Research map / Image generation / Fast diffusion sampling
Few-Shot Image Generation by Conditional Relaxing Diffusion Inversion
Yu Cao, Shaogang Gong
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 CRDI diffusion reconstruction with SGE instead of Image2StyleGAN inversion (eta=15) | FFHQ-pretrained generator · Reconstruction, Bedroom | SSIM: 0.52 -> 0.84 | Improved |
| increase degree of rigidity eta from 1 to 15 (ablation) | CRDI · Reconstruction, Babies | SSIM: 0.55 -> 0.74 | Improved |
| use low eta=1 on a domain similar to source (ablation) | CRDI generation · Babies generation | visual quality | Hurt |
| increase eta beyond optimum (ablation) | CRDI · Babies FID and Intra-LPIPS | FID | Hurt |
| add noise perturbation scheduler to SGE and keep SGE reconstruction (removing either component) (ablation) | CRDI · Babies, Sunglasses, MetFaces | FID | Improved |
| replace GenDA with CRDI under one-shot adaptation | FFHQ source model · Sunglasses 1-shot | FID: 83.70 -> 74.60 | Improved |
| replace GenDA with CRDI at 10-shot | FFHQ source model · Sunglasses 10-shot | FID: 35.64 -> 24.62 | Improved |
| CRDI versus RICK fine-tuning on Babies | CRDI · Babies FID | FID: 39.39 -> 48.52 | Hurt |
| CRDI versus RICK on MetFaces FID | CRDI · MetFaces | FID: 48.53 -> 94.86 | Hurt |
| CRDI versus RICK on MetFaces mode coverage | CRDI · MetFaces | MC-SSIM: 0.69 -> 0.78 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Image2StyleGAN — Image2StyleGAN++: How to Edit the Embedded Images? (CVPR 2020) · better on 3 of 3 shared results
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
- Uni-DAD: Unified Distillation and Adaptation of Diffusion Models for Few-step Few-shot Image Generation (CVPR 2026 · similarity 0.91)
- Revisiting Diffusion Models: From Generative Pre-training to One-Step Generation (ICML 2025 · similarity 0.88)
- A Simple Approach to Unifying Diffusion-based Conditional Generation (ICLR 2025 · similarity 0.88)
- You Only Sample Once: Taming One-Step Text-to-Image Synthesis by Self-Cooperative Diffusion GANs (ICLR 2025 · similarity 0.88)
- Bridging Data Gaps in Diffusion Models with Adversarial Noise-Based Transfer Learning (ICML 2024 · similarity 0.87)
- Understanding the Generalization of Pretrained Diffusion Models on Out-of-Distribution Data (AAAI 2024 · similarity 0.87)
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.
