Add screen recording, crop, and a Cap-style result window

Recording: Ctrl+Shift+Cmd+R (or the menu) opens the same selection
flow — a marquee records that region, clicking a window records it
(following moves). Video is H.264 .mov at 60fps via SCStream ->
AVAssetWriter with the cursor included, auto-saved to ~/Movies and
revealed in Finder on stop. A red border and a floating timer/stop bar
show during recording, both excluded from the video. Audio is wired
behind a flag for later.

The result window gains a Crop button (crop to the selection;
annotations translate and stay editable) and is restyled after Cap:
sliding-pill Screenshot/Text switcher, capsule buttons in Cap blue,
circular icon buttons, and the content on a rounded card — light and
dark.

Generated with OpenCode

Assisted-by: Kimi:k3

Sources/Kapture/App.swift

@@ -33,16 +33,42 @@ final class AppDelegate: NSObject, NSApplicationDelegate {
             }
         }
         HotkeyManager.shared.onTrigger = { CaptureController.shared.begin() }
+        HotkeyManager.shared.onFrozenTrigger = { CaptureController.shared.begin(frozen: true) }
+        HotkeyManager.shared.onRecordTrigger = {
+            if RecordingController.shared.state.isRecording {
+                RecordingController.shared.stop()
+            } else {
+                CaptureController.shared.begin(recording: true)
+            }
+        }
         HotkeyManager.shared.register()
     }
+
+    func applicationWillTerminate(_ notification: Notification) {
+        RecordingController.shared.finalizeForTermination()
+    }
 }
 
 struct MenuView: View {
     @State private var granted = PermissionManager.isGranted
+    @ObservedObject private var recordingState = RecordingController.shared.state
 
     var body: some View {
         Button("Capture & OCR") { CaptureController.shared.begin() }
             .keyboardShortcut("s", modifiers: [.command, .shift])
+            .disabled(recordingState.isRecording)
+
+        Button("Capture Frozen Screen") { CaptureController.shared.begin(frozen: true) }
+            .keyboardShortcut("s", modifiers: [.command, .shift, .control])
+            .disabled(recordingState.isRecording)
+
+        if recordingState.isRecording {
+            Button("Stop Recording") { RecordingController.shared.stop() }
+                .keyboardShortcut("r", modifiers: [.command, .shift, .control])
+        } else {
+            Button("Record Screen…") { CaptureController.shared.begin(recording: true) }
+                .keyboardShortcut("r", modifiers: [.command, .shift, .control])
+        }
 
         Button("Extract from Image File…") { ResultWindow.shared.showOpenPanel() }
 
@@ -119,33 +145,52 @@ enum LoginItem {
     }
 }
 
-func runOCRSelfTest() {
+func renderTestImage(_ text: String, fontSize: CGFloat = 56, width: CGFloat = 1900, height: CGFloat = 220) -> CGImage? {
-    let text = "Kapture OCR test 1234, hello world"
+    let size = NSSize(width: width, height: height)
-    let size = NSSize(width: 1600, height: 220)
     guard let rep = NSBitmapImageRep(
         bitmapDataPlanes: nil, pixelsWide: Int(size.width), pixelsHigh: Int(size.height),
         bitsPerSample: 8, samplesPerPixel: 4, hasAlpha: true, isPlanar: false,
         colorSpaceName: .deviceRGB, bytesPerRow: 0, bitsPerPixel: 0
-    ) else { exit(1) }
+    ) else { return nil }
-
     NSGraphicsContext.saveGraphicsState()
     NSGraphicsContext.current = NSGraphicsContext(bitmapImageRep: rep)
     NSColor.white.setFill()
     NSRect(origin: .zero, size: size).fill()
     let attrs: [NSAttributedString.Key: Any] = [
-        .font: NSFont.boldSystemFont(ofSize: 56),
+        .font: NSFont.systemFont(ofSize: fontSize),
         .foregroundColor: NSColor.black,
     ]
-    text.draw(at: NSPoint(x: 60, y: 75), withAttributes: attrs)
+    // Centre the text vertically so tiny canvases still contain it.
+    let textHeight = ("Ag" as NSString).size(withAttributes: attrs).height
+    let y = max(4, (size.height - textHeight) / 2)
+    text.draw(at: NSPoint(x: 20, y: y), withAttributes: attrs)
     NSGraphicsContext.restoreGraphicsState()
+    return rep.cgImage
+}
 
-    guard let cgImage = rep.cgImage else { exit(1) }
+func runOCRSelfTest() {
-    OCRService.shared.recognize(cgImage) { recognized in
+    let cases: [(expected: String, contains: String, fontSize: CGFloat, width: CGFloat, height: CGFloat)] = [
-        print("expected: \(text)")
+        ("Kapture OCR test 1234, hello world", "Kapture OCR test 1234", 56, 1900, 220),
+        ("你好世界，这是一个测试", "你好世界", 56, 1900, 220),
+        // Tiny text: needs the upscale retry to read anything at all.
+        ("会话导出并发冲突报错 export fail", "导出", 7, 320, 40),
+    ]
+    var index = 0
+
+    func runNext() {
+        guard index < cases.count else { exit(0) }
+        let current = cases[index]
+        index += 1
+        guard let image = renderTestImage(current.expected, fontSize: current.fontSize, width: current.width, height: current.height) else { exit(1) }
+        OCRService.shared.recognize(image) { recognized in
+            print("case:     \(current.expected) @\(Int(current.fontSize))pt")
             print("got:      \(recognized)")
-        let ok = recognized == text
+            let ok = recognized.contains(current.contains)
             print(ok ? "PASS" : "FAIL")
-        exit(ok ? 0 : 1)
+            if !ok { exit(1) }
+            runNext()
         }
+    }
+    runNext()
     RunLoop.main.run()
 }

Sources/Kapture/CaptureController.swift

@@ -76,29 +76,65 @@ final class CaptureController: NSObject {
 
     // MARK: - Session lifecycle
 
-    func begin() {
+    /// Frozen mode: the screen is snapshotted at hotkey press, so the pointer
+    /// and any hover state (tooltips, open menus) end up in the capture.
+    private(set) var isFrozenMode = false
+    /// Recording mode: the selection flow ends in a recording, not a still.
+    private(set) var isRecordingMode = false
+    private var frozenSnapshots: [ScreenCapture.DisplaySnapshot] = []
+
+    func begin(frozen: Bool = false, recording: Bool = false) {
         guard !isSelecting else { return }
+        // No new captures while a recording is running (the selection
+        // overlays would end up in the video).
+        guard !RecordingController.shared.state.isRecording else { return }
         guard PermissionManager.isGranted else {
             PermissionManager.request()
             return
         }
         isSelecting = true
+        isFrozenMode = frozen && !recording
+        isRecordingMode = recording
         previousApp = NSWorkspace.shared.frontmostApplication
-        // Activation is async on macOS 26; wait a tick so the app is active
+        // With the event tap driving input we must NOT activate: stealing
-        // before the overlays appear (otherwise clicks pass through).
+        // focus turns the target app into background mode (dimmed chrome),
+        // and that is what would end up in the screenshot. Activate only on
+        // the fallback path, where panel events are the only input.
+        let usesTap = AXIsProcessTrusted()
+        if !usesTap {
             NSApp.activate(ignoringOtherApps: true)
-        CaptureController.log("begin: activate issued, active=\(NSApp.isActive)")
+        }
+        CaptureController.log("begin: frozen=\(isFrozenMode) recording=\(recording) usesTap=\(usesTap) active=\(NSApp.isActive)")
+
+        if frozen {
+            ScreenCapture.shared.snapshotAllDisplays(includeCursor: true) { [weak self] snapshots in
+                DispatchQueue.main.async {
+                    guard let self, self.isSelecting else { return }
+                    self.frozenSnapshots = snapshots
+                    CaptureController.log("frozen: captured \(snapshots.count) displays")
+                    self.showOverlays()
+                }
+            }
+        } else {
+            showOverlays()
+        }
+    }
+
+    private func showOverlays() {
         DispatchQueue.main.asyncAfter(deadline: .now() + 0.05) { [weak self] in
             guard let self, self.isSelecting else { return }
-            CaptureController.log("begin: active=\(NSApp.isActive) — showing overlays")
             self.overlays = NSScreen.screens.map { screen in
                 let panel = OverlayWindow(screen: screen, controller: self)
-                (panel.contentView as? OverlayView)?.controller = self
+                if let view = panel.contentView as? OverlayView {
+                    view.controller = self
+                    view.frozenImage = self.frozenSnapshots.first(where: { $0.frame == screen.frame })?.image
+                }
                 return panel
             }
-            self.overlays.forEach { $0.makeKeyAndOrderFront(nil) }
+            // orderFrontRegardless: show without taking key/focus.
+            self.overlays.forEach { $0.orderFrontRegardless() }
             self.overlayIDs = self.overlays.map { $0.windowNumber }
-            CaptureController.log("begin: \(self.overlays.count) overlays, ids=\(self.overlayIDs), active=\(NSApp.isActive)")
+            CaptureController.log("begin: \(self.overlays.count) overlays, ids=\(self.overlayIDs) — prior app stays focused")
             self.installEventTap()
         }
     }
@@ -107,9 +143,20 @@ final class CaptureController: NSObject {
     /// corners come from the window's own alpha).
     private func finishWindowPick(_ windowID: CGWindowID, _ rect: NSRect, ownedRisk: Bool) {
         let ids = overlayIDs
+        let recording = isRecordingMode
         hideOverlays()
         isSelecting = false
         restorePreviousApp()
+        if recording {
+            // Owned-popup risk: record the screen region instead of the
+            // window filter, which could composite the parent into the video.
+            RecordingController.shared.begin(
+                region: rect,
+                windowID: ownedRisk ? nil : windowID,
+                excludingWindowIDs: ids
+            )
+            return
+        }
         DispatchQueue.main.async { [self] in
             if ownedRisk {
                 capture(rect, excludingWindowIDs: ids, roundedCorners: true)
@@ -121,11 +168,38 @@ final class CaptureController: NSObject {
 
     private func finishSelection(_ rect: NSRect) {
         let ids = overlayIDs
+        let frozen = frozenSnapshots
+        let wasFrozen = isFrozenMode
+        let recording = isRecordingMode
         hideOverlays()
         guard rect.width >= 2, rect.height >= 2 else {
             cancelSelection()
             return
         }
+
+        if wasFrozen {
+            let midpoint = NSPoint(x: rect.midX, y: rect.midY)
+            let snapshot = frozen.first(where: { $0.frame.contains(midpoint) }) ?? frozen.first
+            isSelecting = false
+            restorePreviousApp()
+            DispatchQueue.main.async { [self] in
+                if let snapshot, let image = ScreenCapture.cropFrozen(snapshot, to: rect) {
+                    CaptureController.log("frozen: cropped \(Int(rect.width))x\(Int(rect.height)) -> \(image.width)x\(image.height)")
+                    ResultWindow.shared.show(image: image)
+                } else {
+                    ResultWindow.shared.show(image: ResultWindow.failureImage("Could not crop the frozen screen."))
+                }
+            }
+            return
+        }
+
+        if recording {
+            isSelecting = false
+            restorePreviousApp()
+            RecordingController.shared.begin(region: rect, windowID: nil, excludingWindowIDs: ids)
+            return
+        }
+
         DispatchQueue.main.async { [self] in
             capture(rect, excludingWindowIDs: ids)
         }
@@ -144,6 +218,9 @@ final class CaptureController: NSObject {
         marqueeRect = nil
         hoverWindow = nil
         marqueeStart = nil
+        frozenSnapshots = []
+        isFrozenMode = false
+        isRecordingMode = false
     }
 
     private func restorePreviousApp() {
@@ -262,9 +339,21 @@ final class CaptureController: NSObject {
         return nil
     }
 
+    /// System dialogs (permission prompts, System Settings) must always
+    /// receive their clicks even while we're selecting.
+    private func isSystemDialogFrontmost() -> Bool {
+        guard let front = NSWorkspace.shared.frontmostApplication else { return false }
+        let id = front.bundleIdentifier ?? ""
+        return id.hasPrefix("com.apple.security") || id.hasPrefix("com.apple.SystemSettings") || id == "com.apple.systempreferences"
+    }
+
     /// Returns the event to pass through, or nil to consume it.
     func handleTap(_ type: CGEventType, _ event: CGEvent) -> CGEvent? {
         guard isSelecting else { return event }
+        if isSystemDialogFrontmost() {
+            CaptureController.log("tap: system dialog frontmost — passing through \(type)")
+            return event
+        }
 
         switch type {
         case .tapDisabledByUserInput, .tapDisabledByTimeout:
@@ -289,6 +378,8 @@ final class CaptureController: NSObject {
             }
             let appKitPoint = cgPointToAppKit(NSPoint(x: event.location.x, y: event.location.y))
             if type == .mouseMoved {
+                // Frozen mode is marquee-only: no window hover detection.
+                if isFrozenMode { return event }
                 let window = topmostWindowContaining(appKitPoint)
                 if window?.frame != hoverFrame {
                     hoverWindow = window
@@ -301,8 +392,11 @@ final class CaptureController: NSObject {
                 }
                 return event
             }
-            // Drag: grow the marquee
+            // Drag: grow the marquee — but only after real movement, so a
+            // click's micro-jitter keeps the hovered window eligible.
             guard let start = marqueeStart else { return nil }
+            let distance = hypot(appKitPoint.x - start.x, appKitPoint.y - start.y)
+            guard distance >= 4 else { return nil }
             let rect = NSRect(
                 x: min(start.x, appKitPoint.x),
                 y: min(start.y, appKitPoint.y),
@@ -325,12 +419,18 @@ final class CaptureController: NSObject {
             if let rect = marqueeRect, dragDistance >= 4, rect.width >= 2, rect.height >= 2 {
                 CaptureController.log("tap: marquee complete \(rect)")
                 finishSelection(rect)
+            } else if isFrozenMode {
+                CaptureController.log("tap: frozen click without drag — cancel")
+                cancelSelection()
             } else if let hover = hoverWindow {
                 // An overlapping window from the same app means this is an
                 // owned popup — desktopIndependentWindow would composite the
                 // parent too, so capture the screen crop instead.
+                // Owned popups composite their parent even when the windows
+                // don't overlap, so ANY other same-app window on screen means
+                // crop (correct region) instead of the group composite.
                 let ownedRisk = trackedWindows.contains {
-                    $0.id != hover.id && $0.pid == hover.pid && $0.frame.intersects(hover.frame)
+                    $0.id != hover.id && $0.pid == hover.pid
                 }
                 CaptureController.log("tap: window pick id=\(hover.id) pid=\(hover.pid) ownedRisk=\(ownedRisk) \(hover.frame)")
                 finishWindowPick(hover.id, hover.frame, ownedRisk: ownedRisk)
@@ -369,6 +469,41 @@ final class ScreenCapture {
 
     private init() {}
 
+    /// Estimate the window's corner radius (in image pixels) by scanning the
+    /// top-left corner: the crop's corner pixels are the desktop background;
+    /// app content starts where the color departs from it.
+    static func detectCornerRadius(of image: CGImage) -> CGFloat? {
+        let w = image.width, h = image.height
+        guard w > 30, h > 30 else { return nil }
+        var px = [UInt8](repeating: 0, count: w * h * 4)
+        guard let c = CGContext(data: &px, width: w, height: h, bitsPerComponent: 8, bytesPerRow: w * 4,
+            space: CGColorSpaceCreateDeviceRGB(),
+            bitmapInfo: CGImageAlphaInfo.premultipliedLast.rawValue) else { return nil }
+        c.draw(image, in: CGRect(x: 0, y: 0, width: w, height: h))
+
+        func pixel(x: Int, yTop: Int) -> (r: Int, g: Int, b: Int) {
+            let row = yTop * w * 4 + x * 4
+            return (Int(px[row]), Int(px[row + 1]), Int(px[row + 2]))
+        }
+        let bg = pixel(x: 0, yTop: 0)
+        func diffFromBg(_ p: (r: Int, g: Int, b: Int)) -> Int {
+            abs(p.r - bg.r) + abs(p.g - bg.g) + abs(p.b - bg.b)
+        }
+        var radii: [CGFloat] = []
+        for (limit, axis) in [(h / 3, 0), (w / 3, 1)] {
+            for i in 0..<limit {
+                let p = axis == 0 ? pixel(x: 0, yTop: i) : pixel(x: i, yTop: 0)
+                if diffFromBg(p) > 24 {
+                    radii.append(CGFloat(i))
+                    break
+                }
+            }
+        }
+        guard radii.count == 2 else { return nil }
+        let r = (radii[0] + radii[1]) / 2
+        return (r >= 3 && r <= 40) ? r : nil
+    }
+
     @available(macOS 14.0, *)
     private func crop(_ image: CGImage, to cgRect: CGRect, display: SCDisplay, roundedCorners: Bool = false) -> CGImage? {
         // cgRect is in SCDisplay space (bottom-left origin, Y flipped vs AppKit).
@@ -399,29 +534,123 @@ final class ScreenCapture {
         // Apply a rounded-corner alpha mask for window captures, matching
         // macOS window chrome (≈10pt radius, scaled to image pixels).
         if roundedCorners {
-            let radius = 10 * scaleX
+            // The screen crop contains the app's real rounded corners against
-            let maskRect = CGRect(x: 0, y: 0, width: result.width, height: result.height)
+            // the desktop — measure the radius from the corner pixels, then
-            guard let maskCtx = CGContext(
+            // mask with it (supersampled for smooth edges).
+            let detected = Self.detectCornerRadius(of: result)
+            let radius = min(max((detected.map { $0 * 1.23 + 2.6 }) ?? (10 * scaleX), 4), 40)
+            CaptureController.log("corners: detected=\(detected.map { String(format: "%.1f", $0) } ?? "nil") using=\(Int(radius))px")
+            // Anti-aliased gray mask: supersample 3x, downsample, then draw
+            // the image through it ONCE at 1x (no content resampling blur).
+            let ss = 3
+            let W = result.width * ss, H = result.height * ss
+            guard let hiMask = CGContext(
+                data: nil, width: W, height: H,
+                bitsPerComponent: 8, bytesPerRow: 0,
+                space: CGColorSpaceCreateDeviceGray(),
+                bitmapInfo: CGImageAlphaInfo.none.rawValue
+            ) else { return result }
+            hiMask.setFillColor(CGColor(gray: 1, alpha: 1))
+            hiMask.fill(CGRect(x: 0, y: 0, width: W, height: H))
+            hiMask.setFillColor(CGColor(gray: 0, alpha: 1))
+            let cornerPath = CGMutablePath()
+            cornerPath.addRect(CGRect(x: 0, y: 0, width: W, height: H))
+            cornerPath.addRoundedRect(in: CGRect(x: 0, y: 0, width: W, height: H), cornerWidth: radius * CGFloat(ss), cornerHeight: radius * CGFloat(ss), transform: .identity)
+            hiMask.addPath(cornerPath)
+            hiMask.fillPath(using: .evenOdd)
+            guard let hiMaskImg = hiMask.makeImage(),
+                  let downMask = CGContext(
+                data: nil, width: result.width, height: result.height,
+                bitsPerComponent: 8, bytesPerRow: 0,
+                space: CGColorSpaceCreateDeviceGray(),
+                bitmapInfo: CGImageAlphaInfo.none.rawValue
+            ) else { return result }
+            downMask.interpolationQuality = .high
+            downMask.draw(hiMaskImg, in: CGRect(x: 0, y: 0, width: result.width, height: result.height))
+            guard let mask1x = downMask.makeImage(),
+                  let ctx = CGContext(
                 data: nil, width: result.width, height: result.height,
                 bitsPerComponent: 8, bytesPerRow: 0,
                 space: CGColorSpaceCreateDeviceRGB(),
                 bitmapInfo: CGImageAlphaInfo.premultipliedLast.rawValue
             ) else { return result }
-            maskCtx.setFillColor(CGColor(gray: 1, alpha: 1))
+            ctx.clip(to: CGRect(x: 0, y: 0, width: result.width, height: result.height), mask: mask1x)
-            maskCtx.fill(maskRect)
+            ctx.draw(result, in: CGRect(x: 0, y: 0, width: result.width, height: result.height))
-            maskCtx.setBlendMode(.clear)
+            if let masked = ctx.makeImage() {
-            maskCtx.setFillColor(CGColor(gray: 0, alpha: 1))
-            maskCtx.addPath(CGPath(roundedRect: maskRect, cornerWidth: radius, cornerHeight: radius, transform: nil))
-            maskCtx.fillPath()
-            maskCtx.setBlendMode(.normal)
-            if let mask = maskCtx.makeImage(),
-               let masked = result.masking(mask) {
                 result = masked
             }
         }
         return result
     }
 
+    /// Screen frame (AppKit global points) + full-resolution image per display.
+    struct DisplaySnapshot {
+        let frame: NSRect
+        let image: CGImage
+    }
+
+    /// Snapshot every display right now. Used by frozen mode: the images are
+    /// taken before the overlay appears, so the pointer and hover states
+    /// (tooltips, menus) are baked in.
+    func snapshotAllDisplays(includeCursor: Bool, completion: @escaping ([DisplaySnapshot]) -> Void) {
+        if #available(macOS 14.0, *) {
+            Task {
+                guard let content = try? await SCShareableContent.current else {
+                    completion([])
+                    return
+                }
+                var out: [DisplaySnapshot] = []
+                for display in content.displays {
+                    guard let screen = NSScreen.screens.first(where: {
+                        (($0.deviceDescription[NSDeviceDescriptionKey("NSScreenNumber")] as? NSNumber)?.uint32Value ?? 0) == display.displayID
+                    }) else { continue }
+                    let config = SCStreamConfiguration()
+                    let scale = screen.backingScaleFactor
+                    config.width = Int(CGFloat(display.width) * scale)
+                    config.height = Int(CGFloat(display.height) * scale)
+                    config.showsCursor = includeCursor
+                    let filter = SCContentFilter(display: display, excludingWindows: [])
+                    if let image = try? await SCScreenshotManager.captureImage(contentFilter: filter, configuration: config) {
+                        out.append(DisplaySnapshot(frame: screen.frame, image: image))
+                    }
+                }
+                completion(out)
+            }
+        } else {
+            var out: [DisplaySnapshot] = []
+            for screen in NSScreen.screens {
+                let cgRect = CGRect(
+                    x: screen.frame.minX,
+                    y: (NSScreen.screens.first(where: { $0.frame.origin == .zero })?.frame.height ?? screen.frame.height) - screen.frame.maxY,
+                    width: screen.frame.width,
+                    height: screen.frame.height
+                )
+                if let image = CGWindowListCreateImage(cgRect, .optionOnScreenOnly, kCGNullWindowID, [.bestResolution, .boundsIgnoreFraming]) {
+                    out.append(DisplaySnapshot(frame: screen.frame, image: image))
+                }
+            }
+            completion(out)
+        }
+    }
+
+    /// Crop a selection out of a frozen snapshot (top-left pixel space).
+    static func cropFrozen(_ snapshot: DisplaySnapshot, to rect: NSRect) -> CGImage? {
+        let scaleX = CGFloat(snapshot.image.width) / snapshot.frame.width
+        let scaleY = CGFloat(snapshot.image.height) / snapshot.frame.height
+        guard scaleX > 0, scaleY > 0 else { return nil }
+        let localX = rect.minX - snapshot.frame.minX
+        let localTopY = snapshot.frame.maxY - rect.maxY
+        let px = CGRect(
+            x: localX * scaleX,
+            y: localTopY * scaleY,
+            width: rect.width * scaleX,
+            height: rect.height * scaleY
+        )
+        let safe = px.intersection(CGRect(x: 0, y: 0, width: CGFloat(snapshot.image.width), height: CGFloat(snapshot.image.height))).integral
+        guard safe.width >= 1, safe.height >= 1 else { return nil }
+        return snapshot.image.cropping(to: safe)
+    }
+
     @available(macOS 14.0, *)
     private func displayContaining(_ rect: NSRect, in content: SCShareableContent) -> SCDisplay? {
         let mid = NSPoint(x: rect.midX, y: rect.midY)

Sources/Kapture/HotkeyManager.swift

@@ -4,11 +4,16 @@ import AppKit
 final class HotkeyManager {
     static let shared = HotkeyManager()
 
+    /// ⌘⇧S — live capture (ScreenCaptureKit at capture time).
     var onTrigger: (() -> Void)?
+    /// ⌃⇧⌘S — frozen capture (snapshot at hotkey press).
+    var onFrozenTrigger: (() -> Void)?
+    /// ⌃⇧⌘R — screen recording (region or window); toggles stop.
+    var onRecordTrigger: (() -> Void)?
 
-    private var hotKeyRef: EventHotKeyRef?
+    private var hotKeyRefs: [EventHotKeyRef?] = []
     private var eventHandlerRef: EventHandlerRef?
-    private var hotKeyID = EventHotKeyID(signature: 0x4B505452, id: 1)
+    private let signature: OSType = 0x4B505452
 
     private init() {}
 
@@ -17,18 +22,38 @@ final class HotkeyManager {
             eventClass: OSType(kEventClassKeyboard),
             eventKind: UInt32(kEventHotKeyPressed)
         )
-        InstallEventHandler(GetApplicationEventTarget(), { _, _, _ in
+        InstallEventHandler(GetApplicationEventTarget(), { _, event, _ in
-            HotkeyManager.shared.onTrigger?()
+            var hotKeyID = EventHotKeyID()
+            let status = GetEventParameter(
+                event,
+                EventParamName(kEventParamDirectObject),
+                EventParamType(typeEventHotKeyID),
+                nil,
+                MemoryLayout<EventHotKeyID>.size,
+                nil,
+                &hotKeyID
+            )
+            guard status == noErr else { return noErr }
+            DispatchQueue.main.async {
+                switch hotKeyID.id {
+                case 1: HotkeyManager.shared.onTrigger?()
+                case 2: HotkeyManager.shared.onFrozenTrigger?()
+                case 3: HotkeyManager.shared.onRecordTrigger?()
+                default: break
+                }
+            }
             return noErr
         }, 1, &eventType, nil, &eventHandlerRef)
 
-        RegisterEventHotKey(
+        registerHotkey(id: 1, keyCode: UInt32(kVK_ANSI_S), modifiers: UInt32(cmdKey | shiftKey))
-            UInt32(kVK_ANSI_S),
+        registerHotkey(id: 2, keyCode: UInt32(kVK_ANSI_S), modifiers: UInt32(cmdKey | shiftKey | controlKey))
-            UInt32(cmdKey | shiftKey),
+        registerHotkey(id: 3, keyCode: UInt32(kVK_ANSI_R), modifiers: UInt32(cmdKey | shiftKey | controlKey))
-            hotKeyID,
+    }
-            GetApplicationEventTarget(),
+
-            0,
+    private func registerHotkey(id: UInt32, keyCode: UInt32, modifiers: UInt32) {
-            &hotKeyRef
+        var ref: EventHotKeyRef?
-        )
+        let hotKeyID = EventHotKeyID(signature: signature, id: id)
+        RegisterEventHotKey(keyCode, modifiers, hotKeyID, GetApplicationEventTarget(), 0, &ref)
+        hotKeyRefs.append(ref)
     }
 }

Sources/Kapture/OCRService.swift

@@ -8,17 +8,40 @@ final class OCRService {
 
     func recognize(_ cgImage: CGImage, completion: @escaping (String) -> Void) {
         queue.async {
+            let text = Self.perform(cgImage)
+            // Tiny text (small UI labels) often reads as nothing at native
+            // size; retry upscaled and keep whichever pass read more.
+            guard text.count < 3 else {
+                completion(text)
+                return
+            }
+            var best = text
+            for factor in [CGFloat(2), CGFloat(3)] {
+                guard let upscaled = Self.upscaled(cgImage, factor: factor) else { continue }
+                let candidate = Self.perform(upscaled)
+                if candidate.count > best.count { best = candidate }
+                if best.count >= 3 { break }
+            }
+            completion(best)
+        }
+    }
+
+    private static func perform(_ cgImage: CGImage) -> String {
         let request = VNRecognizeTextRequest()
         request.recognitionLevel = .accurate
         request.usesLanguageCorrection = true
         request.recognitionLanguages = ["en-US", "zh-Hans", "zh-Hant", "ja-JP", "ko-KR"]
+        // Without this, Vision defaults to the first listed language and
+        // renders CJK as Latin garbage (verified: 你好世界 -> "177WR").
+        if #available(macOS 13.0, *) {
+            request.automaticallyDetectsLanguage = true
+        }
 
         let handler = VNImageRequestHandler(cgImage: cgImage, options: [:])
         do {
             try handler.perform([request])
         } catch {
-                completion("")
+            return ""
-                return
         }
 
         let observations = request.results ?? []
@@ -30,7 +53,21 @@ final class OCRService {
                 return a.boundingBox.minX < b.boundingBox.minX
             }
             .compactMap { $0.topCandidates(1).first?.string }
-            completion(lines.joined(separator: "\n"))
+        return lines.joined(separator: "\n")
     }
+
+    private static func upscaled(_ cgImage: CGImage, factor: CGFloat) -> CGImage? {
+        let width = Int(CGFloat(cgImage.width) * factor)
+        let height = Int(CGFloat(cgImage.height) * factor)
+        guard width > 0, height > 0, width <= 20000, height <= 20000 else { return nil }
+        guard let ctx = CGContext(
+            data: nil, width: width, height: height,
+            bitsPerComponent: 8, bytesPerRow: 0,
+            space: CGColorSpaceCreateDeviceRGB(),
+            bitmapInfo: CGImageAlphaInfo.premultipliedLast.rawValue
+        ) else { return nil }
+        ctx.interpolationQuality = .high
+        ctx.draw(cgImage, in: CGRect(x: 0, y: 0, width: CGFloat(width), height: CGFloat(height)))
+        return ctx.makeImage()
     }
 }

Sources/Kapture/OverlayView.swift

@@ -6,8 +6,11 @@ final class OverlayWindow: NSPanel {
     init(screen: NSScreen, controller: CaptureController) {
         let content = OverlayView()
         super.init(
+            // Non-activating: the overlay must NOT steal focus, so the app
+            // being captured keeps its active appearance (undimmed chrome).
+            // Input comes from the CGEventTap, not panel events.
             contentRect: screen.frame,
-            styleMask: [.borderless],
+            styleMask: [.borderless, .nonactivatingPanel],
             backing: .buffered,
             defer: false
         )
@@ -27,6 +30,9 @@ final class OverlayWindow: NSPanel {
 
 final class OverlayView: NSView {
     weak var controller: CaptureController?
+    /// In frozen mode: the still image to display for this screen (already
+    /// full resolution). Nil = live mode (dim the real screen).
+    var frozenImage: CGImage?
 
     override var acceptsFirstResponder: Bool { true }
 
@@ -53,11 +59,90 @@ final class OverlayView: NSView {
         )
     }
 
+    /// Draw the frozen snapshot filling the screen, then replace its centre
+    /// hole with the live... no: the image IS the content, drawn opaque.
+    private func drawFrozen(_ image: CGImage) {
+        guard let ctx = NSGraphicsContext.current?.cgContext else { return }
+        ctx.saveGState()
+        ctx.interpolationQuality = .high
+        // Image is in pixels, view in points — draw into the view bounds.
+        ctx.draw(image, in: bounds)
+        ctx.restoreGState()
+    }
+
+    private func drawSizeLabel(for sel: NSRect, in vf: NSRect) {
+        let label = "\(Int(sel.width)) × \(Int(sel.height))"
+        let attrs: [NSAttributedString.Key: Any] = [
+            .font: NSFont.monospacedDigitSystemFont(ofSize: 12, weight: .medium),
+            .foregroundColor: NSColor.white,
+        ]
+        let labelSize = label.size(withAttributes: attrs)
+        let labelRect = NSRect(
+            x: vf.maxX - labelSize.width - 8,
+            y: vf.minY - labelSize.height - 10,
+            width: labelSize.width + 8,
+            height: labelSize.height + 6
+        )
+        NSColor.black.withAlphaComponent(0.6).setFill()
+        NSBezierPath(roundedRect: labelRect, xRadius: 4, yRadius: 4).fill()
+        label.draw(at: NSPoint(x: labelRect.minX + 4, y: labelRect.minY + 3), withAttributes: attrs)
+    }
+
     override func draw(_ dirtyRect: NSRect) {
+        if let frozenImage {
+            drawFrozen(frozenImage)
+        } else {
+            NSColor.black.withAlphaComponent(0.35).setFill()
+            NSBezierPath(rect: bounds).fill()
+        }
+
+        // Esc hint — the only way out once clicks are consumed by the tap.
+        let hint: String
+        if controller?.isRecordingMode == true {
+            hint = "Drag to select · hover + click a window to record · Esc to cancel"
+        } else if frozenImage != nil {
+            hint = "Frozen screen · drag to select · Esc to cancel"
+        } else {
+            hint = "Drag to select · hover + click a window to capture it · Esc to cancel"
+        }
+        let hintAttrs: [NSAttributedString.Key: Any] = [
+            .font: NSFont.monospacedSystemFont(ofSize: 12, weight: .medium),
+            .foregroundColor: NSColor.white.withAlphaComponent(0.85),
+        ]
+        let hintSize = hint.size(withAttributes: hintAttrs)
+        let hintRect = NSRect(
+            x: (bounds.width - hintSize.width) / 2 - 10,
+            y: bounds.height - 46,
+            width: hintSize.width + 20,
+            height: hintSize.height + 12
+        )
+        NSColor.black.withAlphaComponent(0.65).setFill()
+        NSBezierPath(roundedRect: hintRect, xRadius: 8, yRadius: 8).fill()
+        hint.draw(at: NSPoint(x: hintRect.minX + 10, y: hintRect.minY + 6), withAttributes: hintAttrs)
+
+        guard let controller else { return }
+
+        // Frozen mode: same look as live mode — a dark veil over everything,
+        // with the selection showing the frozen pixels undimmed.
+        if let frozenImage {
             NSColor.black.withAlphaComponent(0.35).setFill()
             NSBezierPath(rect: bounds).fill()
 
-        guard let controller else { return }
+            if let sel = controller.marqueeRect, sel.width > 0, sel.height > 0 {
+                let vf = viewRect(sel)
+                NSGraphicsContext.current?.saveGraphicsState()
+                NSBezierPath(rect: vf).addClip()
+                drawFrozen(frozenImage)
+                NSGraphicsContext.current?.restoreGraphicsState()
+
+                let border = NSBezierPath(rect: vf)
+                border.lineWidth = 1.5
+                NSColor.controlAccentColor.setStroke()
+                border.stroke()
+                drawSizeLabel(for: sel, in: vf)
+            }
+            return
+        }
 
         // Highlight the hovered window (rounded, like real window corners)
         if let frame = controller.hoverFrame {

Sources/Kapture/RecordingController.swift

@@ -0,0 +1,181 @@
+import AppKit
+import SwiftUI
+
+/// Shared recording state, observed by the menu bar menu and control bar.
+final class RecordingState: ObservableObject {
+    @Published var isRecording = false
+    @Published var elapsed: TimeInterval = 0
+}
+
+/// Owns a recording session: the red border around the recorded region and
+/// the floating timer/stop bar. Both chrome windows are excluded from the
+/// video. On stop, the file is auto-saved to ~/Movies and revealed in Finder.
+final class RecordingController {
+    static let shared = RecordingController()
+
+    let state = RecordingState()
+
+    private var recorder: ScreenRecorder?
+    private var chromeWindows: [NSWindow] = []
+    private var timer: Timer?
+
+    /// `region` is in AppKit global points; `windowID` records that window
+    /// instead of the region. Called by CaptureController after the
+    /// selection overlays are hidden.
+    func begin(region: NSRect, windowID: CGWindowID?, excludingWindowIDs: [Int]) {
+        guard !state.isRecording else { return }
+        showChrome(region: region)
+        let excluded = excludingWindowIDs + chromeWindows.map { $0.windowNumber }
+        state.isRecording = true
+
+        let recorder = ScreenRecorder()
+        self.recorder = recorder
+        // Give the window server a beat so the chrome shows up in the
+        // shareable-content snapshot and gets excluded from the video.
+        DispatchQueue.main.asyncAfter(deadline: .now() + 0.15) { [weak self] in
+            guard let self, self.recorder === recorder else { return }
+            recorder.start(region: region, windowID: windowID, excludingWindowIDs: excluded) { [weak self] ok in
+                DispatchQueue.main.async {
+                    guard let self else { return }
+                    if ok {
+                        self.startTimer()
+                    } else {
+                        self.finish(url: nil)
+                        ResultWindow.shared.show(image: ResultWindow.failureImage("Could not start the recording. Check Screen Recording permission."))
+                    }
+                }
+            }
+        }
+    }
+
+    func stop() {
+        guard state.isRecording else { return }
+        state.isRecording = false
+        timer?.invalidate()
+        timer = nil
+        recorder?.stop { [weak self] url in
+            DispatchQueue.main.async { self?.finish(url: url) }
+        }
+    }
+
+    /// Best-effort finalize on quit so the .mov isn't left truncated.
+    /// Blocks briefly — only call from applicationWillTerminate.
+    func finalizeForTermination() {
+        guard let recorder else { return }
+        let semaphore = DispatchSemaphore(value: 0)
+        recorder.stop { _ in semaphore.signal() }
+        _ = semaphore.wait(timeout: .now() + 3)
+    }
+
+    private func finish(url: URL?) {
+        state.isRecording = false
+        hideChrome()
+        recorder = nil
+        state.elapsed = 0
+        if let url {
+            CaptureController.log("rec: saved \(url.path)")
+            NSWorkspace.shared.activateFileViewerSelecting([url])
+        }
+    }
+
+    private func startTimer() {
+        let started = Date()
+        state.elapsed = 0
+        timer = Timer.scheduledTimer(withTimeInterval: 0.5, repeats: true) { [weak self] _ in
+            self?.state.elapsed = Date().timeIntervalSince(started)
+        }
+    }
+
+    // MARK: - Chrome
+
+    private func showChrome(region: NSRect) {
+        // Red border around the recorded region.
+        let border = NSPanel(
+            contentRect: region,
+            styleMask: [.borderless, .nonactivatingPanel],
+            backing: .buffered,
+            defer: false
+        )
+        border.level = .screenSaver
+        border.collectionBehavior = [.canJoinAllSpaces, .fullScreenAuxiliary, .stationary]
+        border.backgroundColor = .clear
+        border.isOpaque = false
+        border.hasShadow = false
+        border.ignoresMouseEvents = true
+        border.contentView = RecordingBorderView()
+        border.orderFrontRegardless()
+
+        // Floating timer/stop bar at the bottom centre of the region.
+        let bar = NSPanel(
+            contentRect: .zero,
+            styleMask: [.borderless, .nonactivatingPanel],
+            backing: .buffered,
+            defer: false
+        )
+        bar.level = .screenSaver
+        bar.collectionBehavior = [.canJoinAllSpaces, .fullScreenAuxiliary, .stationary]
+        bar.backgroundColor = .clear
+        bar.isOpaque = false
+        bar.hasShadow = false
+        let hosting = NSHostingView(rootView: RecordingBarView(state: state) { [weak self] in
+            self?.stop()
+        })
+        let size = hosting.fittingSize
+        bar.setContentSize(size)
+        bar.setFrameOrigin(NSPoint(x: region.midX - size.width / 2, y: region.minY + 14))
+        bar.contentView = hosting
+        bar.orderFrontRegardless()
+
+        chromeWindows = [border, bar]
+    }
+
+    private func hideChrome() {
+        chromeWindows.forEach { $0.orderOut(nil) }
+        chromeWindows = []
+    }
+}
+
+/// Draws the red rounded border around the recorded region.
+private final class RecordingBorderView: NSView {
+    override func draw(_ dirtyRect: NSRect) {
+        let path = NSBezierPath(roundedRect: bounds.insetBy(dx: 2, dy: 2), xRadius: 8, yRadius: 8)
+        NSColor.systemRed.setStroke()
+        path.lineWidth = 3
+        path.stroke()
+    }
+}
+
+/// The floating capsule bar: red dot, elapsed time, stop button.
+struct RecordingBarView: View {
+    @ObservedObject var state: RecordingState
+    let onStop: () -> Void
+
+    private var timeString: String {
+        let total = Int(state.elapsed)
+        return String(format: "%02d:%02d", total / 60, total % 60)
+    }
+
+    var body: some View {
+        HStack(spacing: 10) {
+            Circle()
+                .fill(Color.red)
+                .frame(width: 8, height: 8)
+            Text(timeString)
+                .font(.system(size: 12, weight: .medium).monospacedDigit())
+                .foregroundStyle(.white)
+            Button(action: onStop) {
+                Image(systemName: "stop.fill")
+                    .font(.system(size: 10, weight: .bold))
+                    .foregroundStyle(.white)
+                    .frame(width: 22, height: 22)
+                    .background(Circle().fill(Color.red))
+                    .contentShape(Circle())
+            }
+            .buttonStyle(.plain)
+            .help("Stop recording (⌃⇧⌘R)")
+        }
+        .padding(.horizontal, 12)
+        .padding(.vertical, 7)
+        .background(Capsule().fill(Color.black.opacity(0.78)))
+    }
+}

Sources/Kapture/ResultWindow.swift

@@ -82,7 +82,7 @@ final class ResultWindow {
 
 @MainActor
 final class CaptureModel: ObservableObject {
-    let image: CGImage
+    @Published var image: CGImage
 
     @Published var mode: CaptureMode = .screenshot
     @Published var extractedText: String?
@@ -195,14 +195,15 @@ final class CaptureModel: ObservableObject {
         return rep.cgImage ?? image
     }
 
-    func addStroke(_ points: [CGPoint]) {
+    /// `widthPx` is the stroke thickness in IMAGE PIXELS (the preview
+    /// converts the slider value so exports match what was drawn).
+    func addStroke(_ points: [CGPoint], widthPx: CGFloat) {
         guard points.count > 1 else { return }
         if tool == .redact {
-            // Redaction: a filled black bar from start to end, as thick as the
+            // Redaction: a filled black bar from start to end.
-            // current stroke width (drawn as a rect for clean edges).
+            strokes.append(Stroke(points: points, width: max(widthPx, 12), color: .black, isRedaction: true))
-            strokes.append(Stroke(points: points, width: max(strokeWidth, 12), color: .black, isRedaction: true))
         } else {
-            strokes.append(Stroke(points: points, width: strokeWidth, color: drawColor))
+            strokes.append(Stroke(points: points, width: widthPx, color: drawColor))
         }
         redoStack = []
     }
@@ -214,8 +215,36 @@ final class CaptureModel: ObservableObject {
         return extractedText ?? "Select text in the screenshot, then click “Get Text”."
     }
 
-    func cropImage(to rectInImage: CGRect) -> CGImage? {
+    /// Crop the image to `rectInImage` (image pixel space). Strokes are
-        return image.cropping(to: rectInImage)
+    /// translated into the new coordinate space; strokes fully outside the
+    /// crop are dropped. Selection is cleared.
+    func crop(to rectInImage: CGRect) {
+        let bounds = CGRect(origin: .zero, size: CGSize(width: image.width, height: image.height))
+        let rect = rectInImage.integral.intersection(bounds)
+        guard rect.width >= 2, rect.height >= 2, rect != bounds,
+              let cropped = image.cropping(to: rect) else { return }
+        image = cropped
+        let newBounds = CGRect(origin: .zero, size: rect.size)
+        func translated(_ stroke: Stroke) -> Stroke? {
+            let points = stroke.points.map { CGPoint(x: $0.x - rect.minX, y: $0.y - rect.minY) }
+            let xs = points.map { $0.x }
+            let ys = points.map { $0.y }
+            let strokeBounds = CGRect(
+                x: xs.min()!, y: ys.min()!,
+                width: xs.max()! - xs.min()!, height: ys.max()! - ys.min()!
+            ).insetBy(dx: -stroke.width, dy: -stroke.width)
+            guard strokeBounds.intersects(newBounds) else { return nil }
+            return Stroke(points: points, width: stroke.width, color: stroke.color, isRedaction: stroke.isRedaction)
+        }
+        strokes = strokes.compactMap(translated)
+        redoStack = redoStack.compactMap(translated)
+        selection = nil
+        statusMessage = "Cropped to \(Int(rect.width))×\(Int(rect.height))."
+    }
+
+    func cropToSelection() {
+        guard let sel = selection, hasSelection else { return }
+        crop(to: sel)
     }
 
     func runOCR(over selectionOnly: CGRect? = nil) {
@@ -297,6 +326,139 @@ final class CaptureModel: ObservableObject {
     }
 }
 
+// MARK: - Cap-style design tokens
+
+/// Visual language modeled after the Cap screen recorder (cap.so):
+/// white card surfaces on a light canvas, capsule controls, circular
+/// icon buttons, and Cap's brand blue for primary actions.
+private enum CapStyle {
+    /// Cap's brand blue (#538EFF).
+    static let blue = Color(red: 83 / 255, green: 142 / 255, blue: 1)
+
+    private static func adaptive(_ light: NSColor, _ dark: NSColor) -> Color {
+        Color(nsColor: NSColor(name: nil, dynamicProvider: { appearance in
+            appearance.bestMatch(from: [.darkAqua, .aqua]) == .darkAqua ? dark : light
+        }))
+    }
+
+    /// Pill/track fill (segmented control track, capsule buttons).
+    static let track = adaptive(NSColor(white: 0.93, alpha: 1), NSColor(white: 1, alpha: 0.1))
+    /// Card/content surfaces.
+    static let card = adaptive(.white, NSColor(white: 0.17, alpha: 1))
+    /// Canvas behind the content card.
+    static let canvas = adaptive(NSColor(white: 0.96, alpha: 1), NSColor(white: 0.12, alpha: 1))
+    /// Hairline borders.
+    static let border = adaptive(NSColor(white: 0, alpha: 0.08), NSColor(white: 1, alpha: 0.12))
+    /// Soft blue highlight for active tools (Cap's "On" pill).
+    static let blueTint = adaptive(
+        NSColor(red: 218 / 255, green: 233 / 255, blue: 1, alpha: 1),
+        NSColor(red: 83 / 255, green: 142 / 255, blue: 1, alpha: 0.28)
+    )
+}
+
+private struct CapPressableStyle: ButtonStyle {
+    func makeBody(configuration: Configuration) -> some View {
+        configuration.label
+            .opacity(configuration.isPressed ? 0.7 : 1)
+            .scaleEffect(configuration.isPressed ? 0.96 : 1)
+    }
+}
+
+/// Sliding-pill segmented control (Cap's Screen/Window switcher).
+private struct CapModePicker: View {
+    @Binding var mode: CaptureModel.CaptureMode
+    @Namespace private var selection
+
+    var body: some View {
+        HStack(spacing: 2) {
+            item(.screenshot, title: "Screenshot")
+            item(.text, title: "Text")
+        }
+        .padding(2.5)
+        .background(Capsule().fill(CapStyle.track))
+    }
+
+    private func item(_ value: CaptureModel.CaptureMode, title: String) -> some View {
+        let isSelected = mode == value
+        return Button {
+            withAnimation(.spring(response: 0.28, dampingFraction: 0.85)) { mode = value }
+        } label: {
+            Text(title)
+                .font(.system(size: 12, weight: .medium))
+                .foregroundStyle(isSelected ? Color.primary : Color.secondary)
+                .padding(.horizontal, 13)
+                .padding(.vertical, 4.5)
+                .background {
+                    if isSelected {
+                        Capsule()
+                            .fill(CapStyle.card)
+                            .shadow(color: Color.black.opacity(0.14), radius: 1.5, y: 1)
+                            .matchedGeometryEffect(id: "mode", in: selection)
+                    }
+                }
+                .contentShape(Capsule())
+        }
+        .buttonStyle(CapPressableStyle())
+    }
+}
+
+/// Capsule button; primary renders in Cap blue.
+private struct CapCapsuleButton: View {
+    let title: String
+    let icon: String
+    var primary = false
+    var disabled = false
+    var help: String
+    let action: () -> Void
+
+    var body: some View {
+        Button(action: action) {
+            Label(title, systemImage: icon)
+                .font(.system(size: 12, weight: .medium))
+                .foregroundStyle(primary ? Color.white : Color.primary.opacity(0.85))
+                .padding(.horizontal, 11)
+                .padding(.vertical, 5.5)
+                .background(Capsule().fill(primary ? CapStyle.blue : CapStyle.track))
+                .contentShape(Capsule())
+        }
+        .buttonStyle(CapPressableStyle())
+        .disabled(disabled)
+        .opacity(disabled ? 0.5 : 1)
+        .help(help)
+    }
+}
+
+/// Circular icon button with hover fill and a blue-tinted active state.
+private struct CapIconButton: View {
+    let icon: String
+    var active = false
+    var disabled = false
+    var help: String
+    var shortcut: KeyboardShortcut? = nil
+    let action: () -> Void
+    @State private var hovering = false
+
+    var body: some View {
+        Button(action: action) {
+            Image(systemName: icon)
+                .font(.system(size: 13, weight: .medium))
+                .foregroundStyle(active ? CapStyle.blue : Color.secondary)
+                .frame(width: 29, height: 29)
+                .background(Circle().fill(active ? CapStyle.blueTint : (hovering ? CapStyle.track : Color.clear)))
+                .contentShape(Circle())
+        }
+        .buttonStyle(CapPressableStyle())
+        .keyboardShortcut(shortcut)
+        .onHover { hovering = $0 }
+        .animation(.easeOut(duration: 0.12), value: hovering)
+        .disabled(disabled)
+        .opacity(disabled ? 0.4 : 1)
+        .help(help)
+    }
+}
+
+// MARK: - Result window
+
 struct ResultView: View {
     @ObservedObject var model: CaptureModel
     let onClose: () -> Void
@@ -304,97 +466,62 @@ struct ResultView: View {
     var body: some View {
         VStack(spacing: 0) {
             // Toolbar
-            HStack(spacing: 10) {
+            HStack(spacing: 6) {
-                Picker("", selection: $model.mode) {
+                CapModePicker(mode: $model.mode)
-                    Text("Screenshot").tag(CaptureModel.CaptureMode.screenshot)
-                    Text("Text").tag(CaptureModel.CaptureMode.text)
-                }
-                .pickerStyle(.segmented)
-                .fixedSize()
 
                 Spacer()
 
                 if model.hasSelection {
-                    Button {
+                    CapCapsuleButton(
-                        model.runOCRSelection()
+                        title: "OCR Selection", icon: "text.magnifyingglass",
-                    } label: {
+                        primary: true, disabled: model.isOCRRunning,
-                        Label("OCR Selection", systemImage: "text.magnifyingglass")
+                        help: "Recognize text inside the selected region"
-                    }
+                    ) { model.runOCRSelection() }
-                    .buttonStyle(.borderedProminent)
+
-                    .disabled(model.isOCRRunning)
+                    CapCapsuleButton(
-                    .help("Recognize text inside the selected region")
+                        title: "Crop", icon: "crop",
+                        help: "Crop the image to the selected region"
+                    ) { model.cropToSelection() }
                 }
 
-                Button {
+                CapCapsuleButton(
-                    model.runOCR()
+                    title: "Get All Text", icon: "text.viewfinder",
-                } label: {
+                    primary: !model.hasSelection, disabled: model.isOCRRunning,
-                    Label("Get All Text", systemImage: "text.viewfinder")
+                    help: "Recognize all text in the screenshot"
-                }
+                ) { model.runOCR() }
-                .buttonStyle(.bordered)
-                .disabled(model.isOCRRunning)
-                .help("Recognize all text in the screenshot")
 
-                Divider().frame(height: 18)
+                CapIconButton(
-
+                    icon: model.tool == .draw ? "pencil.tip.crop.circle.fill" : "pencil.tip.crop.circle",
-                Button {
+                    active: model.tool == .draw,
+                    help: "Draw on the screenshot"
+                ) {
                     model.tool = model.tool == .draw ? .select : .draw
                     if model.tool != .select { model.selection = nil }
-                } label: {
-                    Label("Draw", systemImage: model.tool == .draw ? "pencil.tip.crop.circle.fill" : "pencil.tip.crop.circle")
                 }
-                .buttonStyle(.bordered)
-                .foregroundStyle(model.tool == .draw ? .primary : .secondary)
-                .help("Draw on the screenshot")
 
-                Button {
+                CapIconButton(
+                    icon: model.tool == .redact ? "rectangle.fill.on.rectangle.fill" : "rectangle.dashed",
+                    active: model.tool == .redact,
+                    help: "Cover sensitive areas with black bars"
+                ) {
                     model.tool = model.tool == .redact ? .select : .redact
                     if model.tool != .select { model.selection = nil }
-                } label: {
-                    Label("Redact", systemImage: model.tool == .redact ? "rectangle.fill.on.rectangle.fill" : "rectangle.dashed")
                 }
-                .buttonStyle(.bordered)
-                .foregroundStyle(model.tool == .redact ? .primary : .secondary)
-                .help("Cover sensitive areas with black bars")
 
-                Button {
+                CapIconButton(icon: "doc.on.doc", help: "Copy image to clipboard") { model.copyImage() }
-                    model.copyImage()
+                CapIconButton(icon: "square.and.arrow.down", help: "Save screenshot as PNG") { model.savePNG() }
-                } label: {
+                CapIconButton(icon: "xmark", help: "Close") { onClose() }
-                    Label("Copy", systemImage: "doc.on.doc")
             }
-                .buttonStyle(.bordered)
+            .padding(.horizontal, 12)
-                .help("Copy image to clipboard")
+            .padding(.vertical, 9)
 
-                Button {
+            // Tool options (only when a draw tool is active)
-                    model.savePNG()
-                } label: {
-                    Label("Save", systemImage: "square.and.arrow.down")
-                }
-                .buttonStyle(.bordered)
-                .help("Save screenshot as PNG");
-
-                Button {
-                    onClose()
-                } label: {
-                    Image(systemName: "xmark.circle.fill")
-                        .foregroundStyle(.secondary)
-                }
-                .buttonStyle(.plain)
-                .help("Close")
-            }
-            .padding(.horizontal, 14)
-            .padding(.vertical, 10)
-
-            Divider()
-
-            // Tool toolbar (only when a draw tool is active)
             if model.tool != .select {
                 HStack(spacing: 14) {
                     Label(model.tool == .redact ? "Redact" : "Draw", systemImage: model.tool == .redact ? "rectangle.fill" : "pencil.tip")
                         .font(.caption)
                         .foregroundStyle(.secondary)
 
-                    Divider().frame(height: 18)
-
                     if model.tool == .draw {
                         HStack(spacing: 6) {
                             Image(systemName: "circle.fill").font(.system(size: 4))
@@ -404,10 +531,9 @@ struct ResultView: View {
                         .foregroundStyle(.secondary)
                         Slider(value: $model.strokeWidth, in: 1...12, step: 0.5)
                             .frame(width: 120)
+                            .tint(CapStyle.blue)
                             .help("Stroke thickness")
 
-                        Divider().frame(height: 18)
-
                         ColorPicker("", selection: Binding(
                             get: { Color(nsColor: model.drawColor) },
                             set: { model.drawColor = NSColor($0) }
@@ -423,60 +549,44 @@ struct ResultView: View {
 
                     Spacer()
 
-                    Button {
+                    CapIconButton(
-                        model.undoStroke()
+                        icon: "arrow.uturn.backward", disabled: !model.canUndo,
-                    } label: {
+                        help: "Undo last stroke (⌘Z)",
-                        Label("Undo", systemImage: "arrow.uturn.backward")
+                        shortcut: KeyboardShortcut("z", modifiers: .command)
-                    }
+                    ) { model.undoStroke() }
-                    .buttonStyle(.bordered)
-                    .controlSize(.small)
-                    .disabled(!model.canUndo)
-                    .keyboardShortcut("z", modifiers: .command)
-                    .help("Undo last stroke (⌘Z)")
 
-                    Button {
+                    CapIconButton(
-                        model.redoStroke()
+                        icon: "arrow.uturn.forward", disabled: !model.canRedo,
-                    } label: {
+                        help: "Redo stroke (⌘⇧Z)",
-                        Label("Redo", systemImage: "arrow.uturn.forward")
+                        shortcut: KeyboardShortcut("z", modifiers: [.command, .shift])
-                    }
+                    ) { model.redoStroke() }
-                    .buttonStyle(.bordered)
-                    .controlSize(.small)
-                    .disabled(!model.canRedo)
-                    .keyboardShortcut("z", modifiers: [.command, .shift])
-                    .help("Redo stroke (⌘⇧Z)")
 
-                    Button {
+                    CapIconButton(
-                        model.clearAnnotations()
+                        icon: "eraser", disabled: model.strokes.isEmpty,
-                    } label: {
+                        help: "Remove all drawn annotations"
-                        Label("Clear", systemImage: "eraser")
+                    ) { model.clearAnnotations() }
                 }
-                    .buttonStyle(.bordered)
+                .padding(.horizontal, 12)
-                    .controlSize(.small)
+                .padding(.vertical, 6)
-                    .disabled(model.strokes.isEmpty)
-                    .help("Remove all drawn annotations")
-                }
-                .padding(.horizontal, 14)
-                .padding(.vertical, 8)
-                .background(.bar)
-
-                Divider()
             }
 
             // Content
             if model.mode == .screenshot {
                 ScreenshotPane(model: model)
-                    .padding(12)
+                    .padding(10)
+                    .background(CapStyle.canvas)
             } else {
-                ZStack {
                 ScrollView {
                     Text(model.displayText)
                         .font(.system(.body, design: .monospaced))
                         .textSelection(.enabled)
                         .frame(maxWidth: .infinity, alignment: .leading)
-                            .padding(12)
+                        .padding(14)
                 }
-                }
+                .background(CapStyle.card, in: RoundedRectangle(cornerRadius: 10))
-                .frame(maxWidth: .infinity, maxHeight: .infinity)
+                .overlay(RoundedRectangle(cornerRadius: 10).strokeBorder(CapStyle.border, lineWidth: 1))
+                .padding(10)
+                .background(CapStyle.canvas)
             }
 
             // Status bar
@@ -486,7 +596,7 @@ struct ResultView: View {
                         .controlSize(.small)
                 }
                 Text(model.statusMessage ?? (model.mode == .screenshot
-                    ? (model.hasSelection ? "Region selected — click OCR Selection." : "Drag to select a region, or Get All Text.")
+                    ? (model.hasSelection ? "Region selected — OCR it or Crop." : "Drag to select a region, or Get All Text.")
                     : "Selectable text"))
                     .font(.caption)
                     .foregroundStyle(.secondary)
@@ -495,20 +605,17 @@ struct ResultView: View {
                 Spacer()
 
                 if model.mode == .text, model.extractedText != nil {
-                    Button {
+                    CapCapsuleButton(
-                        model.copyText()
+                        title: "Copy Text", icon: "doc.on.doc",
-                    } label: {
+                        help: "Copy recognized text to the clipboard"
-                        Label("Copy Text", systemImage: "doc.on.doc")
+                    ) { model.copyText() }
-                    }
-                    .buttonStyle(.bordered)
-                    .controlSize(.small)
                 }
             }
-            .padding(.horizontal, 14)
+            .padding(.horizontal, 12)
-            .padding(.vertical, 8)
+            .padding(.vertical, 7)
-            .background(.bar)
         }
-        .frame(minWidth: 520, minHeight: 420)
+        .background(Color(nsColor: .windowBackgroundColor))
+        .frame(minWidth: 600, minHeight: 420)
     }
 }
 
@@ -546,14 +653,24 @@ struct ScreenshotPane: View {
     var body: some View {
         GeometryReader { geo in
             let (scale, drawSize, origin) = geometry(for: geo)
+            let cardRect = CGRect(origin: origin, size: drawSize)
+            let cardClip = RoundedRectangle(cornerRadius: 10).path(in: cardRect)
 
             ZStack {
+                // Card behind the image
+                RoundedRectangle(cornerRadius: 10)
+                    .fill(CapStyle.card)
+                    .shadow(color: Color.black.opacity(0.08), radius: 8, y: 2)
+                    .frame(width: drawSize.width, height: drawSize.height)
+                    .position(x: cardRect.midX, y: cardRect.midY)
+
                 Image(nsImage: NSImage(cgImage: model.image, size: .zero))
                     .resizable()
                     .interpolation(imageSize.width >= geo.size.width ? .high : .none)
                     .aspectRatio(contentMode: .fit)
                     .frame(width: drawSize.width, height: drawSize.height)
-                    .position(x: origin.x + drawSize.width / 2, y: origin.y + drawSize.height / 2)
+                    .clipShape(RoundedRectangle(cornerRadius: 10))
+                    .position(x: cardRect.midX, y: cardRect.midY)
 
                 // Completed strokes
                 Canvas { context, _ in
@@ -575,10 +692,12 @@ struct ScreenshotPane: View {
                             for point in stroke.points.dropFirst() {
                                 path.addLine(to: CGPoint(x: origin.x + point.x * scale, y: origin.y + point.y * scale))
                             }
-                            context.stroke(path, with: .color(Color(nsColor: stroke.color)), lineWidth: stroke.width)
+                            // stroke.width is in image pixels; scale to view points
+                            context.stroke(path, with: .color(Color(nsColor: stroke.color)), lineWidth: stroke.width * scale)
                         }
                     }
                 }
+                .clipShape(cardClip)
                 .allowsHitTesting(false)
 
                 // In-progress stroke
@@ -593,9 +712,17 @@ struct ScreenshotPane: View {
                         }
                         context.stroke(path, with: .color(Color(nsColor: model.drawColor)), lineWidth: model.strokeWidth)
                     }
+                    .clipShape(cardClip)
                     .allowsHitTesting(false)
                 }
 
+                // Card hairline border
+                RoundedRectangle(cornerRadius: 10)
+                    .strokeBorder(CapStyle.border, lineWidth: 1)
+                    .frame(width: drawSize.width, height: drawSize.height)
+                    .position(x: cardRect.midX, y: cardRect.midY)
+                    .allowsHitTesting(false)
+
                 if model.tool == .select, let sel = model.selection {
                     let viewRect = CGRect(
                         x: origin.x + sel.minX * scale,
@@ -604,8 +731,8 @@ struct ScreenshotPane: View {
                         height: sel.height * scale
                     )
                     Rectangle()
-                        .fill(Color.accentColor.opacity(0.15))
+                        .fill(CapStyle.blue.opacity(0.15))
-                        .border(Color.accentColor, width: 1.5)
+                        .border(CapStyle.blue, width: 1.5)
                         .frame(width: viewRect.width, height: viewRect.height)
                         .position(x: viewRect.midX, y: viewRect.midY)
                         .allowsHitTesting(false)
@@ -636,7 +763,7 @@ struct ScreenshotPane: View {
                     .onEnded { _ in
                         if model.tool != .select {
                             if currentStroke.count > 1 {
-                                model.addStroke(currentStroke)
+                                model.addStroke(currentStroke, widthPx: model.strokeWidth / max(scale, 0.01))
                             }
                             currentStroke = []
                         } else {

Sources/Kapture/ScreenRecorder.swift

@@ -0,0 +1,203 @@
+import AppKit
+import AVFoundation
+import CoreMedia
+import ScreenCaptureKit
+
+/// Records a screen region (or a single window) to an H.264 .mov via
+/// SCStream -> AVAssetWriter. Works on macOS 13+ (no SCScreenshotManager).
+///
+/// Video-only for now: the audio path is wired behind `includeAudio`
+/// (off) — enabling sound later means setting the flag to true, which
+/// turns on `capturesAudio`, adds an AAC track, and starts appending
+/// audio sample buffers. That path is untested until enabled.
+final class ScreenRecorder: NSObject, SCStreamOutput, SCStreamDelegate {
+    /// Future: also record system audio into the .mov.
+    var includeAudio = false
+
+    private let queue = DispatchQueue(label: "dev.tzgyn.kapture.rec")
+    private var stream: SCStream?
+    private var writer: AVAssetWriter?
+    private var videoInput: AVAssetWriterInput?
+    private var audioInput: AVAssetWriterInput?
+    private var sessionStarted = false
+    private var outputURL: URL?
+
+    /// `region` is in AppKit global points. When `windowID` is given the
+    /// stream records that window directly (and follows it if it moves);
+    /// otherwise it records `region` via a display filter + sourceRect.
+    func start(region: NSRect, windowID: CGWindowID?, excludingWindowIDs: [Int], completion: @escaping (Bool) -> Void) {
+        Task {
+            do {
+                try await prepare(region: region, windowID: windowID, excludingWindowIDs: excludingWindowIDs)
+                completion(true)
+            } catch {
+                CaptureController.log("rec: start failed: \(error.localizedDescription)")
+                completion(false)
+            }
+        }
+    }
+
+    private func prepare(region: NSRect, windowID: CGWindowID?, excludingWindowIDs: [Int]) async throws {
+        let content = try await SCShareableContent.current
+
+        let excludedSet = Set(excludingWindowIDs.map { CGWindowID($0) })
+        let excluded = content.windows.filter { excludedSet.contains($0.windowID) }
+
+        let config = SCStreamConfiguration()
+        let filter: SCContentFilter
+        let pixelWidth: Int
+        let pixelHeight: Int
+
+        let regionCentre = NSPoint(x: region.midX, y: region.midY)
+        let screen = NSScreen.screens.first(where: { $0.frame.contains(regionCentre) }) ?? NSScreen.screens.first
+        let scale = screen?.backingScaleFactor ?? 2
+
+        if let windowID, let scWindow = content.windows.first(where: { $0.windowID == windowID }) {
+            filter = SCContentFilter(desktopIndependentWindow: scWindow)
+            pixelWidth = max(2, Int(scWindow.frame.width * scale) / 2 * 2)
+            pixelHeight = max(2, Int(scWindow.frame.height * scale) / 2 * 2)
+            CaptureController.log("rec: window id=\(windowID) \(pixelWidth)x\(pixelHeight)")
+        } else {
+            guard let screen else {
+                throw NSError(domain: "dev.tzgyn.kapture", code: 1, userInfo: [NSLocalizedDescriptionKey: "No screen for region"])
+            }
+            let cgID = (screen.deviceDescription[NSDeviceDescriptionKey("NSScreenNumber")] as? NSNumber)?.uint32Value ?? 0
+            guard let display = content.displays.first(where: { $0.displayID == cgID }) ?? content.displays.first else {
+                throw NSError(domain: "dev.tzgyn.kapture", code: 2, userInfo: [NSLocalizedDescriptionKey: "No display for region"])
+            }
+            filter = SCContentFilter(display: display, excludingWindows: excluded)
+            // sourceRect is display-local POINTS with a TOP-LEFT origin
+            // (same mapping cropFrozen uses for the frozen snapshots).
+            let local = CGRect(
+                x: region.minX - screen.frame.minX,
+                y: screen.frame.maxY - region.maxY,
+                width: region.width,
+                height: region.height
+            ).intersection(CGRect(x: 0, y: 0, width: display.frame.width, height: display.frame.height))
+            guard local.width >= 2, local.height >= 2 else {
+                throw NSError(domain: "dev.tzgyn.kapture", code: 3, userInfo: [NSLocalizedDescriptionKey: "Region outside display"])
+            }
+            config.sourceRect = local
+            pixelWidth = max(2, Int(local.width * scale) / 2 * 2)
+            pixelHeight = max(2, Int(local.height * scale) / 2 * 2)
+            CaptureController.log("rec: region \(region) sourceRect=\(local) \(pixelWidth)x\(pixelHeight)")
+        }
+
+        config.width = pixelWidth
+        config.height = pixelHeight
+        config.minimumFrameInterval = CMTime(value: 1, timescale: 60)
+        config.pixelFormat = kCVPixelFormatType_32BGRA
+        // Recordings SHOULD include the pointer (unlike stills).
+        config.showsCursor = true
+        config.queueDepth = 6
+        config.capturesAudio = includeAudio
+
+        let movies = FileManager.default.urls(for: .moviesDirectory, in: .userDomainMask).first
+            ?? URL(fileURLWithPath: NSHomeDirectory())
+        let url = movies.appendingPathComponent("Kapture-\(Self.timestamp()).mov")
+        let writer = try AVAssetWriter(url: url, fileType: .mov)
+
+        let videoInput = AVAssetWriterInput(mediaType: .video, outputSettings: [
+            AVVideoCodecKey: AVVideoCodecType.h264,
+            AVVideoWidthKey: pixelWidth,
+            AVVideoHeightKey: pixelHeight,
+        ])
+        videoInput.expectsMediaDataInRealTime = true
+        writer.add(videoInput)
+
+        var audioInput: AVAssetWriterInput?
+        if includeAudio {
+            let input = AVAssetWriterInput(mediaType: .audio, outputSettings: [
+                AVFormatIDKey: kAudioFormatMPEG4AAC,
+                AVSampleRateKey: 48000,
+                AVNumberOfChannelsKey: 2,
+            ])
+            input.expectsMediaDataInRealTime = true
+            writer.add(input)
+            audioInput = input
+        }
+
+        guard writer.startWriting() else {
+            throw writer.error ?? NSError(domain: "dev.tzgyn.kapture", code: 4, userInfo: [NSLocalizedDescriptionKey: "AVAssetWriter failed to start"])
+        }
+
+        let stream = SCStream(filter: filter, configuration: config, delegate: self)
+        try stream.addStreamOutput(self, type: .screen, sampleHandlerQueue: queue)
+        if includeAudio {
+            try stream.addStreamOutput(self, type: .audio, sampleHandlerQueue: queue)
+        }
+        try await stream.startCapture()
+
+        self.stream = stream
+        self.writer = writer
+        self.videoInput = videoInput
+        self.audioInput = audioInput
+        self.outputURL = url
+        CaptureController.log("rec: started -> \(url.lastPathComponent)")
+    }
+
+    /// Stops the stream and finalizes the file. Completion gets the URL on
+    /// success (called on an internal queue — hop to main for UI).
+    func stop(completion: @escaping (URL?) -> Void) {
+        Task {
+            try? await stream?.stopCapture()
+            stream = nil
+            queue.async {
+                guard let writer = self.writer else {
+                    completion(nil)
+                    return
+                }
+                self.videoInput?.markAsFinished()
+                self.audioInput?.markAsFinished()
+                let url = self.outputURL
+                self.writer = nil
+                self.videoInput = nil
+                self.audioInput = nil
+                self.sessionStarted = false
+                writer.finishWriting {
+                    CaptureController.log("rec: finished status=\(writer.status.rawValue)")
+                    completion(writer.status == .completed ? url : nil)
+                }
+            }
+        }
+    }
+
+    // MARK: SCStreamOutput (called on `queue`)
+
+    func stream(_ stream: SCStream, didOutputSampleBuffer sampleBuffer: CMSampleBuffer, of type: SCStreamOutputType) {
+        guard CMSampleBufferDataIsReady(sampleBuffer), let writer, let videoInput else { return }
+        switch type {
+        case .screen:
+            // Skip incomplete frames (macOS 13.1+ exposes a frame status).
+            if #available(macOS 13.1, *) {
+                guard let attachments = (CMSampleBufferGetSampleAttachmentsArray(sampleBuffer, createIfNecessary: false) as? [NSDictionary])?.first,
+                      let dict = attachments as? [String: Any],
+                      let statusRaw = dict[SCStreamFrameInfo.status.rawValue] as? Int,
+                      SCFrameStatus(rawValue: statusRaw) == .complete else { return }
+            }
+            guard videoInput.isReadyForMoreMediaData else { return }
+            if !sessionStarted {
+                sessionStarted = true
+                writer.startSession(atSourceTime: CMSampleBufferGetPresentationTimeStamp(sampleBuffer))
+            }
+            videoInput.append(sampleBuffer)
+        case .audio:
+            guard includeAudio, sessionStarted, let audioInput, audioInput.isReadyForMoreMediaData else { return }
+            audioInput.append(sampleBuffer)
+        default:
+            break
+        }
+    }
+
+    // MARK: SCStreamDelegate
+
+    func stream(_ stream: SCStream, didStopWithError error: Error) {
+        CaptureController.log("rec: stream stopped with error: \(error.localizedDescription)")
+    }
+
+    private static func timestamp() -> String {
+        let formatter = DateFormatter()
+        formatter.dateFormat = "yyyyMMdd-HHmmss"
+        return formatter.string(from: Date())
+    }
+}

build-app.sh

@@ -108,9 +108,9 @@ cat > "$APP/Contents/Info.plist" <<'PLIST'
     <key>CFBundleIconFile</key>
     <string>Icon</string>
     <key>CFBundleShortVersionString</key>
-    <string>1.8.2</string>
+    <string>1.11.0</string>
     <key>CFBundleVersion</key>
-    <string>82</string>
+    <string>110</string>
     <key>LSUIElement</key>
     <true/>
     <key>NSHighResolutionCapable</key>