AI Paper Atlas

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

CoDi: Subject-Consistent and Pose-Diverse Text-to-Image Generation

Zhanxin Gao, Beier Zhu, Liangyao, Jian Yang, Ying Tai

ICLR 2026 · Image generation › Text-to-image diffusion · approach: Layout-to-image control · 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
add Identity Transport (IT) to vanilla SDXL (ablation)SDXL · Subject-consistent T2I benchmarkDINO-v2 consistency: 0.8010 -> 0.8207Improved
add Identity Transport (IT) to vanilla SDXL (ablation)SDXL · Subject-consistent T2I benchmarkpose diversity: 0.0772 -> 0.0800Improved
add Identity Refinement (IR) to vanilla SDXL (ablation)SDXL · Subject-consistent T2I benchmarkDINO-v2 consistency: 0.8010 -> 0.8618Improved
add Identity Refinement (IR) alone to vanilla SDXL (ablation)SDXL · Subject-consistent T2I benchmarkpose diversity: 0.0772 -> 0.0675Hurt
add IR after IT (full CoDi) versus IT only (ablation)CoDi · Subject-consistent T2I benchmarkDINO-v2 consistency: 0.8207 -> 0.8514Improved
add IT before IR (full CoDi) versus IR only (ablation)CoDi · Subject-consistent T2I benchmarkpose diversity: 0.0675 -> 0.0758Improved
add IT before IR (full CoDi) versus IR only (ablation)CoDi · Subject-consistent T2I benchmarkDreamSim: 0.2044 -> 0.2136Hurt
increase alpha (top-percent reference features) from 30% to 70% (ablation)CoDi · Subject-consistent T2I benchmarkDINO-v2 consistency / pose diversityMixed
vary stage transition step t from 2 to 30 (ablation)CoDi · Subject-consistent T2I benchmarkDINO-v2 consistency / pose diversityNo clear effect
use CoDi instead of ConsiStoryConsiStory · Subject-consistent T2I benchmarkDreamSim: 0.2336 -> 0.2136Improved
use CoDi instead of ConsiStoryConsiStory · Subject-consistent T2I benchmarkpose diversity: 0.0621 -> 0.0758Improved
use CoDi instead of ConsiStoryConsiStory · Subject-consistent T2I benchmarkprompt fidelity: 0.9148 -> 0.9041Hurt

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