WEVSR: Video Diffusion Generators for Real-World Video Super-Resolution with Wavelet-Enhanced VAE Encoder

Yuying Chen, Zhirui Liu, Linyan Jiang, Qifan Gao, Xianguo Zhang, Jianhou Gan, Wenqi Ren
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:15574-15584, 2026.

Abstract

Recent advances in video diffusion models have demonstrated remarkable generative capability, yet adapting these large pretrained text-to-video (T2V) models to video super-resolution (VSR) typically encounters challenges, such as artifacts introduced by complex degradations in real-world scenarios and compromised fidelity due to the strong generative capacity of the powerful T2V models. We present WEVSR, a novel approach that adapts a pretrained flow-matching video diffusion transformer to VSR. First, we design a task-oriented adaptation strategy that leverages timestep sampling and noise augmentation to enhance detail restoration while preserving structural stability. Second, we propose a lightweight multi-level discrete wavelet transform (DWT) front-end for the VAE encoder, injecting explicit frequency priors into the latent space without modifying the pretrained decoder. Extensive experiments across multiple VSR benchmarks demonstrate that WEVSR achieves state-of-the-art performance against existing approaches. Code and models will be released here.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-chen26ce, title = {{WEVSR}: Video Diffusion Generators for Real-World Video Super-Resolution with Wavelet-Enhanced {VAE} Encoder}, author = {Chen, Yuying and Liu, Zhirui and Jiang, Linyan and Gao, Qifan and Zhang, Xianguo and Gan, Jianhou and Ren, Wenqi}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {15574--15584}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/chen26ce/chen26ce.pdf}, url = {https://proceedings.mlr.press/v306/chen26ce.html}, abstract = {Recent advances in video diffusion models have demonstrated remarkable generative capability, yet adapting these large pretrained text-to-video (T2V) models to video super-resolution (VSR) typically encounters challenges, such as artifacts introduced by complex degradations in real-world scenarios and compromised fidelity due to the strong generative capacity of the powerful T2V models. We present WEVSR, a novel approach that adapts a pretrained flow-matching video diffusion transformer to VSR. First, we design a task-oriented adaptation strategy that leverages timestep sampling and noise augmentation to enhance detail restoration while preserving structural stability. Second, we propose a lightweight multi-level discrete wavelet transform (DWT) front-end for the VAE encoder, injecting explicit frequency priors into the latent space without modifying the pretrained decoder. Extensive experiments across multiple VSR benchmarks demonstrate that WEVSR achieves state-of-the-art performance against existing approaches. Code and models will be released here.} }
Endnote
%0 Conference Paper %T WEVSR: Video Diffusion Generators for Real-World Video Super-Resolution with Wavelet-Enhanced VAE Encoder %A Yuying Chen %A Zhirui Liu %A Linyan Jiang %A Qifan Gao %A Xianguo Zhang %A Jianhou Gan %A Wenqi Ren %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-chen26ce %I PMLR %P 15574--15584 %U https://proceedings.mlr.press/v306/chen26ce.html %V 306 %X Recent advances in video diffusion models have demonstrated remarkable generative capability, yet adapting these large pretrained text-to-video (T2V) models to video super-resolution (VSR) typically encounters challenges, such as artifacts introduced by complex degradations in real-world scenarios and compromised fidelity due to the strong generative capacity of the powerful T2V models. We present WEVSR, a novel approach that adapts a pretrained flow-matching video diffusion transformer to VSR. First, we design a task-oriented adaptation strategy that leverages timestep sampling and noise augmentation to enhance detail restoration while preserving structural stability. Second, we propose a lightweight multi-level discrete wavelet transform (DWT) front-end for the VAE encoder, injecting explicit frequency priors into the latent space without modifying the pretrained decoder. Extensive experiments across multiple VSR benchmarks demonstrate that WEVSR achieves state-of-the-art performance against existing approaches. Code and models will be released here.
APA
Chen, Y., Liu, Z., Jiang, L., Gao, Q., Zhang, X., Gan, J. & Ren, W.. (2026). WEVSR: Video Diffusion Generators for Real-World Video Super-Resolution with Wavelet-Enhanced VAE Encoder. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:15574-15584 Available from https://proceedings.mlr.press/v306/chen26ce.html.

Related Material