Research map / Video / Video segmentation and tracking
Generative Video Compression with One-Dimensional Latent Representation
Zihan Zheng, Zhaoyang Jia, Naifu Xue, Jiahao Li, Bin Li, Zongyu Guo, Xiaoyi Zhang, Zhenghao Chen, Houqiang Li, Yan Lu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use 1D latent tokens with 1D memory vs GLC-Video | generative video codec · HEVC-B | BD-rate (LPIPS): -60.4% | Improved |
| use GVC1D vs GLC-Video for fidelity | generative video codec · HEVC-B | BD-rate (PSNR): -53.8% PSNR, -45.1% MS-SSIM | Improved |
| add long-term 1D memory (ablation) | GVC1D · UVG | BD-rate (LPIPS): 40.6% -> 0.0% | Improved |
| use 1D memory instead of 2D memory (ablation) | GVC1D · UVG | BD-rate (LPIPS): 16.8% -> 0.0% | Improved |
| add autoregressive entropy model (ablation) | GVC1D without memory · HEVC-B | BD-rate (LPIPS): 67.8% -> 20.1% | Improved |
| use 32x16 tokens instead of 16x32 (ablation) | GVC1D · HEVC-B | BD-rate (LPIPS): 41.7% -> 0.0% | Improved |
| use 32x16 tokens instead of 16x64 (ablation) | GVC1D · HEVC-B | BD-rate (LPIPS): 20.1% -> 0.0% | Improved |
| use 32x16 tokens instead of 32x32 (ablation) | GVC1D · MCL-JCV | BD-rate (LPIPS): 10.2% -> 0.0% | Improved |
| use 32x16 tokens instead of 64x16 (ablation) | GVC1D · UVG | BD-rate (LPIPS): 22.8% -> 0.0% | Improved |
| use GVC1D instead of GLC-Video | video codec decoding · 1080P decoding time | seconds: 0.183 -> 0.207 | Hurt |
Closest papers by meaning
- Generative Neural Video Compression via Video Diffusion Prior (CVPR 2026 · similarity 0.89)
- Generative Latent Coding for Ultra-Low Bitrate Image Compression (CVPR 2024 · similarity 0.89)
- GVCC: Zero-Shot Video Compression via Codebook-Driven Stochastic Rectified Flow (NeurIPS 2026 · similarity 0.87)
- Augmented Deep Contexts for Spatially Embedded Video Coding (CVPR 2025 · similarity 0.87)
- One-Step Diffusion-Based Image Compression with Semantic Distillation (NeurIPS 2025 · similarity 0.86)
- When Worse is Better: Navigating the Compression Generation Trade-off In Visual Tokenization (NeurIPS 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
