Unified Generative Video Compression with Autoregressive and Bidirectional Video Diffusion
To support both stages, we present UniGVC, a Unified Generative Video Codec that supports two decoding modes from a single bitstream.
ProofPaper ↗
Key points
- Video diffusion models can reconstruct visually compelling content from severely compressed representations at ultra-low bitrates, but they impose a rigid deployment trade-off.
- Bidirectional (BI) models achieve high quality through dense full-sequence attention, yet require the entire sequence before reconstruction and incur substantial latency.
- Autoregressive (AR) models enable low-delay streaming, but their restricted temporal context reduces compression efficiency.
- UniGVC-BI employs a BI-DiT with dense attention for quality-oriented random-access decoding, whereas UniGVC-AR uses windowed AR-DiT for low-delay streaming.
Sources (1)
- [1]Unified Generative Video Compression with Autoregressive and Bidirectional Video DiffusionarXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 05:25 AM
To support both stages, we present UniGVC, a Unified Generative Video Codec that supports two decoding modes from a single bitstream.
Video diffusion models can reconstruct visually compelling content from severely compressed representations at ultra-low bitrates, but they impose a rigid deployment trade-off.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 9, 2026Qwen/Qwen-Image-2.1-Turbo
- Oct 8, 2026Conditional Residual Prediction: Improving Autoregressive Video Diffusion without a Bidirectional Teacher
- Oct 8, 2026Bernoulli Flow Models: Self-Consistent Generative Modeling for Binary Data
- Oct 8, 2026No Distillation Needed: Single-Pass Real-Time Talking Heads via Acausal Noise Shaping
- Oct 8, 2026Normalizing Trajectory Models
- Oct 7, 2026Enabling Preference-driven Unlearning in Few-step Distilled Text-to-Image Diffusion Models