research / 5 min read

Render Preview: See What the AI Sees

Interactive SVG-to-PNG preview showing the ControlNet conditioning input. Upload architectural drawings and see how they transform for AI rendering.

    ╔═══════════════════════════════════════════════════════════════════════╗
    ║   RENDER PREVIEW: SEE WHAT THE AI SEES                                ║
    ║                                                                       ║
    ║   ┌─────────────────┐        ┌─────────────────┐        ┌───────────┐ ║
    ║   │    ┌───┐ ┌───┐  │        │    ▓▓▓   ▓▓▓   │        │  photo-   │ ║
    ║   │    │   │ │   │  │   →    │    ▓▓▓   ▓▓▓   │   →    │  realistic│ ║
    ║   │    └───┘ └───┘  │  WASM  │    ▓▓▓▓▓▓▓▓▓   │   AI   │  render   │ ║
    ║   │    ┌─────────┐  │        │    ▓▓▓▓▓▓▓▓▓   │        │           │ ║
    ║   │    │         │  │        │    ▓▓▓▓▓▓▓▓▓   │        │           │ ║
    ║   │    └─────────┘  │        │                 │        │           │ ║
    ║   └─────────────────┘        └─────────────────┘        └───────────┘ ║
    ║      Original SVG              Canon Preview           ControlNet    ║
    ║      (your drawing)          (white on black)          (photorealism)║
    ║                                                                       ║
    ║   "See what the AI sees. Control what the AI does."                   ║
    ╚═══════════════════════════════════════════════════════════════════════╝

Interactive Demo

Upload an SVG floor plan to see how ControlNet will interpret it. The preview shows the Canon-styled conditioning image (white on black) that guides the AI render.

Upload SVG

How It Works

ControlNet uses edge-detected images to guide AI image generation. By converting your architectural SVG to a high-contrast PNG, we preserve the geometry while allowing the AI to render photorealistic materials and lighting.

Validation: Unlike black-box tools, this experiment proves the preview is accurate. Click "Test Render" to send your preview to ControlNet and see the actual result side-by-side. The comparison validates that our preview accurately represents what the AI receives.

SVG

Your Drawing

→

PNG

Canon Colors

→

ControlNet

→

JPG

Photorealistic

Philosophical Alignment

"See what the AI sees. Control what the AI does."

Canon Differentiation

Unlike black-box rendering tools, this preview makes the conditioning input visible. You see exactly what geometry the AI will preserve. The Test Render feature validates this transparency—proving the preview accurately represents ControlNet's input.

Fenestra Comparison

Fenestra offers 30-second renders with natural language editing—fast, but opaque. This tool offers explicit control—slower, but transparent.

Fenestra: "Add people to the terrace" → AI interprets
This: See conditioning → choose presets → understand result