AI Paper Atlas

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

LACON: Training Text-to-Image Model from Uncurated Data

Zhiyang Liang, Ziyu Wan, Hongyu Liu, DONG CHEN, Qiu Shen, Hao Zhu, Dongdong Chen

ECCV 2026 · Image generation › Text-to-image diffusion · approach: Personalized and region-controlled generation · 0 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 quality-attribute conditioning and train on the full 110M unfiltered data (LACON-A) vs filtered Baseline-BSana-1.6B · GenEval 512pxGenEval: 68.0 -> 71.6Improved
add quality-attribute conditioning (LACON-A) vs Baseline-BSana-0.6B · GenEval 512pxGenEval: 62.8 -> 65.6Improved
add quality conditioning vs unconditioned full-data training (Baseline-C)Sana-1.6B · FID 512pxFID: 13.5 -> 11.2Improved
use negative-prompt guidance on full unfiltered data (Baseline-D vs Baseline-C)Sana-1.6B · GenEval 512pxGenEval: 67.0 -> 67.1No clear effect
train on full data without quality conditioning (Baseline-C) vs 65% filtered (Baseline-B)Sana-0.6B · FID 512pxFID: 13.4 -> 15.4Hurt
raise retained data ratio from 5% to 65% (ablation)Sana-1.6B · GenEval 512pxGenEval: 58.3 -> 68.0Improved
raise retained data ratio from 65% to 100% without conditioning (ablation)Sana-1.6B · GenEval 512pxGenEval: 68.0 -> 67.0Hurt
add aggressive per-attribute guidance (LACON-A over LACON-S) (ablation)Sana-1.6B · GenEval 512pxGenEval: 70.9 -> 71.6Improved
LACON-A over LACON-S at 1024 resolution (ablation)Sana-1.6B · DPG 1024pxDPG: 77.6 -> 78.8Improved
apply LACON to autoregressive Qwen3-1.7BQwen3-1.7B · GenEvalGenEval: 66.4 -> 70.3Improved
apply LACON to autoregressive Qwen3-0.6BQwen3-0.6B · FIDFID: 12.4 -> 11.2Improved
use GCC instead of linear interpolation tokens (ablation)LACON-A · GenEval, Sana-1.6BGenEval: 69.5 -> 71.6Improved

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