Research map / Image generation / Fast diffusion sampling
DBMSolver: A Training-free Diffusion Bridge Sampler for High-Quality Image-to-Image Translation
Sankarshana Venugopal, Mohammad Mostafavi, Jonghyun Choi
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 |
|---|---|---|---|
| increase DBMSolver NFE from 6 to 20 | DBMSolver · DIODE-Outdoor 256 | FID: 3.38 -> 2.06 | Improved |
| increase DBMSolver NFE from 6 to 20 | DBMSolver · Edges2Handbags 64 | FID: 0.93 -> 0.53 | Improved |
| increase DBMSolver NFE from 6 to 20 | DBMSolver · Face2Comics 256 | FID: 3.04 -> 0.96 | Improved |
| increase DBMSolver NFE from 6 to 20 | DBMSolver · ImageNet inpainting 256 | FID: 4.98 -> 4.07 | Improved |
| use DBMSolver (6 NFE) instead of Hybrid Heun (119 NFE) | DBM sampling · DIODE-Outdoor 256 | FID: 4.43 -> 3.38 | Improved |
| use DBMSolver (20 NFE) instead of DBIM-3 (20 NFE) | DBM sampling · Edges2Handbags 64 | FID: 1.45 -> 0.53 | Improved |
| use DBMSolver (20 NFE) instead of DBIM-3 (20 NFE) | DBM sampling · ImageNet inpainting 256 | FID: 4.07 -> 4.07 | No clear effect |
| use DBMSolver (6 NFE) instead of DBIM-1 (6 NFE) | DBM sampling · ImageNet inpainting 256 | FID: 5.36 -> 4.98 | Improved |
| use DBMSolver (6 NFE) instead of DBIM-1 (6 NFE) | DBM sampling · CelebAMask-HQ 256 | FID: 44.92 -> 34.76 | Improved |
| use DBMSolver (6 NFE) instead of ODES3 (28 NFE) | DBM sampling · Edges2Handbags 64 | FID: 0.54 -> 0.93 | Hurt |
| use training-free DBMSolver (6 NFE) instead of distilled CDBM (2 NFE) | DBM sampling · DIODE-Outdoor 256 | FID: 3.66 -> 3.38 | Improved |
| use 2nd-order (k=2) instead of 1st-order solution (ablation) | DBMSolver · supplementary ablation | FID | Improved |
Closest papers by meaning
- Diffusion Bridge Implicit Models (ICLR 2025 · similarity 0.90)
- Inverse Bridge Matching Distillation (ICML 2025 · similarity 0.90)
- Denoising Diffusion Bridge Models (ICLR 2024 · similarity 0.89)
- Exploring the Design Space of Diffusion Bridge Models (NeurIPS 2025 · similarity 0.89)
- SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis (ICLR 2024 · similarity 0.88)
- A Simple Early Exiting Framework for Accelerated Sampling in Diffusion Models (ICML 2024 · 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.
