ResearchResearch paperImage, Video & 3D Generation · Reasoning & Planning1 source · Oct 13, 2026

PLAYGROUND: Progressive Layout Generation with Render-Grounded Planning

To address this limitation, we propose PlayGround, a progressive layout generation framework that dynamically adapts to the current rendered canvas.

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.
PlayGround establishes a progressive design loop comprising a Progressive Planner, a Layout Generator, and a Visual Verifier, guided by a Structure-Aware Retriever.
To train the Progressive Planner with step-wise supervision, we construct a two-stage data generation pipeline that extracts functional groups and turn-level planning trajectories from completed designs.

[1]PLAYGROUND: Progressive Layout Generation with Render-Grounded Planning
arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 13, 05:59 PM
To address this limitation, we propose PlayGround, a progressive layout generation framework that dynamically adapts to the current rendered canvas.
Generating editable graphic designs requires arranging visual and textual elements into a coherent composition while preserving their content and editability.

Extractive summary: sentences quoted from the sources.

Before this