abdullah4ai/visionos-patterns

visionOS Platform Patterns

Three Scene Types

1. Windows — 2D SwiftUI content in a glass panel (default)
2. Volumes — 3D content in a bounded box
3. Immersive Spaces — full spatial experience around the user

RealityKit Entity Loading

import RealityKit

// In a RealityView
RealityView { content in
    if let entity = try? await Entity(named: "Scene", in: realityKitContentBundle) {
        // Add collision and input for interaction
        entity.components.set(CollisionComponent(shapes: [.generateBox(size: [0.1, 0.1, 0.1])]))
        entity.components.set(InputTargetComponent())
        content.add(entity)
    }
}

Entity Interaction

Entities need CollisionComponent + InputTargetComponent for user interaction:

RealityView { content in
    let sphere = ModelEntity(mesh: .generateSphere(radius: 0.1))
    sphere.components.set(CollisionComponent(shapes: [.generateSphere(radius: 0.1)]))
    sphere.components.set(InputTargetComponent())
    content.add(sphere)
} update: { content in
    // Update entities
}
.gesture(
    TapGesture()
        .targetedToAnyEntity()
        .onEnded { value in
            // Handle tap on entity
        }
)

Spatial Gestures

// Tap gesture on entities
.gesture(
    TapGesture()
        .targetedToAnyEntity()
        .onEnded { value in
            handleTap(on: value.entity)
        }
)

// Drag gesture on entities
.gesture(
    DragGesture()
        .targetedToAnyEntity()
        .onChanged { value in
            value.entity.position = value.convert(value.location3D, from: .local, to: .scene)
        }
)

ARKit in Immersive Spaces

ARKit is only available in Full Space (not Shared Space):

ImmersiveSpace(id: "full-immersive") {
    RealityView { content in
        // ARKit features available here
    }
}
.immersionStyle(selection: .constant(.full), in: .full)

Observable Pattern

Use @Observable over ObservableObject:

@Observable
class AppModel {
    var items: [Item] = []
    var selectedItem: Item?
    var isImmersive = false
}

// In App
@State private var model = AppModel()

WindowGroup {
    ContentView()
        .environment(model)
}

Concurrency

MainActor by default in Swift 6.2. Use detached Tasks for background work:

// Already on MainActor by default
func loadData() {
    Task.detached {
        let data = await fetchData()
        await MainActor.run {
            self.items = data
        }
    }
}

Hover and Focus

Eye tracking drives hover states automatically:

Button("Action") { }
    .hoverEffect() // Required for all interactive elements
    .hoverEffect(.highlight) // Highlight variant
    .hoverEffect(.lift) // Lift variant

Widgets

visionOS 26 widgets are spatial — they snap to walls and tables:

// Standard WidgetKit implementation works
// visionOS renders them as spatial glass panels

Deprecated API Alternatives

• Use .foregroundStyle() not .foregroundColor()
• Use .clipShape(.rect(cornerRadius:)) not .cornerRadius()
• Use @Observable not ObservableObject

NOT Available on visionOS

• No camera access (enterprise API only)
• No haptic feedback (CoreHaptics)
• No HealthKit
• No direct gaze data (privacy protected)
• No UIScreen.main
• No MapKit with full map view
• No NFC or Bluetooth LE scanning
• No App Clips
• No share extensions
• No Safari extensions

Rules

1. Three scene types: Windows (default), Volumes (3D bounded), Immersive Spaces (full spatial)
2. Entities need CollisionComponent + InputTargetComponent for interaction
3. ARKit only works in Full Space immersive experiences
4. Use @Observable over ObservableObject
5. ALL interactive elements MUST have .hoverEffect()
6. Use spatial gestures (.targetedToAnyEntity()) for 3D interaction
7. MainActor by default — use Task.detached for background work
8. Use .foregroundStyle() not .foregroundColor()
9. Use .clipShape(.rect(cornerRadius:)) not .cornerRadius()