AI Paper Atlas

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

Extending One-Step Image Generation from Class Labels to Text via Discriminative Text Representation

Chenxi Zhao, Chen Zhu, Xiaokun Feng, Aiming Hao, Jiashu Zhu, Jiachen Lei, Jiahong Wu, Xiangxiang Chu, Jufeng Yang

CVPR 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters · 3 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
train with MeanFlow objective (EMF) instead of standard flow matching, 1-stepBLIP3o-NEXT · GenEval overallGenEval score: 0.46 -> 0.74Improved
train with MeanFlow (EMF), 4-stepBLIP3o-NEXT · GenEval overallGenEval score: 0.86 -> 0.90Improved
train with MeanFlow (EMF), 1-stepBLIP3o-NEXT · DPG-Bench overallDPG-Bench score: 57.05 -> 77.36Improved
train with MeanFlow (EMF), 1-stepBLIP3o-NEXT · HPS-v2.1 averageHPS-v2.1 score: 18.54 -> 25.77Improved
use 4 EMF steps instead of 30 BLIP3o-NEXT stepsEMF · GenEval overallGenEval score: 0.91 -> 0.90No clear effect
increase EMF sampling steps from 1 to 4 (ablation)EMF · GenEval overallGenEval score: 0.74 -> 0.90Improved
increase EMF sampling steps from 4 to 8 (ablation)EMF · DPG-Bench overallDPG-Bench score: 81.20 -> 81.94Improved
MeanFlow training with fewer sampling steps (20 -> 4) (ablation)SANA-1.5 · GenEval overallGenEval score: 0.83 -> 0.50Hurt
fine-tune the text encoder on SFT data (ablation)SANA-1.5 flow matching · GenEval overallGenEval score: 0.81 -> 0.85Improved
fine-tune the text encoder on SFT data before MeanFlow, 4 steps (ablation)SANA-1.5 MeanFlow · GenEval overallGenEval score: 0.50 -> 0.47No clear effect
use BLIP3o-NEXT text encoder instead of T5 (ablation)text-retrieval analysis · COCO2017 retrievalDINO similarity: 0.634 -> 0.734Improved
use BLIP3o-NEXT text encoder instead of T5 (ablation)sub-sequence analysis · DPG-Bench promptscosine similarity: 0.893 -> 0.999Improved

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