Search papers, labs, and topics across Lattice.
This paper introduces Generative Transmission (GenTrans), a novel approach to video communication that shifts the focus from fidelity-oriented signal reconstruction to optimizing for perceptual utility and system usability under ultra-low-bandwidth and weak-network conditions. By framing video transmission as a joint optimization problem that incorporates bandwidth, computation, and memory, GenTrans leverages generative models and cross-clip memory reuse to enhance transmission efficiency and robustness. Experimental results demonstrate that GenTrans achieves significant improvements in transmission and decoding efficiency while maintaining high perceptual quality, outperforming traditional methods in challenging network environments.
GenTrans redefines video communication by prioritizing perceptual utility over pixel-perfect fidelity, achieving remarkable efficiency in ultra-low-bandwidth scenarios.
Under the AI Flow framework, communication is shifting from transmitting fidelity-oriented information flows toward delivering task-oriented and perception-oriented token flows across heterogeneous network resources. Video communication is a fundamental component of modern information networks. However, under ultra-low-bandwidth and weak-network conditions, conventional video coding and transmission methods, which are primarily optimized for pixel-level fidelity, often struggle to balance visual usability, transmission efficiency, and robustness to unstable links. With the rapid advancement of generativemodels, video communication is also moving from precise signal reconstruction toward receiver-side perceptual utility and system-level usability. In this paper, we propose Generative Transmission (GenTrans) for video communication under ultra-low-bandwidth and weak-network conditions. Built upon Generative Video Compression (GVC), GenTrans formulates video transmission as a joint optimization problem involving bandwidth, computation, and memory, rather than treating it merely as a signal coding task. By leveraging generative priors, cross-clip memory reuse, runtime state reuse, and weak-network-aware transport, GenTrans significantly reduces transmission overhead while enabling visually coherent and practically useful reconstruction. Experimental results show that GenTrans supports effective video transmission under ultra-low-bitrate and weak-network conditions, achieving improved transmission efficiency, decoding efficiency, and robustness while preserving perceptual quality.
