Architectural Program Creator

Architectural Program Creator is a browser-based spatial planning tool for turning a list of spaces and their relationships into an interactive 3D program diagram.

It is designed for early-stage architectural thinking, when the goal is to test adjacencies, relative sizes, clustering, and overall spatial composition before moving into detailed modeling.

The app reads:

• a program CSV with space dimensions and metadata
• a relationship matrix CSV with positive and negative adjacency weights

It then places those spaces as 3D blocks inside a configurable boundary, visualizes them in real time, and exports the arrangement as JSON.

Screenshots

Full interface view:

3D viewport detail:

Files

• program_creator.html: main application
• space_parameters.csv: minimal example program input
• relationship_matrix.csv: minimal example relationship matrix

What It Does

• Loads a program table from CSV
• Loads a relationship matrix from CSV
• Packs and repositions spaces inside a configurable box
• Supports isometric view, labels, and PNG export
• Exports the current 3D arrangement as JSON

Run Locally

This project does not require a build step.

1. Open program_creator.html in a modern browser.
2. If your browser blocks local file features, serve the folder with a simple static server instead.
3. Load space_parameters.csv and relationship_matrix.csv through the file inputs in the sidebar.

Notes:

• The viewer loads three.js from https://unpkg.com, so an internet connection is required unless you vendor that dependency locally.
• The HTML references an optional HelveticaNowDisplay-Regular.ttf file in the same folder. If it is missing, the app falls back to system fonts.
• The bundled CSVs are intentionally small examples, not a full project dataset.

Program CSV

The loader accepts flexible column names. Core fields are:

• code
• space_type
• family
• count
• length_m or length
• depth_m or depth
• ceiling_height_m, height_m, or height

Optional fields:

• inseparable_group or similar group identifiers
• preferred position columns such as preferredx, preferredy, preferredz

For rows with count > 1, the app generates numbered instances automatically.

Relationship Matrix CSV

Expected format:

• first column: row type id
• remaining columns: target type ids
• cell values: numeric weights

Interpretation:

• positive values pull space types closer together
• negative values push them apart
• 0 means no relationship

The app normalizes relationship values internally.

Export

The Export JSON button outputs:

• box dimensions
• view settings
• type definitions
• per-space geometry, color, center point, and rotation

Notes

• This repository currently contains a single-file HTML application.
• If you want fully offline use, download three.min.js locally and update the script tag in program_creator.html.

License

This project is source-available for personal, educational, research, and non-commercial evaluation use only.

Commercial use requires separate permission from the repository owner. See LICENSE.txt.