Search papers, labs, and topics across Lattice.
ReGenVC is an innovative end-to-end generative video codec that achieves real-time decoding of talking-head videos at an ultra-low bitrate of approximately 26 kB for a 77-frame sequence. By utilizing a neurally compressed first frame, per-frame pose keypoints, and a four-step distilled diffusion transformer, it significantly outperforms traditional codecs like x264/x265, which require 250-280 kB for similar quality. The system maintains sharp video quality at low bitrates while achieving 24 fps output on an 8-GPU node, making it suitable for live streaming applications without frame underruns.
ReGenVC achieves artifact-free video reconstruction at one-tenth the bitrate of traditional codecs, enabling real-time streaming without frame drops.
We present ReGenVC, an end-to-end generative video codec that compresses talking-head video to an ultra-low bitrate and decodes it in real time. The encoder reduces a source clip to a compact bitstream -- a neurally compressed first frame, per-frame pose keypoints, and metadata -- totaling about 26 kB for a 77-frame sequence. The decoder is a four-step distilled diffusion transformer that reconstructs the video conditioned on the transmitted pose and reference frame. Compared with x264/x265, ReGenVC reduces the bitrate to roughly one tenth of that required by traditional codecs (about 26 kB vs. 250--280 kB for essentially artifact-free reconstruction); at a matched ultra-low bitrate, conventional codecs collapse into blocking artifacts while ReGenVC stays sharp by exploiting a strong generative prior. The central obstacle to deploying such a codec is decoder latency: multi-step sampling with transformer and VAE components is too slow for interactive use. We make the decoder real-time through four-step distillation and three model-preserving system techniques: (i) eight-GPU unified sequence parallelism (Ulysses&Ring), (ii) a spatially-split VAE, and (iii) a three-stage overlapped pipeline; an analytical timing model characterizes the real-time feasibility region. On an 8-GPU node, the system sustains 24 fps output (972 ms per 25-frame window, within the 1000 ms budget), enabling a live browser stream without observed frame underruns. A hybrid CPU-GPU deployment further runs the encoder on the CPU at 24 fps and offloads the decoder-side one-shot conditioning encoders to the CPU, reducing the per-GPU memory peak from 21.1 GB to about 7.7 GB. To our knowledge, ReGenVC is the first end-to-end generative video codec to combine ultra-low-bitrate encoding with real-time decoding on an 8-GPU system.
