hu, Zhenyu
Liu, Jiayue
Wang, Guo
Qi, Lu
contents Video super-resolution (VSR) seeks to reconstruct high-resolution frames from low-resolution inputs. While diffusion-based methods have substantially improved perceptual quality, extending them to video remains challenging for two reasons: strong generative priors can introduce temporal instability, and multi-frame diffusion pipelines are often too expensive for practical deployment. To address both challenges simultaneously, we propose InstaVSR, a lightweight diffusion framework for efficient video super-resolution. InstaVSR combines three ingredients: (1) a pruned one-step diffusion backbone that removes several costly components from conventional diffusion-based VSR pipelines, (2) recurrent training with flow-guided temporal regularization to improve frame-to-frame stability, and (3) dual-space adversarial learning in latent and pixel spaces to preserve perceptual quality after backbone simplification. On an NVIDIA RTX 4090, InstaVSR processes a 30-frame video at 2K$\times$2K resolution in under one minute with only 7 GB of memory usage, substantially reducing the computational cost compared to existing diffusion-based methods while maintaining favorable perceptual quality with significantly smoother temporal transitions.
format Preprint
id arxiv_https___arxiv_org_abs_2603_26134
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle InstaVSR: Taming Diffusion for Efficient and Temporally Consistent Video Super-Resolution
Hu, Jintong
Chen, Bin
Hu, Zhenyu
Liu, Jiayue
Wang, Guo
Qi, Lu
Computer Vision and Pattern Recognition
Video super-resolution (VSR) seeks to reconstruct high-resolution frames from low-resolution inputs. While diffusion-based methods have substantially improved perceptual quality, extending them to video remains challenging for two reasons: strong generative priors can introduce temporal instability, and multi-frame diffusion pipelines are often too expensive for practical deployment. To address both challenges simultaneously, we propose InstaVSR, a lightweight diffusion framework for efficient video super-resolution. InstaVSR combines three ingredients: (1) a pruned one-step diffusion backbone that removes several costly components from conventional diffusion-based VSR pipelines, (2) recurrent training with flow-guided temporal regularization to improve frame-to-frame stability, and (3) dual-space adversarial learning in latent and pixel spaces to preserve perceptual quality after backbone simplification. On an NVIDIA RTX 4090, InstaVSR processes a 30-frame video at 2K$\times$2K resolution in under one minute with only 7 GB of memory usage, substantially reducing the computational cost compared to existing diffusion-based methods while maintaining favorable perceptual quality with significantly smoother temporal transitions.
title InstaVSR: Taming Diffusion for Efficient and Temporally Consistent Video Super-Resolution
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.26134