AI Paper Atlas

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

DCoAR: Deep Concept Injection into Unified Autoregressive Models for Personalized Text-to-Image Generation

Fangtai Wu, Mushui Liu, Weijie He, Zhao Wang, Yunlong Yu

CVPR 2026 · Image generation › Text-to-image diffusion · approach: Personalized and region-controlled generation · 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 Dual Prior Preservation loss to LMCL (ablation)DCoAR (LMCL) · DreamBenchDINO: 0.6610 -> 0.7142Improved
add Dual Prior Preservation loss to LMCL (ablation)DCoAR (LMCL) · DreamBenchCLIP-I: 0.7647 -> 0.8019Improved
add Dual Prior Preservation loss to LMCL (ablation)DCoAR (LMCL) · DreamBenchCLIP-T: 0.3096 -> 0.2968Hurt
add Context-Aware Self-Regularization loss to LMCL (ablation)DCoAR (LMCL) · DreamBenchDINO: 0.6610 -> 0.7194Improved
add Context-Aware Self-Regularization loss to LMCL (ablation)DCoAR (LMCL) · DreamBenchCLIP-T: 0.3096 -> 0.3192Improved
add CASR on top of LMCL + DPP (ablation)DCoAR · DreamBenchCLIP-I: 0.8019 -> 0.8151Improved
add KL divergence term to class NTP loss in DPP (ablation)DCoAR DPP · DreamBenchCLIP-T: 0.3011 -> 0.3184Improved
add class NTP loss to the KL term in DPP (ablation)DCoAR DPP · DreamBenchCLIP-I: 0.7794 -> 0.8151Improved
insert context tokens into first 24 layers instead of 9 (ablation)DCoAR · DreamBenchsubject fidelity (CLIP-I, DINO)Hurt
add Identity Mask in subject-style composition (ablation)DCoAR · Subject-style generationqualitative concept contaminationImproved
use both text and image tokens instead of only one modality (ablation)DCoAR · DreamBench (qualitative)identity and editabilityImproved
DCoAR vs ZipLoRA on subject alignment in style taskDCoAR · Subject-style (StyleDrop styles)subject alignment: 0.604 vs. 0.655Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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