ConsistCompose：面向图像合成的统一多模态布局控制
ConsistCompose: Unified Multimodal Layout Control for Image Composition
把布局坐标直接嵌入语言提示、并用坐标感知的无分类器引导实现空间控制，为视觉研究者提供了一种无需任务专用分支即可复用到底层生成模型上的布局可控多实例生成方案，同时配套的 3.4M 布局与身份标注数据集可用于训练与评测布局条件生成。
原文依据 embeds layout coordinates directly into language prompts, enabling layout-controlled multi-instance image generation
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论将视觉理解与图像生成耦合起来的统一多模态模型已取得快速进展，但多数系统仍聚焦于视觉定位——即将语言与图像区域对齐——而其生成侧对应能力，即面向布局可控多实例生成的语言嵌入布局定位生成（LELG），仍研究不足，并限制了精确的构图控制。我们提出 ConsistCompose，一个统一多模态框架，它将布局坐标直接嵌入语言提示中，从而在单一生成接口内实现基于交错图文输入的布局可控多实例图像生成。我们进一步构建了 ConsistCompose3M，这是一个包含布局与身份标注的 340 万多实例生成数据集（260 万文本引导数据对与 80 万图像引导数据对），为布局条件生成提供大规模监督。在该框架内，LELG 通过实例坐标绑定提示与坐标感知的无分类器引导得以实例化，从而将语言布局线索转化为精确的空间控制，且无需任务专用分支。在 COCO-Position 与 MS-Bench 上的实验表明，ConsistCompose 在布局控制基线上显著提升了空间精度，同时保持了身份保真度与具有竞争力的通用多模态理解能力，为布局可控的多模态图像生成建立了统一范式。
英文原文摘要 1428 字符
Unified multimodal models that couple visual understanding with image generation have advanced rapidly, yet most systems still focus on visual grounding--aligning language with image regions--while their generative counterpart, linguistic-embedded layout-grounded generation(LELG) for layout-controllable multi-instance generation, remains underexplored and limits precise compositional control. We present ConsistCompose, a unified multimodal framework that embeds layout coordinates directly into language prompts, enabling layout-controlled multi-instance image generation from Interleaved Image-Text within a single generative interface. We further construct ConsistCompose3M, a 3.4M multi-instance generation dataset with layout and identity annotations (2.6M text-guided and 0.8M image-guided data pairs) that provides large-scale supervision for layout-conditioned generation. Within this framework, LELG is instantiated through instance-coordinate binding prompts and coordinate-aware classifier-free guidance, which translate linguistic layout cues into precise spatial control without task-specific branches. Experiments on COCO-Position and MS-Bench show that ConsistCompose substantially improves spatial accuracy over layout-controlled baselines while preserving identity fidelity and competitive general multimodal understanding, establishing a unified paradigm for layout-controllable multimodal image generation.