Watermarkd pipeline

Watermarkd’s pipeline is built to be predictable: users import an image, configure overlays, and produce an export at a known size and quality. This doc captures the current steps and constraints.

Pipeline stages

1. Import – load UIImage, normalize orientation, downscale if needed.
2. Overlay config – text, size, color, position, opacity.
3. Preview – render in SwiftUI Canvas (or CALayer fallback).
4. Export – draw into Core Graphics context at target resolution.

Import behavior

• Honor EXIF rotation; user should never see a rotated preview.
• Large images (>12–15 MP) may be auto-downscaled for performance, not visibly in UI.
• Color space is converted to sRGB for consistent output.

Overlay model

Each overlay is a struct:

struct Overlay {
    var text: String
    var position: CGPoint   // normalized 0–1
    var scale: CGFloat      // relative to image width
    var color: UIColor
    var opacity: CGFloat
}

Positions remain device-agnostic by staying normalized.

Rendering

• Preview uses SwiftUI drawing, not final export quality.
• Export uses Core Graphics for crisp text.
• All overlays are composited last to avoid any filtering from image resizing.

Export flow

// pseudocode outline
beginContext(size: targetSize, scale: 1.0)

draw(baseImage)

for overlay in overlays:
    drawText(overlay.text,
             at: overlay.position * targetSize,
             scale: overlay.scale,
             color: overlay.color.withAlpha(opacity))

return contextImage

Constraints

• Export must match preview positioning within 1–2 px.
• Keep UI controls minimal — avoid multi-step menus.
• Ensure fast export even on older devices.

Future considerations

• Multi-layer overlays (icons, QR, shapes).
• Batch export.
• Non-destructive presets.