AI Paper Atlas

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

Self-Evaluation Unlocks Any-Step Text-to-Image Generation

Xin Yu, Xiaojuan Qi, Zhengqi Li, Kai Zhang, Richard Zhang, Zhe Lin, Eli Shechtman, Tianyu Wang, Yotam Nitzan

CVPR 2026 · Image generation › Text-to-image diffusion · approach: Personalized and region-controlled generation · 4 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 energy-preserving target normalization (ablation)Self-E (0.5B, 100k iterations) · GenEval, 4 stepsOverall: 0.6156 -> 0.6381Improved
add energy-preserving target normalization (ablation)Self-E (0.5B, 100k iterations) · GenEval, 2 stepsOverall: 0.5555 -> 0.5439Hurt
add energy-preserving target normalization (ablation)Self-E (0.5B, 100k iterations) · GenEval, 50 stepsOverall: 0.7018 -> 0.7160Improved
include auxiliary (fake score) term from the start of training (ablation)Self-E (0.5B, 100k iterations) · GenEval, 2 stepsOverall: 0.5439 -> 0.3307Hurt
include auxiliary term from the start of training (ablation)Self-E (0.5B, 100k iterations) · GenEval, 50 stepsOverall: 0.7160 -> 0.6166Hurt
re-introduce auxiliary term in later training stages (ablation)Self-E (2B) · 2-step qualitativeartifact severityImproved
use self-evaluation loss versus flow matching only0.5B model, 300k iterations · GenEval, 2 stepsOverall: 0.2523 -> 0.6097Improved
use self-evaluation loss versus flow matching only0.5B model, 300k iterations · GenEval, 50 stepsOverall: 0.7311 -> 0.7543Improved
use self-evaluation versus IMM native few-step training0.5B model, 300k iterations · GenEval, 4 stepsOverall: 0.5994 -> 0.7121Improved
increase inference stepsSelf-E (2B) · GenEvalOverall: 0.753 -> 0.781 -> 0.785 -> 0.815 (2/4/8/50 steps)Improved
compared to TiM (concurrent any-step model) at few stepsSelf-E (2B) · GenEval, 2 stepsOverall: 0.6338 -> 0.7531Improved
Self-E vs FLUX.1-Dev at 50 stepsSelf-E (2B) · GenEval, 50 stepsOverall: 0.7966 -> 0.8151Improved

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