Automatic Design Draft Recognition and Floor Generation in the Tongtian Tower Platform

The article focuses on eliminating conventional development constraints by creating an online pipeline that transforms design drafts into floor layouts with a single click, requiring no code and offering low entry barriers for users to become floor style producers.

It introduces the background of the Tongtian Tower platform, which aims to improve efficiency in marketing activity construction by automating the link between design and floor production, and provides an overview of the system’s goals.

What is Design Draft Automatic Recognition? The system uses MCube to determine whether a template needs to be fetched, loads the template, converts the product into a view‑tree, parses expressions and events, binds them, and finally renders the view to the screen.

The project, named “Design‑Front‑End Integration,” parses Sketch design drafts (JSON metadata) to generate floor layouts automatically, reducing manual effort and enabling designers and operators to create floors without referencing annotations.

During the 618 promotion, the solution was applied to the Qinglu main venue, supporting various material types (single‑row, double‑row, carousel, etc.) and achieving a 40% increase in construction efficiency.

Core Design Philosophy leverages existing resources and user habits by combining RELAAAY asset management, the Tongtian Tower flexible floor building system, and design standards, resulting in a streamlined workflow that reduces steps and boosts efficiency.

Core Architecture includes automatic design‑draft recognition, canvas element dragging and preview, visual fine‑tuning, advanced configurations (navigation, tracking), and one‑click floor saving with asset mapping.

The solution addresses three main challenges: breaking platform linkages, automating design‑draft recognition, and associating generated elements with data fields. It uses a combination of style extraction from the design draft and existing material pools to bind data fields without requiring users to understand underlying data structures.

Implementation steps involve retrieving design‑draft JSON via RELAAAY’s API, dehydrating and rehydrating data to remove redundant layers, recognizing layout and elements based on grouping rules, and generating DSL‑compatible floor protocols.

Future work will focus on improving recognition accuracy, enhancing design‑draft fidelity, optimizing the canvas interaction, expanding the material pool, and further automating the floor‑building process.