Scany

Privacy-first document scanner for iPhone. Point the camera at a page, hold still, and Scany detects the paper in any orientation, waits until the shot is steady, auto-captures, straightens the perspective, and saves a PDF to your Google Drive or iCloud. There is no Scany account and no Scany server. Pages are processed on the phone.

Version 1.0 is approved and is being published on the App Store. Product site: scany.eyesrhythm.com. Privacy: scany.eyesrhythm.com/privacy.

The source in this tree is under the PolyForm Noncommercial License 1.0.0. You may use, modify, and share it for a noncommercial purpose, provided you keep LICENSE and the required notice below. Commercial use is not included. Third-party libraries keep their own licenses. See THIRD_PARTY_NOTICES.md.

Required Notice: Copyright Pranay Soni (https://github.com/pranaycv)

This is a custom AVFoundation + Vision + Core Image pipeline — not a wrapper around VNDocumentCameraViewController — so stabilize-then-auto-capture and continuous multi-page scanning work the way the product requires.

Requirements

• A Mac with Xcode 15+ (Xcode 16 recommended)
• Supported devices: iPhone 12, 12 mini, 12 Pro, 12 Pro Max, and every newer iPhone
• Minimum iOS: 15.0 (the floor for VNDetectDocumentSegmentationRequest plus this camera pipeline). iPhone 12 shipped with iOS 14 and can update; install iOS 15 or later.
• A physical iPhone is required for camera scanning. The Simulator can launch the app but cannot exercise live page detection.
• An Apple ID / development team for device signing

Open and run on a real iPhone

1. Clone this repo and open Scany.xcodeproj in Xcode.
   (Optional) If you use XcodeGen, project.yml can regenerate the project with xcodegen generate.
2. Select the Scany scheme and your connected iPhone.
3. In the project editor → Signing & Capabilities, choose your Team. The bundle ID is com.prnysoni.Scany.
4. Plug in the iPhone, unlock it, and trust the computer if asked.
5. Press Run (⌘R). On first launch, iOS will ask for Camera access — allow it.
6. If iOS blocks the developer app: Settings → Privacy & Security → Developer Mode (if prompted), then Settings → General → VPN & Device Management → trust your certificate.

Test scanning on a real iPhone 12 or newer. The Simulator will show the home screen but cannot exercise live page detection.

Permissions

Permission	Why
Camera (NSCameraUsageDescription)	Live preview, Vision page detection, auto-capture
Motion (NSMotionUsageDescription)	Keep captured pages upright from how you hold the phone
Photo Library	Not required. Cloud upload is the primary export. Share sheet is a fallback.

How to use

1. Tap Start scan.
2. Place a page in the frame. A live overlay appears on the detected quad.
3. Hold still. A yellow overlay appears once the page is stable, then Scany captures automatically (shutter sound + haptic) and looks for the next page.
4. The shutter button is optional — use it only if auto-capture does not fire.
5. Thumbnails along the bottom show the session. Long-press a thumbnail to delete.
6. Tap Done when finished. On the review screen you can delete / reorder pages, turn on Auto-enhance all pages, then Save to Drive / Save to iCloud. After a successful upload, clear the local page images. System share sheet remains a secondary fallback. There is no on-device document library.

Guidance shown in-camera:

• Position the page in the frame — no document found
• Move closer — page is too small
• Hold steady… — detected but still jittering
• Low light — add more light — frame is too dark
• Hold still until it looks sharp — Laplacian variance is low (motion / misfocus)

After each capture there is a ~1.35s cooldown so the same page is not captured twice.

Cancel mid-session asks to discard if any pages exist.

Architecture

HomeView → ScanSessionView → ReviewView → Drive / iCloud (share sheet fallback)
                │
                ├─ CameraSession          AVCaptureSession (video + photo)
                ├─ DocumentDetector       VNDetectDocumentSegmentationRequest
                │                         + VNDetectRectanglesRequest fallback
                ├─ QuadStabilizer         jitter / fill / blur / low-light
                ├─ PerspectiveCorrector   CIPerspectiveCorrection
                ├─ OrientationCorrector   Vision text score at 0/90/180/270°
                ├─ ImageEnhancer          CIDocumentEnhancer + contrast + unsharp
                ├─ PageCaptureProcessor   Warp + upright + write JPEG
                ├─ PDFExporter            UIGraphicsPDFRenderer
                └─ CloudManager           Google Drive API v3 + iCloud Drive

Piece	Role
Camera/CameraSession.swift	Back camera, 4K when available, video frames for Vision, stills for capture
Detection/DocumentDetector.swift	Document quad in Vision normalized coordinates (any paper rotation)
Detection/QuadStabilizer.swift	Yellow overlay after 5 stable frames, auto-capture after 20 more
Processing/PerspectiveCorrector.swift	Warp the quad to a rectangle
Processing/OrientationCorrector.swift	Pick the upright rotation from text; else prefer portrait
Processing/ImageEnhancer.swift	Document-style enhance applied to all pages when enabled once
Processing/PageCaptureProcessor.swift	Flatten, upright, JPEG write on a background queue
Export/PDFExporter.swift	Multi-page PDF builder
Cloud/CloudManager.swift	Connect accounts, remember folder, upload, disconnect
Cloud/GoogleDriveService.swift	Google Sign-In (SPM) or PKCE + Drive API v3
Cloud/ICloudDriveService.swift	iCloud Drive Documents container (iCloud Drive → Scany)
Cloud/SimulatedCloudStore.swift	Sandbox upload when client IDs are still placeholders
UI/Scan/ScanViewModel.swift	Ephemeral session only — pages are cleared after upload

Captured pages are JPEGs in a temporary session folder only. After a successful cloud upload you can clear them. Scany does not keep a history of past scans.

Project layout

Scany.xcodeproj/          Xcode project + shared scheme
project.yml               XcodeGen spec (optional)
Scany/
  ScanyApp.swift          App entry + Home / Scan / Review routing
  Info.plist              Display name, camera usage, iOS 15 minimum, arm64
  PrivacyInfo.xcprivacy   Privacy manifest
  Assets.xcassets/        App icon + accent
  Cloud/                  Drive + iCloud + Keychain
  Camera/ Detection/ Processing/ Export/ Models/ UI/

Display name: Scany. iPhone only, iOS 15.0+, arm64, SwiftUI, Swift 5.

The project does not require iOS 17. SwiftUI APIs that shipped later (NavigationStack, navigationDestination, two-parameter onChange, scrollContentBackground, UIColor(Color)) are #available-gated with iOS 15/16 fallbacks in UI/Compatibility.swift. Document detection uses Vision segmentation on iOS 15+, with VNDetectRectanglesRequest as the real fallback when no page quad is returned.

Cloud export (Google Drive + iCloud)

Scans live in a Google Drive or iCloud Drive folder you choose. Settings (gear on Home, or first export) lets you:

1. Connect Google Drive and/or Connect iCloud
2. Pick or create a destination folder (remembered for later exports)
3. See account + folder status, Change folder, or Disconnect

Review’s primary button is Save to Drive / Save to iCloud (or a chooser if both are connected). Share sheet is secondary.

The iOS OAuth client ID in CloudConfig.swift is the public client for the shipped app, bundle ID com.prnysoni.Scany. An iOS client has no secret; Google restricts it to that bundle ID. A fork that changes the bundle ID needs its own client. Do not commit a client secret, refresh token, or service-account key.

If Google.clientID still starts with YOUR_, Connect Drive writes files to the app sandbox under Documents/CloudUploadSimulation/. iCloud uses the real iCloud account on a device; Simulator falls back to the sandbox if iCloud is signed out.

Privacy policy (App Store + Google consent): https://scany.eyesrhythm.com/privacy

App Store listing copy, privacy answers, and archive steps: AppStore/LISTING.md

1. Fill in Scany/Cloud/CloudConfig.swift

Google.clientID = iOS OAuth client ID ending in .apps.googleusercontent.com

isConfigured is false while the value still starts with YOUR_. iCloud needs no client ID.

2. Google Cloud Console

1. Create (or pick) a project → APIs & Services → enable Google Drive API.
2. Credentials → Create credentials → OAuth client ID → application type iOS.
3. Bundle ID: com.prnysoni.Scany.
4. Copy the client ID into CloudConfig.Google.clientID.
5. Add the reversed client ID as a URL scheme in Info.plist and set GIDClientID to the same client ID.
6. OAuth consent screen: add scope https://www.googleapis.com/auth/drive.file (files this app creates). userinfo.email is used for the Settings label. The app may stay in Testing; users will see Google’s unverified-app warning.

Redirect used by the PKCE fallback: {reversed-client-id}:/oauth2redirect/google.

3. iCloud Drive

Connect iCloud opens the Files picker. Choose an iCloud Drive folder (create one named Scany if you want). No iCloud entitlement is required, so a free personal Apple team can still run on device. PDFs are written into that folder.

4. Info.plist URL schemes

Scheme	Purpose
com.googleusercontent.apps.<YOUR_IOS_CLIENT_PREFIX>	Google Sign-In / PKCE return
scany	Extra custom scheme

5. Swift packages (Xcode resolves these on open)

Package	URL	Product	License
Google Sign-In	https://github.com/google/GoogleSignIn-iOS	GoogleSignIn	Apache 2.0

Google auth uses the official SDK when the package is linked (#if canImport). If it is missing, Drive falls back to PKCE + ASWebAuthenticationSession. Drive API v3 uploads go through URLSession. Google tokens and the remembered folder are stored in the Keychain. iCloud uses a security-scoped Files bookmark and NSFileCoordinator.

Notes

• Signing: leave Automatically manage signing on and pick your team.
• Camera scanning still needs a physical iPhone. Cloud Settings + simulated upload work in Simulator.
• Torch is available on the scan screen for dim rooms.
• Tap the preview to focus/expose on that point.
• Auto-enhance is a document-level switch. New pages captured after you turn it on are enhanced automatically.

License

Scany’s own code is under the PolyForm Noncommercial License 1.0.0, the same terms as EyesRhythm. Noncommercial use, modification, and sharing are allowed if you keep LICENSE and this notice. Commercial use is not included.

Required Notice: Copyright Pranay Soni (https://github.com/pranaycv)

Questions: prnysoni@gmail.com. Security reports go to the same address. See SECURITY.md.