AI Paper Atlas

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

EruDiff: Refactoring Knowledge in Diffusion Models for Advanced Text-to-Image Synthesis

Xiefan Guo, Xinzhu Ma, Haoxiang Ma, Zihao Zhou, Di Huang

ECCV 2026 · Image generation › Text-to-image diffusion · approach: Compositional text-to-image guidance · 1 citations (OpenAlex)

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 changedWhereMeasuredEffect
add EruDiff (DK-DM + NO-RL) to FLUX.1-devFLUX.1-dev · WISEWiScore overall: 0.50 -> 0.64Improved
train EruDiff directly on WISE prompts instead of Knowledge-10KFLUX.1-dev + EruDiff · WISEWiScore overall: 0.64 -> 0.73Improved
add EruDiff to FLUX.1-devFLUX.1-dev · Science-T2I SSciScore: 23.31 -> 33.70Improved
add EruDiff to Qwen-ImageQwen-Image · Science-T2I SSciScore: 25.38 -> 35.43Improved
add EruDiff to Qwen-ImageQwen-Image · Science-T2I CSciScore: 31.56 -> 38.48Improved
replace GPT-4o prompt rewriting with EruDiffFLUX.1-dev · Science-T2I SSciScore: 32.90 -> 33.70Improved
add NO-RL to EruDiff (ablation)EruDiff on FLUX.1-dev · Science-T2I SSciScore: 31.99 -> 33.70Improved
add NO-RL to EruDiff (ablation)EruDiff on FLUX.1-dev · Science-T2I CSciScore: 32.82 -> 34.54Improved
add positive-sample learning to NO-RL (ablation)EruDiff on FLUX.1-dev · Science-T2I SSciScore: 33.70 -> 33.83No clear effect
add positive-sample learning to NO-RL (ablation)EruDiff on FLUX.1-dev · Science-T2I CSciScore: 34.54 -> 34.19No clear effect
use SciScore reward instead of CLIPScore for NO-RL filtering (ablation)EruDiff on FLUX.1-dev · Science-T2I SSciScore: 32.93 -> 33.70Improved
add EruDiff instead of GRPOFLUX.1-dev · Science-T2I SSciScore: 31.68 -> 33.70Improved

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