PulseAugur
LIVE 12:24:36
research · [1 source] ·
0
research

GVCC uses generative models for zero-shot video compression

Researchers have introduced GVCC, a novel zero-shot framework for video compression that leverages pretrained generative video models as decoders. This approach enables high-fidelity reconstruction at ultra-low bitrates by transmitting a generation trajectory rather than relying on traditional regression methods. GVCC converts deterministic flow samplers into stochastic processes, allowing information to be encoded via per-step innovations, and is demonstrated in text-to-video, autoregressive image-to-video, and first-last-frame-to-video configurations. AI

Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →

IMPACT Presents a new method for high-fidelity video compression at ultra-low bitrates using generative models.

RANK_REASON This is a research paper detailing a new method for video compression.

Read on arXiv cs.AI →

COVERAGE [1]

  1. arXiv cs.AI TIER_1 · Ziyue Zeng, Xun Su, Haoyuan Liu, Bingyu Lu, Yui Tatsumi, Hiroshi Watanabe ·

    GVCC: Zero-Shot Video Compression via Codebook-Driven Stochastic Rectified Flow

    arXiv:2603.26571v3 Announce Type: replace-cross Abstract: At ultra-low bitrates, high-fidelity reconstruction requires sampling plausible videos from the posterior rather than regressing to oversmoothed conditional means. We propose Generative Video Codebook Codec (GVCC), a zero-…