AI Paper Atlas

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

DreamStory: Open-Domain Story Visualization by LLM-Guided Multi-Subject Consistent Diffusion

Huiguo He, Huan Yang, Zixi Tuo, Yuan Zhou, Qiuyue Wang, Yuhang Zhang, Zeyu Liu, Wenhao Huang, Hongyang Chao, Jian Ping Yin

TPAMI 2025 · Image generation › Text-to-image diffusion · approach: Text-driven synthesis and semantic guidance · 35 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 MMCA (masked mutual cross-attention) (ablation)Playground baseline · DS-500 2-SubjectD&C-DS: 0.3996 -> 0.5301Improved
add MMSA (masked mutual self-attention) (ablation)Playground baseline · DS-500 2-SubjectD&C-DS: 0.3996 -> 0.4233Improved
add MMSA to MMCA (ablation)Playground + MMCA · DS-500 2-SubjectD&C-DS: 0.5301 -> 0.5444Improved
add MMSA to MMCA (ablation)Playground + MMCA · DS-500 3-SubjectD&C-DS: 0.2186 -> 0.2335Improved
add MSD modules to Playground (ablation)Playground · DS-500 2-SubjectCLIP-T: 0.3818 -> 0.3779Hurt
add MSD modules to SDXLSDXL · DS-500 2-SubjectDS: 0.5045 -> 0.6048Improved
add MSD modules to SDXLSDXL · DS-500 3-SubjectD&C-DS: 0.1241 -> 0.1778Improved
use Playground instead of SDXL as backbone (ablation)DreamStory · DS-500 2-SubjectD&C-DS: 0.3848 -> 0.5444Improved
use DreamStory instead of MuDIMuDI · DS-500 2-SubjectD&C-DS: 0.4410 -> 0.5444Improved
use ChatGPT4 instead of LLaMA3-1B for subject annotation (ablation)LLM story director · DS-500 3-Subjectannotation accuracy (%): 80.13 -> 91.28Improved
use smaller MobileSAM instead of Grounded-SAM (ablation)DreamStory · DS-500 2-SubjectD&C-DS: 0.5444 -> 0.5386No clear effect
apply MMSA to decoder layers instead of encoder layers (ablation)DreamStory · qualitativeappearance consistencyImproved

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