InfVSR: Toward Consistency-Driven Streaming Generative Video Super-Resolution

Real-world videos often extend over thousands of frames. Existing generative video super-resolution (VSR) approaches, however, face two persistent challenges when processing long sequences: (1) inefficiency due to the heavy cost of multi-step denoising for full-length sequences; and (2) poor consistency hindered by temporal decomposition that causes artifacts and discontinuities. To break these limits, we propose InfVSR, which reformulates VSR as an autoregressive-one-step-diffusion paradigm, and enables streaming inference with video diffusion priors. First, we adapt the pretrained DiT into a causal structure, maintaining both local and global coherence via rolling KV-cache and joint visual guidance. Second, we distill the diffusion process into a single step efficiently, with patch-wise pixel supervision and cross-chunk distribution matching. To fill the gap in long-form video evaluation, we build a new benchmark tailored for extended sequences and further introduce semantic-level metrics to comprehensively assess temporal consistency. Our method pushes the frontier of long-form VSR, achieves state-of-the-art quality with enhanced semantic consistency, and delivers up to 58x speed-up over existing methods such as MGLD-VSR. Our code and models are available at https://github.com/Kai-Liu001/InfVSR .

This paper introduces InfVSR, a new method for enhancing long videos efficiently and consistently. Instead of processing an entire video at once, InfVSR restores the video piece by piece while carrying useful information forward from previous parts. This allows it to handle very long videos with stable memory usage. The method also uses shared visual references to keep objects, backgrounds, and motion more consistent over time. To better test long-video enhancement, this work also builds a new benchmark containing 1000-frame real-world videos. Experiments show that InfVSR produces high-quality and temporally consistent results while being much faster than previous generative video restoration methods, making long-video super-resolution more practical for real applications.