Training-Free Object-Background Compositional T2I via Dynamic Spatial Guidance and Multi-Path Pruning
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866918439455555584 |
|---|---|
| author | Deng, Yang Mould, David Rosin, Paul L. Lai, Yu-Kun |
| author_facet | Deng, Yang Mould, David Rosin, Paul L. Lai, Yu-Kun |
| contents | Existing text-to-image diffusion models, while excelling at subject synthesis, exhibit a persistent foreground bias that treats the background as a passive and under-optimized byproduct. This imbalance compromises global scene coherence and constrains compositional control. To address the limitation, we propose a training-free framework that restructures diffusion sampling to explicitly account for foreground-background interactions. Our approach consists of two key components. First, Dynamic Spatial Guidance introduces a soft, time step dependent gating mechanism that modulates foreground and background attention during the diffusion process, enabling spatially balanced generation. Second, Multi-Path Pruning performs multi-path latent exploration and dynamically filters candidate trajectories using both internal attention statistics and external semantic alignment signals, retaining trajectories that better satisfy object-background constraints. We further develop a benchmark specifically designed to evaluate object-background compositionality. Extensive evaluations across multiple diffusion backbones demonstrate consistent improvements in background coherence and object-background compositional alignment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_09850 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Training-Free Object-Background Compositional T2I via Dynamic Spatial Guidance and Multi-Path Pruning Deng, Yang Mould, David Rosin, Paul L. Lai, Yu-Kun Computer Vision and Pattern Recognition Existing text-to-image diffusion models, while excelling at subject synthesis, exhibit a persistent foreground bias that treats the background as a passive and under-optimized byproduct. This imbalance compromises global scene coherence and constrains compositional control. To address the limitation, we propose a training-free framework that restructures diffusion sampling to explicitly account for foreground-background interactions. Our approach consists of two key components. First, Dynamic Spatial Guidance introduces a soft, time step dependent gating mechanism that modulates foreground and background attention during the diffusion process, enabling spatially balanced generation. Second, Multi-Path Pruning performs multi-path latent exploration and dynamically filters candidate trajectories using both internal attention statistics and external semantic alignment signals, retaining trajectories that better satisfy object-background constraints. We further develop a benchmark specifically designed to evaluate object-background compositionality. Extensive evaluations across multiple diffusion backbones demonstrate consistent improvements in background coherence and object-background compositional alignment. |
| title | Training-Free Object-Background Compositional T2I via Dynamic Spatial Guidance and Multi-Path Pruning |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2604.09850 |