Skip to main content

Quick Start

This guide walks you through the full Planova workflow: uploading a floor plan, reviewing the AI parse, exploring the 3D scene, and exporting the result.

1. Launch the App

Start Planova in development mode:

pnpm tauri dev

The app opens to the Project Dashboard.

2. Create a New Project

Click the New Project button on the dashboard. You will be prompted to fill in:

  • Name -- a descriptive name for your project (e.g. "Apartment 3BR")
  • Description -- optional notes
  • Style -- choose one of the six interior styles:
    • Modern Luxury
    • Cream
    • Nordic
    • New Chinese
    • Wabi-Sabi
    • Industrial

Click Create to proceed. You will be taken to the project detail page.

:::tip Style Can Be Changed Later You can change the style preset at any time from the 3D viewer toolbar. Each style applies different materials, colors, and furniture sets. :::

3. Upload a Floor Plan Image

On the project detail page, click the upload area or drag and drop a floor plan file.

Supported formats:

  • JPG / JPEG
  • PNG
  • PDF

Maximum file size is 50 MB.

:::info Best Results For the most accurate parsing, use a clean, high-resolution floor plan with clearly visible room labels, wall lines, and door/window symbols. Hand-drawn sketches can work but may produce lower quality results. :::

4. Wait for AI Parsing

Once uploaded, Planova automatically begins the parsing pipeline. You will see progress indicators for each stage:

  1. Preprocessing -- image is cleaned and normalized
  2. VLM Parse -- the AI vision model analyzes the floor plan
  3. Normalize -- raw detections are converted to a scene graph
  4. Repair -- geometry issues are fixed
  5. Validate -- the scene is scored for quality
  6. 3D Generation -- furniture is placed and materials applied

The entire process typically takes 30--90 seconds depending on your LLM provider and the complexity of the floor plan.

5. Review Parse Quality

After parsing completes, you will see the ReviewGate screen. This shows:

  • Quality scores -- metrics like wall completeness, room closure, and furniture coverage, each displayed as a percentage
  • Debug overlay -- a rendered image showing detected walls, rooms, doors, and windows overlaid on your original floor plan
  • Scene preview -- a quick look at the generated 3D scene

You have two options:

ActionWhen to Use
AcceptQuality scores look good, proceed to the 3D viewer
RetryScores are low or the parse missed rooms; re-run the pipeline

:::warning Low Quality Scores If wall completeness or room closure is below 70%, consider retrying or uploading a cleaner floor plan image. Low scores mean the 3D geometry will have gaps or incorrect room shapes. :::

6. Explore the 3D Scene

Once you accept the parse, you enter the 3D Viewer. There are three interaction modes:

Orbit Mode (default)

  • Left-click drag to rotate the camera around the scene
  • Scroll wheel to zoom in/out
  • Right-click drag to pan
  • Reset Camera button to return to the default view

Walk Mode

  • Click Walk in the toolbar to enter first-person mode
  • WASD or Arrow keys to move
  • Mouse to look around
  • Press Esc to exit walk mode

Edit Mode

  • Click Edit in the toolbar
  • Click on any object (furniture, wall, etc.) to select it
  • Use the controls to move, rotate, or delete selected objects
  • Toggle ceilings on/off for a better view during editing

7. Change Style Preset

In the viewer toolbar, click the style selector to switch between presets. Each preset changes:

  • Wall and floor materials/textures
  • Ceiling style
  • Furniture models and colors
  • Lighting mood

The scene regenerates with the new style while preserving the room layout.

8. Export or Render

From the 3D viewer toolbar you have two export options:

Export as GLB

Click Export GLB to save the 3D scene as a standard .glb file. This file can be imported into:

  • Blender
  • Unity
  • Unreal Engine
  • Any 3D tool that supports the GLB/GLTF format

AI Render

Click AI Render to generate a photorealistic image of the current view:

  1. Optionally edit the rendering prompt to describe the mood or lighting you want
  2. Click Generate
  3. Wait for the image generation model to produce the render
  4. The rendered image is saved to your project

:::info AI Rendering Requires Configuration AI rendering uses a separate image generation provider. Make sure you have configured the Image provider in Settings with a compatible model (e.g. DALL-E, Stable Diffusion). :::