AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

D2PO: Optimizing Diffusion Samplers via Dynamic Preference

Jinkyu Kim, Jinyoung Choi, Bohyung Han

ECCV 2026 · Image generation › Text-to-image diffusion · approach: Preference optimization for diffusion · 0 citations (OpenAlex)

Read the paper

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 changedWhereMeasuredEffect
add dynamic preference (2N-step self-refined teacher instead of fixed N+1-step teacher) (ablation)D2PO · SD v1.5 COCO iPNDM, 4 stepsFID: 16.91 -> 15.69Improved
add dynamic preference (ablation)D2PO · SD v1.5 COCO iPNDM, 5 stepsFID: 13.70 -> 13.38Improved
use score-based energy instead of LPIPS (ablation)D2PO · SD v1.5 COCO iPNDM, 4 stepsFID: 17.88 -> 15.69Improved
use score-based energy instead of LPIPS (ablation)D2PO · SD v1.5 COCO iPNDM, 5 stepsFID: 14.59 -> 13.38Improved
use score-based energy instead of LPIPS (ablation)D2PO · SD v1.5 COCO iPNDM, 4 stepsAesthetic score: 5.1796 -> 5.2024Improved
use epoch-wise copy instead of EMA for reference timestep schedule (ablation)D2PO · SD v1.5 COCO iPNDM, 4 stepsFID: 15.75 -> 15.69Improved
use epoch-wise copy instead of EMA for reference timestep schedule (ablation)D2PO · SD v1.5 COCO iPNDM, 4 stepsAesthetic score: 5.1937 -> 5.2024Improved
D2PO instead of LD3 sampler optimizationiPNDM sampler · SD v1.5 COCO iPNDM, 4 stepsFID: 17.60 -> 15.69Improved
D2PO instead of LD3 at 7 stepsiPNDM sampler · SD v1.5 COCO iPNDM, 7 stepsFID: 12.41 -> 12.71Hurt
D2PO instead of LD3iPNDM sampler · SD v1.5 COCO iPNDM, 7 stepsHPSv2: 0.2434 -> 0.2513Improved
D2PO instead of LD3iPNDM 3M sampler · ImageNet-256 latent, 4 stepsFID: 9.19 -> 7.28Improved
D2PO instead of LD3InstaFlow sampler · InstaFlow COCO, 2 stepsFID: 63.55 -> 40.68Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy