Research map / Image generation / Text-to-image diffusion
CoD: A Diffusion Foundation Model for Image Compression
Zhaoyang Jia, Zihan Zheng, Naifu Xue, Jiahao Li, Bin Li, Zongyu Guo, Xiaoyi Zhang, 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 |
|---|---|---|---|
| add auxiliary loss to flow matching loss (ablation) | CoD training · Kodak 512x512, 0.0039 bpp (training ablation) | PSNR: 9.83 dB -> 15.20 dB | Improved |
| add auxiliary loss to flow matching loss (ablation) | CoD training · Kodak 512x512, 0.0039 bpp (training ablation) | FID: 76.8 -> 48.3 | Improved |
| add unified training to flow matching loss (ablation) | CoD training · Kodak 512x512, 0.0039 bpp (training ablation) | PSNR: 9.83 dB -> 15.83 dB | Improved |
| add unified training to flow matching loss (ablation) | CoD training · Kodak 512x512, 0.0039 bpp (training ablation) | FID: 76.8 -> 48.4 | Improved |
| add unified post-training on top of auxiliary loss (ablation) | CoD training · Kodak 512x512, 0.0039 bpp (training ablation) | FID: 48.3 -> 46.0 | Improved |
| add unified post-training on top of auxiliary loss (ablation) | CoD training · Kodak 512x512, 0.0039 bpp (training ablation) | PSNR: 15.20 dB -> 16.20 dB | Improved |
| switch from V-prediction to X-prediction (ablation) | pixel-space CoD · Kodak 512x512, 0.0039 bpp (pixel-space prediction target) | PSNR: 16.21 dB -> 17.29 dB | Improved |
| switch from V-prediction to X-prediction (ablation) | pixel-space CoD · Kodak 512x512, 0.0039 bpp (pixel-space prediction target) | FID: 46.0 -> 44.8 | Improved |
| replace decoupled DDT head with JiT pure DiT (ablation) | pixel-space X-prediction CoD · Kodak 512x512, 0.0039 bpp (pixel-space prediction target) | PSNR: 17.29 dB -> 16.89 dB | Hurt |
| use X-prediction CoD inside DiffC (ablation) | DiffC with pixel-space CoD · Kodak 512x512 | FID and PSNR | Hurt |
| finetune MS-ILLM decoder with CoD-based DMD loss (ablation) | MS-ILLM · Kodak 512x512, 0.011 bpp (MS-ILLM decoder finetuning) | FID: 92.5 -> 80.92 | Improved |
| finetune MS-ILLM decoder with CoD-based DMD loss (ablation) | MS-ILLM · Kodak 512x512, 0.011 bpp (MS-ILLM decoder finetuning) | PSNR: 21.43 dB -> 21.08 dB | Mixed |
Closest papers by meaning
- OSCAR: One-Step Diffusion Codec Across Multiple Bit-rates (NeurIPS 2025 · similarity 0.91)
- Lossy Compression with Pretrained Diffusion Models (ICLR 2025 · similarity 0.90)
- PICD: Versatile Perceptual Image Compression with Diffusion Rendering (CVPR 2025 · similarity 0.90)
- Progressive Compression with Universally Quantized Diffusion Models (ICLR 2025 · similarity 0.90)
- One-Step Diffusion-Based Image Compression with Semantic Distillation (NeurIPS 2025 · similarity 0.89)
- Bridging the Gap between Gaussian Diffusion Models and Universal Quantization for Image Compression (CVPR 2025 · similarity 0.89)
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.
