Research map / Image generation / Text-to-image diffusion
LAMIC: Layout-Aware Multi-Image Composition via Scalability of Multimodal Diffusion Transformer
Yuzhuo Chen, Zehua Ma, Jianhua Wang, Kai Kang, Shunyu Yao, Weiming Zhang
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add Group Isolation Attention (ablation) | LAMIC · XVerseBench-extended (two-reference) | ID-S: 39.15 -> 78.04 | Improved |
| add Group Isolation Attention (ablation) | LAMIC · XVerseBench-extended (four-reference) | IN-R: 66.95 -> 89.81 | Improved |
| add Region-Modulated Attention (ablation) | LAMIC · XVerseBench-extended (two-reference) | IN-R: 81.77 -> 92.39 | Improved |
| add Region-Modulated Attention (ablation) | LAMIC · XVerseBench-extended (three-reference) | IN-R: 87.45 -> 91.90 | Improved |
| add Region-Modulated Attention (ablation) | LAMIC · XVerseBench-extended (two-reference) | AES: 54.91 -> 53.59 | Hurt |
| add Region-Modulated Attention (ablation) | LAMIC · XVerseBench-extended (four-reference) | ID-S: 65.81 -> 70.25 | Improved |
| raise first-stage step ratio from 0.05 to 0.20 (ablation) | LAMIC · XVerseBench-extended (two-reference) | IN-R: 92.39 -> 95.26 | Improved |
| raise first-stage step ratio from 0.05 to 0.20 (ablation) | LAMIC · XVerseBench-extended (two-reference) | AES: 53.59 -> 51.90 | Hurt |
| raise first-stage step ratio from 0.05 to 0.20 (ablation) | LAMIC · XVerseBench-extended (three-reference) | DPG: 91.95 -> 86.48 | Hurt |
| raise first-stage step ratio from 0.05 to 0.20 (ablation) | LAMIC · XVerseBench-extended (four-reference) | FI-R: 20.81 -> 20.71 | No clear effect |
| use LAMIC vs layout-unaware baselines for FI-R | multi-reference composition · XVerseBench-extended (four-reference) | FI-R: 23.71 -> 20.81 | No clear effect |
| LAMIC vs UNO on aesthetics | multi-reference composition · XVerseBench-extended (two-reference) | AES: 59.52 -> 53.59 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- XVerse — XVerse: Consistent Multi-Subject Control of Identity and Semantic Attributes via DiT Modulation (NeurIPS 2025) · better on 5 of 5 shared results
- OmniGen — OmniGen: Unified Image Generation (CVPR 2025) · better on 1 of 1 shared result
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
- InstanceAssemble: Layout-Aware Image Generation via Instance Assembling Attention (NeurIPS 2025 · similarity 0.88)
- HiCo: Hierarchical Controllable Diffusion Model for Layout-to-image Generation (NeurIPS 2024 · similarity 0.87)
- SpotActor: Training-Free Layout-Controlled Consistent Image Generation (AAAI 2025 · similarity 0.87)
- SSMG: Spatial-Semantic Map Guided Diffusion Model for Free-Form Layout-to-Image Generation (AAAI 2024 · similarity 0.87)
- Control and Realism: Best of Both Worlds in Layout-to-Image without Training (ICML 2025 · similarity 0.87)
- PLACE: Adaptive Layout-Semantic Fusion for Semantic Image Synthesis (CVPR 2024 · similarity 0.86)
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.
