aiqbench analysis
LiveVVT addresses the latency bottleneck in diffusion-based video virtual try-on by introducing a rolling streaming framework that maintains high visual fidelity through bounded bidirectional modeling. By utilizing complementary memory mechanisms for temporal dynamics and global appearance, the system achieves 26x lower latency and 11x higher throughput compared to complete-clip dependence models. This advancement enables practical, real-time deployment of high-quality garment try-on applications without sacrificing synthesis quality.
LiveVVT: High-Fidelity Video Virtual Try-On in Real Time
arXiv:2608.26714v1 Announce Type: cross Abstract: 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.
Abstract: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.
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2608.26714 [cs.CV]
(or arXiv:2608.26714v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2608.26714 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 290 words
Related coverage
[AINews] not much happened today
Latent Space · Oct 3, 2026
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
At SIGGRAPH, NVIDIA Advances Graphics and Simulation With Agentic and Physical AI
NVIDIA · Jul 20, 2026
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · Oct 13, 2026