Search papers, labs, and topics across Lattice.
This paper introduces LiveVVT, a novel rolling streaming diffusion framework for high-fidelity video virtual try-on (VVT) that addresses the latency and computational challenges of existing methods. By maintaining bounded bidirectional modeling within a causal recurrent generation process, LiveVVT effectively preserves local interactions while generating video chunks in real time. Experimental results show that LiveVVT achieves superior generation quality with significantly reduced latency and increased throughput compared to similar-sized models, enabling practical applications in real-time VVT scenarios.
Achieving high-fidelity video virtual try-on in real time, LiveVVT reduces latency by 26x while enhancing generation quality.
Diffusion-based Video Virtual Try-On (VVT) achieves high visual fidelity through bidirectional spatio-temporal modeling, but complete-clip dependence incurs prohibitive latency and computational overhead in practical continuous deployment. Naively enforcing causality disrupts pretrained bidirectional priors and substantially degrades synthesis quality. We introduce LiveVVT, a rolling streaming diffusion framework that preserves bounded bidirectional modeling within causal recurrent generation. Within a fixed-size window, LiveVVT jointly denoises multiple video chunks under bounded look-ahead, preserving local bidirectional interactions while emitting one clean chunk per iteration. Beyond the window, two complementary memories sustain long-term consistency: a bounded temporal memory propagates recent dynamics and occlusion context, whereas a persistent global appearance memory, constructed once from the target garment and a frontal try-on keyframe, anchors garment details and dressed appearance throughout the stream. We further introduce a progressive distillation framework integrating bidirectional VVT learning, teacher-trajectory regression for causal few-step adaptation, and Collaborative Matching Distillation, which couples teacher-distribution matching with rolling flow matching on real videos to align optimization with recurrent inference. Experiments on paired and unpaired long-sequence benchmarks demonstrate superior generation quality over similarly sized models, with $26\times$ lower latency and $11\times$ higher throughput, enabling high-fidelity real-time streaming VVT.
