Research map / Image generation / Fast diffusion sampling
DICT: Data Injection and Contrastive Trajectory Refinement for Conditional Image Generation with Diffusion Models
Chunnan Shang, Zhizhong Wang, Xin Zhang, Hongwei Wang
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 Data Injection (noise-perturbed condition injected in early steps) (ablation) | DICT · Text-driven style transfer | Style Loss and CLIP Loss | Improved |
| add Data Injection (ablation) | DICT · Text-driven style transfer | Text Score | No clear effect |
| add Data Injection (ablation) | DICT · FFHQ super-resolution and deblurring | PSNR, SSIM, LPIPS | Improved |
| add Contrastive Trajectory Refinement (ablation) | DICT · Style transfer, super-resolution and deblurring | quality metrics | Improved |
| use DICT instead of TFG on text score | DICT · Style transfer | Text Score: 0.3092 -> 0.2952 | Hurt |
| use DICT instead of StyleShot | DICT · Style transfer | Style Loss: 0.6747 -> 0.6313 | Improved |
| use DICT instead of StyleShot | DICT · Style transfer | CLIP Loss: 6.0415 -> 4.2334 | Improved |
| use DICT instead of FPS-SMC | DICT · FFHQ super-resolution | PSNR: 27.6305 -> 28.8600 | Improved |
| use DICT instead of FPS-SMC | DICT · FFHQ super-resolution | SSIM: 0.8283 -> 0.8233 | Hurt |
| use DICT instead of DCDP | DICT · FFHQ deblurring | PSNR: 27.9110 -> 27.5794 | Hurt |
| use DICT instead of DCDP | DICT · FFHQ deblurring | LPIPS: 0.2325 -> 0.2236 | Improved |
| vary Data Injection steps T1 too low or too high (ablation) | DICT · Style transfer | visual quality | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 5 of 5 on most shared results.
- StyleStudio — StyleStudio: Text-Driven Style Transfer with Selective Control of Style Elements (CVPR 2025) · better on 3 of 3 shared results
- CSGO — CSGO: Content-Style Composition in Text-to-Image Generation (NeurIPS 2025) · better on 3 of 3 shared results
- SITCOM — SITCOM: Step-wise Triple-Consistent Diffusion Sampling For Inverse Problems (ICML 2025) · better on 1 of 1 shared result
- FlowDPS — FlowDPS : Flow-Driven Posterior Sampling for Inverse Problems (ICCV 2025) · better on 1 of 1 shared result
- FreeDoM — FreeDoM: Training-Free Energy-Guided Conditional Diffusion Model (ICCV 2023) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- A Simple Approach to Unifying Diffusion-based Conditional Generation (ICLR 2025 · similarity 0.91)
- CoDi: Conditional Diffusion Distillation for Higher-Fidelity and Faster Image Generation (CVPR 2024 · similarity 0.90)
- NoiseCtrl: A Sampling-Algorithm-Agnostic Conditional Generation Method for Diffusion Models (CVPR 2025 · similarity 0.90)
- DEFT: Efficient Fine-tuning of Diffusion Models by Learning the Generalised $h$-transform (NeurIPS 2024 · similarity 0.89)
- High-Fidelity Diffusion-Based Image Editing (AAAI 2024 · similarity 0.89)
- REG: Rectified Gradient Guidance for Conditional Diffusion Models (ICML 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.
