7 papers
OmniVR: Joint Video-Audio Conditional Generation for Restoring Degraded Historical Films
Xin Lu, Zihao Fan, Mingchen Zhong +3
Historical films suffer from co-occurring visual and audio degradations---blur, noise, flicker, hiss, clipping, and dropout---yet existing methods restore each modality independent…
Ultra Flash: Scaling Real-Time Streaming Video Generation to High Resolutions
Luxury, Jie Huang, Zihao Fan +25
While recent autoregressive video diffusion models achieve remarkable streaming quality, they remain confined to low resolutions (e.g., 480P), leaving efficient, scalable, real-tim…
IR-Flow: Bridging Discriminative and Generative Image Restoration via Rectified Flow
Zihao Fan, Xin Lu, Jie Xiao +3
In image restoration, single-step discriminative mappings often lack fine details via expectation learning, whereas generative paradigms suffer from inefficient multi-step sampling…
Bird-SR: Bidirectional Reward-Guided Diffusion for Real-World Image Super-Resolution
Zihao Fan, Xin Lu, Yidi Liu +4
Powered by multimodal text-to-image priors, diffusion-based super-resolution excels at synthesizing intricate details; however, models trained on synthetic low-resolution (LR) and…
FinPercep-RM: A Fine-grained Reward Model and Co-evolutionary Curriculum for RL-based Real-world Super-Resolution
Yidi Liu, Zihao Fan, Jie Huang +6
Reinforcement Learning with Human Feedback (RLHF) has proven effective in image generation field guided by reward models to align human preferences. Motivated by this, adapting RLH…
Elucidating and Endowing the Diffusion Training Paradigm for General Image Restoration
Xin Lu, Xueyang Fu, Jie Xiao +3
While diffusion models demonstrate strong generative capabilities in image restoration (IR) tasks, their complex architectures and iterative processes limit their practical applica…