Research map / Video / Video generation
Next-Frame Decoding for Ultra-Low-Bitrate Image Compression with Video Diffusion Priors
Yunuo Chen, Chuqin Zhou, Jiangchuan Li, Xiaoyue Ling, Bing He, Jincheng Dai, Li Song, Guo 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 |
|---|---|---|---|
| add Stage I interval adaptation (ablation) | NeFIC · DIV2K | BD-rate LPIPS: -52.32 -> -61.71 | Improved |
| add Stage II one-step bypass (ablation) | NeFIC · DIV2K | BD-rate LPIPS/FID: LPIPS -47.01 -> -61.71; FID -53.72 -> -49.94 | Mixed |
| add conditional anchor encoding (ablation) | NeFIC · DIV2K | BD-rate LPIPS: -56.81 -> -61.71 | Improved |
| add bypass refinement (ablation) | NeFIC · DIV2K | BD-rate LPIPS: -48.01 -> -61.71 | Improved |
| add auxiliary anchor loss (ablation) | NeFIC · DIV2K | BD-rate LPIPS: -19.78 -> -61.71 | Improved |
| add CLIP-based GAN loss (ablation) | NeFIC · DIV2K | BD-rate FID/LPIPS: FID -30.31 -> -49.94; LPIPS -63.09 -> -61.71 | Mixed |
| raise LoRA rank/alpha from 64/64 to 256/256 (ablation) | NeFIC · DIV2K | BD-rate LPIPS: -44.18% -> -61.71% | Improved |
| raise LoRA rank/alpha from 64/64 to 256/256 (ablation) | NeFIC · DIV2K | BD-rate FID: -59.91 -> -49.94 | Hurt |
| add Stage II one-step generation (vs 50-step denoising) (ablation) | NeFIC · DIV2K | decoding time on Kodak (s): 30.61 -> 0.88 | Improved |
| add Stage II one-step generation (vs 50-step denoising) (ablation) | NeFIC · DIV2K | decoding time on DIV2K (s): 296.02 -> 6.83 | Improved |
| use CogVideoX VDM with anchor instead of Flux token concatenation (ablation) | NeFIC · DIV2K | BD-rate LPIPS: -42.18 -> -61.71 | Improved |
| add anchor token concatenation to Flux-dev-12B (vs no anchor) (ablation) | Flux-dev-12B codec · DIV2K | BD-rate LPIPS: -32.01 -> -42.18 | Improved |
Closest papers by meaning
- Ultra Lowrate Image Compression with Semantic Residual Coding and Compression-aware Diffusion (ICML 2025 · similarity 0.91)
- Towards image compression with perfect realism at ultra-low bitrates (ICLR 2024 · similarity 0.91)
- Decouple Distortion from Perception: Region Adaptive Diffusion for Extreme-low Bitrate Perception Image Compression (CVPR 2025 · similarity 0.88)
- Bridging the Gap between Gaussian Diffusion Models and Universal Quantization for Image Compression (CVPR 2025 · similarity 0.88)
- OSCAR: One-Step Diffusion Codec Across Multiple Bit-rates (NeurIPS 2025 · similarity 0.88)
- Progressive Compression with Universally Quantized Diffusion Models (ICLR 2025 · similarity 0.88)
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.
