[论文] PLAYGROUND: Progressive Layout Generation with Render-Grounded Plannin...

论文概要

研究领域: CV 作者: Jaeho Lee, Junhwan Heo, Junghyun Park, WonJun Moon, Eunju Yang, Seungho Jang, Seongchan Kim, Seungryong Kim 发布时间: 2026-10-13 arXiv: 2610.18865

中文摘要

生成可编辑的图形设计需要将视觉和文本元素排列成连贯的构图，同时保留其内容和可编辑性。尽管近期基于 MLLM 的方法采用迭代生成和视觉细化来模拟真实设计工作流，但其高层计划在执行前就已预先确定，限制了设计演化过程中的适应性。为此我们提出 PlayGround——一个渐进式布局生成框架，能动态适应当前渲染画布。PlayGround 建立了一个渐进式设计循环，包含渐进式规划器、布局生成器和视觉验证器，由结构感知检索器引导。给定初始无序元素集，结构感知检索器首先推断粗略布局结构以检索结构相似的参考设计；随后每轮中，渐进式规划器检查当前渲染画布和剩余元素，形成下一个功能组并规划其构图；布局生成器预测可编辑的布局属性；视觉验证器评估渲染候选并在需要时提供细化反馈。为训练渐进式规划器进行逐步监督，我们构建了从完成的设计中提取功能组和轮次级规划轨迹的两阶段数据生成管线。在 Crello 和 LICA 上的实验证明了 PlayGround 相对于现有方法的有效性。

原文摘要

Generating editable graphic designs requires arranging visual and textual elements into a coherent composition while preserving their content and editability. Although recent MLLM-based approaches employ iterative generation and visual refinement to reflect the real-world design workflows, their high-level plans are predetermined prior to execution, limiting their adaptability as the design evolves. To address this limitation, we propose PlayGround, a progressive layout generation framework that dynamically adapts to the current rendered canvas. PlayGround establishes a progressive design loop comprising a Progressive Planner, a Layout Generator, and a Visual Verifier, guided by a Structure-Aware Retriever. Given an initial set of unordered elements, the Structure-Aware Retriever initially...

#论文 #arXiv #CV #小凯