From d12507632d410ff492ea4f9545c5e68ff9d72b59 Mon Sep 17 00:00:00 2001
From: EtienneLescot <etiennelescot@gmail.com>
Date: Thu, 1 Oct 2026 16:25:43 +0200
Subject: [PATCH 1/2] feat(camera): square camera by default, "Original" keeps
 the camera's own proportions

The picture-in-picture camera now starts square. The other shape, which follows the
camera's own aspect (a portrait camera stays portrait), is labelled "Original" instead
of "Rectangle".

Each shape has its own default roundness, 70% for Square and 40% for Original. An
untouched roundness follows the shape when it changes; a tuned one is kept. Projects
saved with a shape keep the look they had.

The editor's drag box was laid out from a fixed 4:3 camera while the native canvas drew
the real one. Both now read webcamBoxSourceSize, so the box matches the picture.
---
 src/components/ai-edition/PreviewCanvas.tsx   | 33 +++++++---
 .../ai-edition/RightPanes.layout.test.tsx     | 20 ++++++-
 .../ai-edition/RightPanes.tooltips.test.tsx   |  5 +-
 src/components/ai-edition/RightPanes.tsx      | 13 +++-
 .../video-editor/projectPersistence.test.ts   |  2 +-
 src/components/video-editor/types.ts          |  2 +-
 src/i18n/locales/ar/settings.json             |  4 +-
 src/i18n/locales/cs/settings.json             |  4 +-
 src/i18n/locales/de/settings.json             |  4 +-
 src/i18n/locales/en/settings.json             |  4 +-
 src/i18n/locales/es/settings.json             |  4 +-
 src/i18n/locales/fr/settings.json             |  4 +-
 src/i18n/locales/it/settings.json             |  4 +-
 src/i18n/locales/ja-JP/settings.json          |  4 +-
 src/i18n/locales/ko-KR/settings.json          |  4 +-
 src/i18n/locales/pt-BR/settings.json          |  4 +-
 src/i18n/locales/ru/settings.json             |  4 +-
 src/i18n/locales/tr/settings.json             |  4 +-
 src/i18n/locales/vi/settings.json             |  4 +-
 src/i18n/locales/zh-CN/settings.json          |  4 +-
 src/i18n/locales/zh-TW/settings.json          |  4 +-
 .../ai-edition/store/editorSettings.test.ts   |  5 +-
 src/lib/compositeLayout.ts                    |  2 +-
 src/lib/projectDefaults.ts                    | 40 ++++++++++---
 src/native/sceneDescription.test.ts           |  4 +-
 src/native/sceneDescription.ts                | 60 ++++++++++++-------
 website/docs/editing-timeline.md              |  2 +-
 website/docs/guides/product-demo-video.md     |  2 +-
 .../current/editing-timeline.md               |  2 +-
 .../current/guides/product-demo-video.md      |  2 +-
 .../current/editing-timeline.md               |  2 +-
 .../current/guides/product-demo-video.md      |  2 +-
 .../current/editing-timeline.md               |  2 +-
 .../current/guides/product-demo-video.md      |  2 +-
 .../current/editing-timeline.md               |  2 +-
 .../current/guides/product-demo-video.md      |  2 +-
 .../current/editing-timeline.md               |  2 +-
 .../current/guides/product-demo-video.md      |  2 +-
 .../current/editing-timeline.md               |  2 +-
 .../current/guides/product-demo-video.md      |  2 +-
 .../current/editing-timeline.md               |  2 +-
 .../current/guides/product-demo-video.md      |  2 +-
 42 files changed, 181 insertions(+), 97 deletions(-)

diff --git a/src/components/ai-edition/PreviewCanvas.tsx b/src/components/ai-edition/PreviewCanvas.tsx
index b5b72167b..79d95e5ed 100644
--- a/src/components/ai-edition/PreviewCanvas.tsx
+++ b/src/components/ai-edition/PreviewCanvas.tsx
@@ -23,7 +23,7 @@
 // as the user resizes the workbench.
 
 import type { PointerEvent as ReactPointerEvent } from "react";
-import { useEffect, useMemo, useRef, useState } from "react";
+import { useEffect, useMemo, useRef, useState, useSyncExternalStore } from "react";
 import {
 	type CameraFullscreenRegion,
 	type CropRegion,
@@ -62,6 +62,12 @@ import { webcamAnchorAt } from "@/lib/projectDefaults";
 import { wallpaperStyle } from "@/lib/wallpaper";
 import { getCssClipPath } from "@/lib/webcamMaskShapes";
 import { computeCameraFullscreenProgress } from "@/lib/zoomMath/cameraFullscreenUtils";
+import { webcamBoxSourceSize } from "@/native/sceneDescription";
+import {
+	getWebcamNativeSize,
+	getWebcamNativeSizeRevision,
+	subscribeWebcamNativeSize,
+} from "@/native/webcamSizeCache";
 import { AnnotationLayer } from "./AnnotationLayer";
 import { NativeCompositorOverlay } from "./NativeCompositorOverlay";
 import styles from "./NewEditorShell.module.css";
@@ -119,11 +125,6 @@ interface PreviewCanvasProps {
 // (the <video> contain-fits its true ratio within a box sized for the wrong
 // one), on top of the intentional `settings.padding` margin.
 const SCREEN_SOURCE_SIZE = { width: 1920, height: 1080 };
-// ponytail: live preview defaults until the camera <video> reports its real
-// dimensions via loadedmetadata. 4:3 is the legacy default — typical webcams
-// capture at 1.33, and using a 16:9 default collapses vertical-stack to a
-// degenerate full-bleed camera with 0px screen height.
-const WEBCAM_SOURCE_SIZE = { width: 960, height: 720 };
 
 export function PreviewCanvas(props: PreviewCanvasProps) {
 	const te = useScopedT("editor");
@@ -241,6 +242,23 @@ export function PreviewCanvas(props: PreviewCanvasProps) {
 		[assets, props.clips, props.currentTimeSec],
 	);
 	const activeClipHasCamera = Boolean(activeCameraTrack?.visible && activeCameraTrack.sourcePath);
+	// The drag box takes the camera's real proportions, the ones the native canvas draws
+	// (a portrait camera in "Original" is a portrait box), not a fixed 4:3.
+	const webcamSizeRevision = useSyncExternalStore(
+		subscribeWebcamNativeSize,
+		getWebcamNativeSizeRevision,
+		() => 0,
+	);
+	// biome-ignore lint/correctness/useExhaustiveDependencies: the revision re-reads the probed-size cache
+	const webcamSourceSize = useMemo(
+		() =>
+			webcamBoxSourceSize(
+				activeCameraTrack,
+				activeCameraTrack?.sourcePath ? getWebcamNativeSize(activeCameraTrack.sourcePath) : null,
+				settings.webcamCropRegion,
+			),
+		[activeCameraTrack, settings.webcamCropRegion, webcamSizeRevision],
+	);
 
 	const formatFill = useMemo(() => (document ? isFormatFillActive(document) : false), [document]);
 	const layout = useMemo(() => {
@@ -273,7 +291,7 @@ export function PreviewCanvas(props: PreviewCanvasProps) {
 			maxContentSize,
 			// Same box as the scene: a filled format gives the screen the whole padded area.
 			screenSize: formatFill ? maxContentSize : croppedScreenSize,
-			webcamSize: preset === "no-webcam" ? null : WEBCAM_SOURCE_SIZE,
+			webcamSize: preset === "no-webcam" ? null : webcamSourceSize,
 			layoutPreset: preset,
 			webcamSizePreset: settings.webcamSizePreset,
 			// Picture-in-picture only: the block layouts place and round their own camera.
@@ -287,6 +305,7 @@ export function PreviewCanvas(props: PreviewCanvasProps) {
 		screenNativeSize,
 		cropRegion,
 		activeClipHasCamera,
+		webcamSourceSize,
 		settings.webcamLayoutPreset,
 		settings.webcamMaskShape,
 		settings.webcamSizePreset,
diff --git a/src/components/ai-edition/RightPanes.layout.test.tsx b/src/components/ai-edition/RightPanes.layout.test.tsx
index 718abef73..75fd21fe0 100644
--- a/src/components/ai-edition/RightPanes.layout.test.tsx
+++ b/src/components/ai-edition/RightPanes.layout.test.tsx
@@ -237,8 +237,10 @@ describe("LayoutPane picture-in-picture camera", () => {
 		const shapes = within(screen.getByRole("group", { name: "Camera shape" })).getAllByRole(
 			"button",
 		);
-		expect(shapes.map((b) => b.textContent)).toEqual(["Rectangle", "Square"]);
-		fireEvent.click(screen.getByRole("button", { name: "Square" }));
+		expect(shapes.map((b) => b.textContent)).toEqual(["Square", "Original"]);
+		fireEvent.click(within(shapes[1]).getByText("Original"));
+		expect(stored()).toMatchObject({ webcamMaskShape: "rectangle" });
+		fireEvent.click(within(shapes[0]).getByText("Square"));
 		expect(stored()).toMatchObject({ webcamMaskShape: "square" });
 		fireEvent.change(screen.getByRole("slider", { name: "Roundness" }), {
 			target: { value: "100" },
@@ -247,6 +249,20 @@ describe("LayoutPane picture-in-picture camera", () => {
 		expect(stored()).toMatchObject({ webcamMaskShape: "square", webcamRoundness: 1 });
 	});
 
+	it("moves an untouched roundness to each shape's own default, and keeps a tuned one", () => {
+		renderLayout(seedProject(true));
+		const shapes = within(screen.getByRole("group", { name: "Camera shape" })).getAllByRole(
+			"button",
+		);
+		const roundness = () => screen.getByRole("slider", { name: "Roundness" });
+		expect(roundness()).toHaveValue("70");
+		fireEvent.click(shapes[1]);
+		expect(stored()).toMatchObject({ webcamMaskShape: "rectangle", webcamRoundness: 0.4 });
+		fireEvent.change(roundness(), { target: { value: "55" } });
+		fireEvent.click(shapes[0]);
+		expect(stored()).toMatchObject({ webcamMaskShape: "square", webcamRoundness: 0.55 });
+	});
+
 	it("places the camera on one of eight anchors, bottom right by default", () => {
 		renderLayout(seedProject(true));
 		const grid = screen.getByRole("group", { name: "Position" });
diff --git a/src/components/ai-edition/RightPanes.tooltips.test.tsx b/src/components/ai-edition/RightPanes.tooltips.test.tsx
index f31cbade5..46795a985 100644
--- a/src/components/ai-edition/RightPanes.tooltips.test.tsx
+++ b/src/components/ai-edition/RightPanes.tooltips.test.tsx
@@ -174,8 +174,11 @@ describe("inspector labels", () => {
 	it("hangs no tooltip on the camera background labels or the crop zoom value", () => {
 		const root = mount(<LayoutPane />);
 
+		// "Original" names a camera shape too: neither carries a tooltip.
 		for (const name of ["Original", "Blur", "Cutout", "Custom"]) {
-			expect(screen.getByRole("button", { name }).querySelector("[title]")).toBeNull();
+			for (const button of screen.getAllByRole("button", { name })) {
+				expect(button.querySelector("[title]")).toBeNull();
+			}
 		}
 		expect(root.querySelector('[title="Zoom"]')).toBeNull();
 	});
diff --git a/src/components/ai-edition/RightPanes.tsx b/src/components/ai-edition/RightPanes.tsx
index cd67a7e40..fe1f00bf6 100644
--- a/src/components/ai-edition/RightPanes.tsx
+++ b/src/components/ai-edition/RightPanes.tsx
@@ -119,6 +119,7 @@ import {
 } from "@/lib/cursor/cursorThemes";
 import { gradientSeedColor, oneColorGradient } from "@/lib/gradientBuilder";
 import {
+	DEFAULT_WEBCAM_ROUNDNESS,
 	FRAME_THEMES,
 	type FrameTheme,
 	RECORDING_FRAMES,
@@ -2688,8 +2689,8 @@ const NATIVE_WEBCAM_BASE_PCT = 16.7;
 
 // The camera's two proportions. Its rounding is the slider under them, and each icon draws it.
 const CAMERA_SHAPES = [
-	{ value: "rectangle", labelKey: "layout.shapes.rectangle", x: 3, y: 6, w: 18, h: 12 },
 	{ value: "square", labelKey: "layout.shapes.square", x: 4, y: 4, w: 16, h: 16 },
+	{ value: "rectangle", labelKey: "layout.shapes.original", x: 3, y: 6, w: 18, h: 12 },
 ] as const satisfies ReadonlyArray<{ value: WebcamMask } & Record<string, unknown>>;
 
 const ANCHOR_KEYS: Record<WebcamAnchor, string> = {
@@ -2948,7 +2949,13 @@ export function LayoutPane() {
 							value={settings.webcamMaskShape}
 							disabled={layoutControlsDisabled}
 							onChange={(shape) => {
-								void set({ webcamMaskShape: shape });
+								// An untouched roundness follows the shape to its own default; a tuned one stays.
+								const untouched =
+									settings.webcamRoundness === DEFAULT_WEBCAM_ROUNDNESS[settings.webcamMaskShape];
+								void set({
+									webcamMaskShape: shape,
+									...(untouched ? { webcamRoundness: DEFAULT_WEBCAM_ROUNDNESS[shape] } : {}),
+								});
 								if (isNativeCompositorActive()) {
 									setNativeParam("webcamShape", shape);
 								}
@@ -2962,7 +2969,7 @@ export function LayoutPane() {
 							value={Math.round(settings.webcamRoundness * 100)}
 							min={0}
 							max={100}
-							defaultValue={Math.round(DEFAULT_EDITOR_SETTINGS.webcamRoundness * 100)}
+							defaultValue={Math.round(DEFAULT_WEBCAM_ROUNDNESS[settings.webcamMaskShape] * 100)}
 							suffix="%"
 							disabled={layoutControlsDisabled}
 							onChange={(next) => setLive({ webcamRoundness: next / 100 })}
diff --git a/src/components/video-editor/projectPersistence.test.ts b/src/components/video-editor/projectPersistence.test.ts
index f4ed230a6..ed146c59a 100644
--- a/src/components/video-editor/projectPersistence.test.ts
+++ b/src/components/video-editor/projectPersistence.test.ts
@@ -127,7 +127,7 @@ describe("projectPersistence media compatibility", () => {
 		expect(normalizeProjectEditor({ webcamMaskShape: "rounded" }).webcamMaskShape).toBe("rounded");
 		expect(
 			normalizeProjectEditor({ webcamMaskShape: "not-a-real-shape" as never }).webcamMaskShape,
-		).toBe("rectangle");
+		).toBe("square");
 	});
 
 	it("normalizes webcam mirroring safely", () => {
diff --git a/src/components/video-editor/types.ts b/src/components/video-editor/types.ts
index 1d3027a53..89c523bea 100644
--- a/src/components/video-editor/types.ts
+++ b/src/components/video-editor/types.ts
@@ -20,7 +20,7 @@ export const DEFAULT_WEBCAM_LAYOUT_PRESET: WebcamLayoutPreset = "picture-in-pict
 
 export type WebcamMaskShape = "rectangle" | "circle" | "square" | "rounded";
 
-export const DEFAULT_WEBCAM_MASK_SHAPE: WebcamMaskShape = "rectangle";
+export const DEFAULT_WEBCAM_MASK_SHAPE: WebcamMaskShape = "square";
 
 export const DEFAULT_WEBCAM_MIRRORED = false;
 
diff --git a/src/i18n/locales/ar/settings.json b/src/i18n/locales/ar/settings.json
index 499b16f5f..78854c26c 100644
--- a/src/i18n/locales/ar/settings.json
+++ b/src/i18n/locales/ar/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "تأطير كاميرا الويب",
 		"webcamFramingDrag": "اسحب الإطار لتحريكه، واسحب زاويةً للتكبير.",
 		"shapes": {
-			"rectangle": "مستطيل",
-			"square": "مربع"
+			"square": "مربع",
+			"original": "الأصلي"
 		},
 		"webcamPosition": "الموضع",
 		"anchors": {
diff --git a/src/i18n/locales/cs/settings.json b/src/i18n/locales/cs/settings.json
index 3032960cf..c0f0c73eb 100644
--- a/src/i18n/locales/cs/settings.json
+++ b/src/i18n/locales/cs/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Oříznutí kamery",
 		"webcamFramingDrag": "Přetažením rámečku jej posunete, přetažením rohu přiblížíte.",
 		"shapes": {
-			"rectangle": "Obdélník",
-			"square": "Čtverec"
+			"square": "Čtverec",
+			"original": "Původní"
 		},
 		"webcamPosition": "Poloha",
 		"anchors": {
diff --git a/src/i18n/locales/de/settings.json b/src/i18n/locales/de/settings.json
index 47edd219e..3e6af58b1 100644
--- a/src/i18n/locales/de/settings.json
+++ b/src/i18n/locales/de/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Webcam-Ausschnitt",
 		"webcamFramingDrag": "Ziehe den Rahmen, um ihn zu verschieben, oder eine Ecke, um zu zoomen.",
 		"shapes": {
-			"rectangle": "Rechteck",
-			"square": "Quadrat"
+			"square": "Quadrat",
+			"original": "Original"
 		},
 		"webcamPosition": "Position",
 		"anchors": {
diff --git a/src/i18n/locales/en/settings.json b/src/i18n/locales/en/settings.json
index 1ce746841..867827cc1 100644
--- a/src/i18n/locales/en/settings.json
+++ b/src/i18n/locales/en/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Webcam crop",
 		"webcamFramingDrag": "Drag the frame to move it, or a corner to zoom.",
 		"shapes": {
-			"rectangle": "Rectangle",
-			"square": "Square"
+			"square": "Square",
+			"original": "Original"
 		},
 		"webcamPosition": "Position",
 		"anchors": {
diff --git a/src/i18n/locales/es/settings.json b/src/i18n/locales/es/settings.json
index 447e65b30..7d906c4e0 100644
--- a/src/i18n/locales/es/settings.json
+++ b/src/i18n/locales/es/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Encuadre de cámara",
 		"webcamFramingDrag": "Arrastra el marco para moverlo, o una esquina para hacer zoom.",
 		"shapes": {
-			"rectangle": "Rectángulo",
-			"square": "Cuadrado"
+			"square": "Cuadrado",
+			"original": "Original"
 		},
 		"webcamPosition": "Posición",
 		"anchors": {
diff --git a/src/i18n/locales/fr/settings.json b/src/i18n/locales/fr/settings.json
index 4f9501816..7561410e2 100644
--- a/src/i18n/locales/fr/settings.json
+++ b/src/i18n/locales/fr/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Cadrage de la webcam",
 		"webcamFramingDrag": "Faites glisser le cadre pour le déplacer, ou un coin pour zoomer.",
 		"shapes": {
-			"rectangle": "Rectangle",
-			"square": "Carré"
+			"square": "Carré",
+			"original": "Original"
 		},
 		"webcamPosition": "Position",
 		"anchors": {
diff --git a/src/i18n/locales/it/settings.json b/src/i18n/locales/it/settings.json
index e4952c231..555c0dbb3 100644
--- a/src/i18n/locales/it/settings.json
+++ b/src/i18n/locales/it/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Inquadratura webcam",
 		"webcamFramingDrag": "Trascina il riquadro per spostarlo, o un angolo per lo zoom.",
 		"shapes": {
-			"rectangle": "Rettangolo",
-			"square": "Quadrato"
+			"square": "Quadrato",
+			"original": "Originale"
 		},
 		"webcamPosition": "Posizione",
 		"anchors": {
diff --git a/src/i18n/locales/ja-JP/settings.json b/src/i18n/locales/ja-JP/settings.json
index 22f09e78b..dba5bf24f 100644
--- a/src/i18n/locales/ja-JP/settings.json
+++ b/src/i18n/locales/ja-JP/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "ウェブカメラの構図",
 		"webcamFramingDrag": "枠をドラッグして移動、角をドラッグしてズームします。",
 		"shapes": {
-			"rectangle": "長方形",
-			"square": "正方形"
+			"square": "正方形",
+			"original": "オリジナル"
 		},
 		"webcamPosition": "位置",
 		"anchors": {
diff --git a/src/i18n/locales/ko-KR/settings.json b/src/i18n/locales/ko-KR/settings.json
index d57d98046..cc6882854 100644
--- a/src/i18n/locales/ko-KR/settings.json
+++ b/src/i18n/locales/ko-KR/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "웹캠 구도",
 		"webcamFramingDrag": "프레임을 드래그해 옮기고, 모서리를 드래그해 확대/축소하세요.",
 		"shapes": {
-			"rectangle": "직사각형",
-			"square": "정사각형"
+			"square": "정사각형",
+			"original": "원본"
 		},
 		"webcamPosition": "위치",
 		"anchors": {
diff --git a/src/i18n/locales/pt-BR/settings.json b/src/i18n/locales/pt-BR/settings.json
index 835e972bc..56bde8ca9 100644
--- a/src/i18n/locales/pt-BR/settings.json
+++ b/src/i18n/locales/pt-BR/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Enquadramento da webcam",
 		"webcamFramingDrag": "Arraste o quadro para movê-lo, ou um canto para dar zoom.",
 		"shapes": {
-			"rectangle": "Retângulo",
-			"square": "Quadrado"
+			"square": "Quadrado",
+			"original": "Original"
 		},
 		"webcamPosition": "Posição",
 		"anchors": {
diff --git a/src/i18n/locales/ru/settings.json b/src/i18n/locales/ru/settings.json
index 576b2494d..f16d8fc16 100644
--- a/src/i18n/locales/ru/settings.json
+++ b/src/i18n/locales/ru/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Кадрирование веб-камеры",
 		"webcamFramingDrag": "Перетащите рамку, чтобы сдвинуть её, или угол, чтобы изменить масштаб.",
 		"shapes": {
-			"rectangle": "Прямоугольник",
-			"square": "Квадрат"
+			"square": "Квадрат",
+			"original": "Исходная"
 		},
 		"webcamPosition": "Положение",
 		"anchors": {
diff --git a/src/i18n/locales/tr/settings.json b/src/i18n/locales/tr/settings.json
index ae3389387..1f500503e 100644
--- a/src/i18n/locales/tr/settings.json
+++ b/src/i18n/locales/tr/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Webcam kadrajı",
 		"webcamFramingDrag": "Taşımak için çerçeveyi, yakınlaştırmak için bir köşesini sürükleyin.",
 		"shapes": {
-			"rectangle": "Dikdörtgen",
-			"square": "Kare"
+			"square": "Kare",
+			"original": "Orijinal"
 		},
 		"webcamPosition": "Konum",
 		"anchors": {
diff --git a/src/i18n/locales/vi/settings.json b/src/i18n/locales/vi/settings.json
index 283f9c9db..ce6f50f7c 100644
--- a/src/i18n/locales/vi/settings.json
+++ b/src/i18n/locales/vi/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "Khung hình webcam",
 		"webcamFramingDrag": "Kéo khung để di chuyển, hoặc kéo một góc để thu phóng.",
 		"shapes": {
-			"rectangle": "Chữ nhật",
-			"square": "Vuông"
+			"square": "Vuông",
+			"original": "Gốc"
 		},
 		"webcamPosition": "Vị trí",
 		"anchors": {
diff --git a/src/i18n/locales/zh-CN/settings.json b/src/i18n/locales/zh-CN/settings.json
index 8ed665e5a..bfe73d781 100644
--- a/src/i18n/locales/zh-CN/settings.json
+++ b/src/i18n/locales/zh-CN/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "摄像头构图",
 		"webcamFramingDrag": "拖动取景框可移动，拖动一角可缩放。",
 		"shapes": {
-			"rectangle": "矩形",
-			"square": "正方形"
+			"square": "正方形",
+			"original": "原始"
 		},
 		"webcamPosition": "位置",
 		"anchors": {
diff --git a/src/i18n/locales/zh-TW/settings.json b/src/i18n/locales/zh-TW/settings.json
index 1c8912e74..481582a13 100644
--- a/src/i18n/locales/zh-TW/settings.json
+++ b/src/i18n/locales/zh-TW/settings.json
@@ -286,8 +286,8 @@
 		"webcamFraming": "攝影機構圖",
 		"webcamFramingDrag": "拖曳取景框可移動，拖曳一角可縮放。",
 		"shapes": {
-			"rectangle": "矩形",
-			"square": "正方形"
+			"square": "正方形",
+			"original": "原始"
 		},
 		"webcamPosition": "位置",
 		"anchors": {
diff --git a/src/lib/ai-edition/store/editorSettings.test.ts b/src/lib/ai-edition/store/editorSettings.test.ts
index fb1382898..54b9bae3a 100644
--- a/src/lib/ai-edition/store/editorSettings.test.ts
+++ b/src/lib/ai-edition/store/editorSettings.test.ts
@@ -258,7 +258,10 @@ describe("patchEditorSettings", () => {
 		expect(read({ webcamMaskShape: "square" })).toEqual(["square", 0.3]);
 		// A stored roundness wins over the one the shape implied, and stays in 0..1.
 		expect(read({ webcamMaskShape: "circle", webcamRoundness: 0.2 })).toEqual(["square", 0.2]);
-		expect(read({ webcamRoundness: 4 })).toEqual(["rectangle", 1]);
+		expect(read({ webcamRoundness: 4 })).toEqual(["square", 1]);
+		// A stored rectangle keeps the rounding it drew; nothing stored is the factory square.
+		expect(read({ webcamMaskShape: "rectangle" })).toEqual(["rectangle", 0.3]);
+		expect(read({})).toEqual(["square", 0.7]);
 	});
 
 	it("reads a camera size outside the slider's 15–50% into it", () => {
diff --git a/src/lib/compositeLayout.ts b/src/lib/compositeLayout.ts
index ad52114d4..044270ce9 100644
--- a/src/lib/compositeLayout.ts
+++ b/src/lib/compositeLayout.ts
@@ -429,7 +429,7 @@ export function computeCompositeLayout(params: {
 		webcamSizePreset = 25,
 		webcamAnchor = "bottom-right",
 		webcamMaskShape = "rectangle",
-		webcamRoundness = DEFAULT_WEBCAM_ROUNDNESS,
+		webcamRoundness = DEFAULT_WEBCAM_ROUNDNESS.rectangle,
 		frame = "none",
 	} = params;
 	const { width: canvasWidth, height: canvasHeight } = canvasSize;
diff --git a/src/lib/projectDefaults.ts b/src/lib/projectDefaults.ts
index 7246175ff..6071c9b62 100644
--- a/src/lib/projectDefaults.ts
+++ b/src/lib/projectDefaults.ts
@@ -128,15 +128,23 @@ export function readWebcamAnchor(anchor: unknown, legacyPosition: unknown): Webc
 	return "bottom-right";
 }
 
-/** The camera's proportions. Its roundness is a setting of its own, `webcamRoundness`. */
+/**
+ * The camera's proportions: a square crop, or the camera's own ("rectangle", shown as
+ * "Original": a portrait camera stays portrait). Its roundness is a setting of its own,
+ * `webcamRoundness`.
+ */
 export type WebcamMask = "rectangle" | "square";
 
 /**
  * The camera's corner rounding, 0 square to 1 fully round, as a fraction of half the camera's
  * short side. A fraction, so the same value draws the same shape at any size and resolution;
- * at 1 a square camera is a circle.
+ * at 1 a square camera is a circle. Each proportion starts at its own: a square takes more
+ * rounding than a wide camera before it reads as a blob.
  */
-export const DEFAULT_WEBCAM_ROUNDNESS = 0.3;
+export const DEFAULT_WEBCAM_ROUNDNESS: Record<WebcamMask, number> = { square: 0.7, rectangle: 0.4 };
+
+/** What a stored `rectangle` or `square` drew before the roundness was a setting. */
+const LEGACY_WEBCAM_ROUNDNESS = 0.3;
 
 /**
  * The picture-in-picture camera's size, in percent of the frame's short side: the camera's
@@ -149,15 +157,29 @@ export const WEBCAM_SIZE_MAX = 50;
 /**
  * Reads a stored camera shape and roundness. `circle` and `rounded` were a proportion and a
  * rounding folded into one value; they split here into the two settings, the way
- * `readRecordingFrame` splits the old window themes. A stored roundness wins.
+ * `readRecordingFrame` splits the old window themes. A stored roundness wins; with no shape
+ * stored at all, both are the factory default.
  */
 export function readWebcamMask(
 	shape: unknown,
 	roundness: unknown,
 ): { shape: WebcamMask; roundness: number } {
-	const fromShape = shape === "circle" ? 1 : shape === "rounded" ? 0.6 : DEFAULT_WEBCAM_ROUNDNESS;
+	const proportion: WebcamMask =
+		shape === "square" || shape === "circle"
+			? "square"
+			: shape === "rectangle" || shape === "rounded"
+				? "rectangle"
+				: DEFAULT_PROJECT_APPEARANCE.webcamMaskShape;
+	const fromShape =
+		shape === "circle"
+			? 1
+			: shape === "rounded"
+				? 0.6
+				: shape === "rectangle" || shape === "square"
+					? LEGACY_WEBCAM_ROUNDNESS
+					: DEFAULT_WEBCAM_ROUNDNESS[proportion];
 	return {
-		shape: shape === "square" || shape === "circle" ? "square" : "rectangle",
+		shape: proportion,
 		roundness:
 			typeof roundness === "number" && Number.isFinite(roundness)
 				? Math.min(1, Math.max(0, roundness))
@@ -181,7 +203,7 @@ export interface ProjectAppearanceDefaults {
 	borderRadius: number;
 	padding: number;
 	webcamLayoutPreset: "picture-in-picture" | "vertical-stack" | "dual-frame" | "no-webcam";
-	/** The camera's proportions: its own ("rectangle") or cropped square. See `readWebcamMask`. */
+	/** The camera's proportions: cropped square or its own ("rectangle"). See `readWebcamMask`. */
 	webcamMaskShape: WebcamMask;
 	/** 0 square corners to 1 fully round. See `DEFAULT_WEBCAM_ROUNDNESS`. */
 	webcamRoundness: number;
@@ -227,8 +249,8 @@ export const DEFAULT_PROJECT_APPEARANCE: ProjectAppearanceDefaults = {
 	borderRadius: 40,
 	padding: 50,
 	webcamLayoutPreset: "picture-in-picture",
-	webcamMaskShape: "rectangle",
-	webcamRoundness: DEFAULT_WEBCAM_ROUNDNESS,
+	webcamMaskShape: "square",
+	webcamRoundness: DEFAULT_WEBCAM_ROUNDNESS.square,
 	webcamMirrored: false,
 	webcamReactiveZoom: true,
 	webcamSizePreset: 25,
diff --git a/src/native/sceneDescription.test.ts b/src/native/sceneDescription.test.ts
index 2e5a70443..3f85af1b7 100644
--- a/src/native/sceneDescription.test.ts
+++ b/src/native/sceneDescription.test.ts
@@ -1373,7 +1373,7 @@ describe("buildSceneDescription.settings mapping", () => {
 					timelineEndSec: 1,
 				}),
 			],
-			legacyEditor: { webcamSizePreset: 25 },
+			legacyEditor: { webcamSizePreset: 25, webcamMaskShape: "rectangle" },
 		});
 		const scene = buildSceneDescription(doc);
 		const rect = scene.layout.webcamRect;
@@ -1436,7 +1436,7 @@ describe("buildSceneDescription.settings mapping", () => {
 					timelineEndSec: 1,
 				}),
 			],
-			legacyEditor: { webcamSizePreset: 25 },
+			legacyEditor: { webcamSizePreset: 25, webcamMaskShape: "rectangle" },
 		});
 		const scene = buildSceneDescription(doc);
 		const rect = scene.layout.webcamRect;
diff --git a/src/native/sceneDescription.ts b/src/native/sceneDescription.ts
index d308087ca..1f3d36314 100644
--- a/src/native/sceneDescription.ts
+++ b/src/native/sceneDescription.ts
@@ -735,6 +735,37 @@ export function annotationFootageRect(
 	);
 }
 
+type Size = { width: number; height: number };
+
+/**
+ * The size the camera box is laid out from: the camera's own proportions, cropped.
+ *
+ * The order matters. `camera.width/height` comes FIRST because it is the only source
+ * every caller has: it is in the document, so the export dialog and the CLI runner read
+ * it exactly as the preview does. `probed` is second, as a fresher-than-disk override for
+ * the window before the backfill has written the dimensions: it is what a mounted <video>
+ * just reported, and only the preview can ever supply it. The hardcoded 4:3 is last and
+ * is only reached for a document that predates the field, opened with no camera mounted.
+ *
+ * That ordering is the fix: the box used to depend on WHO was asking rather than on what
+ * was recorded, so a 16:9 camera was framed 16:9 in the preview and 4:3 in the export.
+ * The scene and the editor's drag box both read the same answer now.
+ */
+export function webcamBoxSourceSize(
+	camera: { width?: number; height?: number } | null | undefined,
+	probed: Size | null,
+	crop: Size,
+): Size {
+	const source =
+		camera?.width && camera?.height
+			? { width: camera.width, height: camera.height }
+			: (probed ?? { width: 960, height: 720 });
+	return {
+		width: Math.max(1, Math.round(source.width * crop.width)),
+		height: Math.max(1, Math.round(source.height * crop.height)),
+	};
+}
+
 /** Serialize a document into a {@link SceneDescription}. Pure — no per-frame math. */
 export function buildSceneDescription(
 	document: AxcutDocument,
@@ -1089,30 +1120,13 @@ export function buildSceneDescription(
 	 * The camera source SHAPE of a clip, resolved the same way and for the same reasons
 	 * as `screenSourceSizeOf` above: per clip, because two clips need not have been
 	 * recorded with the same camera.
-	 *
-	 * The order matters. `cameraTrack.width/height` comes FIRST because it is the only
-	 * source every caller has: it is in the document, so the export dialog and the CLI
-	 * runner read it exactly as the preview does. `webcamSourceSize` is second, as a
-	 * fresher-than-disk override for the window before the backfill has written the
-	 * dimensions — it is what a mounted <video> just reported, and only the preview can
-	 * ever supply it. The hardcoded 4:3 is last and is now only reached for a document
-	 * that predates the field, opened somewhere with no camera element mounted.
-	 *
-	 * That ordering is the fix: the box used to depend on WHO was asking rather than on
-	 * what was recorded, so a 16:9 camera was framed 16:9 in the preview and 4:3 in the
-	 * export. Everything below reads the same answer now.
 	 */
-	const webcamSourceSizeOf = (clip: AxcutClip) => {
-		const camera = assetById.get(clip.assetId)?.cameraTrack;
-		const source =
-			camera?.width && camera?.height
-				? { width: camera.width, height: camera.height }
-				: (webcamSourceSize ?? { width: 960, height: 720 });
-		return {
-			width: Math.max(1, Math.round(source.width * settings.webcamCropRegion.width)),
-			height: Math.max(1, Math.round(source.height * settings.webcamCropRegion.height)),
-		};
-	};
+	const webcamSourceSizeOf = (clip: AxcutClip) =>
+		webcamBoxSourceSize(
+			assetById.get(clip.assetId)?.cameraTrack,
+			webcamSourceSize,
+			settings.webcamCropRegion,
+		);
 	const layoutForClip = (
 		screenSize: { width: number; height: number },
 		hasCamera: boolean,
diff --git a/website/docs/editing-timeline.md b/website/docs/editing-timeline.md
index d7ea69b84..8bb042ac7 100644
--- a/website/docs/editing-timeline.md
+++ b/website/docs/editing-timeline.md
@@ -31,7 +31,7 @@ A floating icon rail sits over the preview. Five facets:
 | Facet | What it controls |
 |---|---|
 | **Composition** | A background section (image, solid color, or gradient behind your recording; upload your own image or pick from presets), with an **Animation** row (None, Drift, Aurora, Waves) that moves gradients and images alike, and a background blur from 0 to 100%. Then the frame: shadow (None, Light, Medium, Strong), padding, corner roundness, and motion blur. Its **Format** row sets the output shape for preview and export: **Auto** (the default for new projects), which wraps the frame around your recording and camera layout with an even padding border, your clips' own shapes under **Original**, plus 16:9, 9:16, 1:1, 4:3, 4:5, 16:10, and 10:16. |
-| **Camera layout** | Webcam composite: picture-in-picture, vertical stack, dual frame, or no webcam. Mirror and "shrink on zoom." For picture-in-picture: camera shape (rectangle or square), roundness (fully round, a square camera is a circle), size, and position: one of eight spots along the edge, always the same distance from it. Drag the webcam on the canvas and it snaps to the nearest one. |
+| **Camera layout** | Webcam composite: picture-in-picture, vertical stack, dual frame, or no webcam. Mirror and "shrink on zoom." For picture-in-picture: camera shape (square, or original: the camera's own proportions, so a portrait camera stays portrait), roundness (fully round, a square camera is a circle), size, and position: one of eight spots along the edge, always the same distance from it. Drag the webcam on the canvas and it snaps to the nearest one. |
 | **Audio** | The output level, applied the same way in the preview and the export. |
 | **Cursor** | Only meaningful for recordings made in the editable cursor mode, on Windows, macOS, or Linux. Show/hide and auto-hide, the cursor style with its **3D cursor** switch, the cursor types (each type the video shows, drawn as recorded or as the arrow), sliders for size (up to four times the default), smoothing, and motion blur, click bounce (None, Light, Strong), and **Click impact**, which pushes the screen back on each click under every camera. |
 | **Transcript** | The aggregated transcript across every clip, editable — see [Transcript editing](./captions.md#transcript-editing). Its **Captions** button turns captions on, styles them, and translates them — see [Captions & transcript](./captions.md#captions). |
diff --git a/website/docs/guides/product-demo-video.md b/website/docs/guides/product-demo-video.md
index 3c021ab40..9ebd33611 100644
--- a/website/docs/guides/product-demo-video.md
+++ b/website/docs/guides/product-demo-video.md
@@ -52,7 +52,7 @@ The webcam is recorded to its own file, so its placement is an editing decision
 
 - **Picture in picture**, **Vertical stack**, **Dual frame**, or **No webcam**.
 - For every layout: mirror, and a crop of the camera image.
-- For **Picture in picture** only: **Camera shape** (Rectangle or Square), **Roundness** (at 100% a square camera is a circle), a size from 15 to 50% (25% by default), **Position** (a corner or the middle of an edge, bottom right by default), and **Shrink on zoom**, on by default, which makes the camera smaller while a zoom plays so it does not cover the detail. Drag the camera on the canvas and it snaps to the nearest position.
+- For **Picture in picture** only: **Camera shape** (Square by default, or Original, which keeps the camera's own proportions), **Roundness** (70% for Square and 40% for Original by default; at 100% a square camera is a circle), a size from 15 to 50% (25% by default), **Position** (a corner or the middle of an edge, bottom right by default), and **Shrink on zoom**, on by default, which makes the camera smaller while a zoom plays so it does not cover the detail. Drag the camera on the canvas and it snaps to the nearest position.
 - **Camera background**: Original, Blur, Cutout or Custom. Cutout removes the background without a green screen, using a segmentation model that runs on your CPU. This section only appears when the segmentation runtime loads on your machine.
 
 For an intro or outro, press `C` to add a **Full Camera** segment: the camera fills the whole frame for that span.
diff --git a/website/i18n/de/docusaurus-plugin-content-docs/current/editing-timeline.md b/website/i18n/de/docusaurus-plugin-content-docs/current/editing-timeline.md
index 914c17a3b..6c6fefdf3 100644
--- a/website/i18n/de/docusaurus-plugin-content-docs/current/editing-timeline.md
+++ b/website/i18n/de/docusaurus-plugin-content-docs/current/editing-timeline.md
@@ -31,7 +31,7 @@ Alles Folgende beschreibt den Modus **Edit**: oben eine Vorschau mit anpassbarer
 | Tab | Was er steuert |
 |---|---|
 | **Composition** | Ein Abschnitt für den Hintergrund (Bild, Volltonfarbe oder Verlauf hinter deiner Aufnahme; eigenes Bild hochladen oder eine Vorlage wählen), mit einer Zeile **Animation** (None, Drift, Aurora, Waves), die Verläufe und Bilder gleichermaßen bewegt, und einer Hintergrundunschärfe von 0 bis 100 %. Dann der Rahmen: Schatten (None, Light, Medium, Strong), Innenabstand, Eckenrundung und Bewegungsunschärfe. Die Zeile **Format** legt die Ausgabeform für Vorschau und Export fest: **Auto** (Standard für neue Projekte), bei dem sich das Bild mit einem gleichmäßigen Innenabstand als Rand um deine Aufnahme und das Webcam-Layout legt, die eigenen Formen deiner Clips unter **Original**, dazu 16:9, 9:16, 1:1, 4:3, 4:5, 16:10 und 10:16. |
-| **Camera layout** | Webcam-Komposition: Bild-im-Bild, vertikal gestapelt, Doppelrahmen oder keine Webcam. Spiegeln und „Shrink on Zoom“. Für Bild-im-Bild: Kameraform (Rechteck oder Quadrat), Rundung (ganz rund wird eine quadratische Kamera zum Kreis), Größe und Position: einer von acht Plätzen am Rand, immer im gleichen Abstand dazu. Zieh die Webcam auf der Arbeitsfläche, und sie rastet am nächsten Platz ein. |
+| **Camera layout** | Webcam-Komposition: Bild-im-Bild, vertikal gestapelt, Doppelrahmen oder keine Webcam. Spiegeln und „Shrink on Zoom“. Für Bild-im-Bild: Kameraform (Quadrat, oder Original: die eigenen Proportionen der Kamera, eine Hochformat-Kamera bleibt also hochkant), Rundung (ganz rund wird eine quadratische Kamera zum Kreis), Größe und Position: einer von acht Plätzen am Rand, immer im gleichen Abstand dazu. Zieh die Webcam auf der Arbeitsfläche, und sie rastet am nächsten Platz ein. |
 | **Audio** | Der Ausgabepegel, in Vorschau und Export gleich angewendet. |
 | **Cursor** | Nur sinnvoll für Aufnahmen im bearbeitbaren Cursormodus, unter Windows, macOS oder Linux. Ein-/Ausblenden und automatisches Ausblenden, der Cursorstil mit seinem Schalter **3D cursor**, die Cursortypen (jeder Typ, den das Video zeigt, wie aufgezeichnet oder als Pfeil gezeichnet), Regler für Größe (bis zum Vierfachen des Standards), Glättung und Bewegungsunschärfe, Klick-Bounce (None, Light, Strong) und **Click impact**, das den Bildschirm bei jedem Klick nach hinten drückt, unter jeder Kamera. |
 | **Transcript** | Das zusammengeführte Transkript aller Clips, bearbeitbar, siehe [Bearbeitung über das Transkript](./captions.md#transcript-editing). Die Schaltfläche **Captions** darin schaltet Untertitel ein, gestaltet und übersetzt sie, siehe [Untertitel & Transkript](./captions.md#captions). |
diff --git a/website/i18n/de/docusaurus-plugin-content-docs/current/guides/product-demo-video.md b/website/i18n/de/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
index e0540f1d9..3e6b96b31 100644
--- a/website/i18n/de/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
+++ b/website/i18n/de/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
@@ -52,7 +52,7 @@ Die Webcam wird in eine eigene Datei aufgenommen. Ihre Platzierung ist also eine
 
 - **Picture in Picture**, **Vertical Stack**, **Dual Frame** oder **No Webcam**.
 - Für jedes Layout: Spiegeln und ein Zuschnitt des Kamerabilds.
-- Nur für **Picture in Picture**: **Camera Shape** (Rectangle oder Square), **Roundness** (bei 100 % wird eine quadratische Kamera zum Kreis), eine Größe von 15 bis 50 % (standardmäßig 25 %), **Position** (eine Ecke oder die Mitte einer Kante, standardmäßig unten rechts) und **Shrink on Zoom**, standardmäßig an: Es verkleinert die Kamera, während ein Zoom läuft, damit sie das Detail nicht verdeckt. Zieh die Kamera auf der Arbeitsfläche, und sie rastet an der nächsten Position ein.
+- Nur für **Picture in Picture**: **Camera Shape** (standardmäßig Square, oder Original, das die eigenen Proportionen der Kamera behält), **Roundness** (standardmäßig 70 % für Square und 40 % für Original; bei 100 % wird eine quadratische Kamera zum Kreis), eine Größe von 15 bis 50 % (standardmäßig 25 %), **Position** (eine Ecke oder die Mitte einer Kante, standardmäßig unten rechts) und **Shrink on Zoom**, standardmäßig an: Es verkleinert die Kamera, während ein Zoom läuft, damit sie das Detail nicht verdeckt. Zieh die Kamera auf der Arbeitsfläche, und sie rastet an der nächsten Position ein.
 - **Camera Background**: Original, Blur, Cutout oder Custom. Cutout entfernt den Hintergrund ohne Greenscreen, mit einem Segmentierungsmodell, das auf deiner CPU läuft. Dieser Abschnitt erscheint nur, wenn sich die Segmentierungs-Laufzeit auf deinem Rechner laden lässt.
 
 Für ein Intro oder Outro drückst du `C`, um ein **Full Camera**-Segment hinzuzufügen: Die Kamera füllt in diesem Abschnitt das ganze Bild.
diff --git a/website/i18n/es/docusaurus-plugin-content-docs/current/editing-timeline.md b/website/i18n/es/docusaurus-plugin-content-docs/current/editing-timeline.md
index 986e6bdc1..d065f84dc 100644
--- a/website/i18n/es/docusaurus-plugin-content-docs/current/editing-timeline.md
+++ b/website/i18n/es/docusaurus-plugin-content-docs/current/editing-timeline.md
@@ -31,7 +31,7 @@ Sobre la vista previa hay una barra flotante de íconos con cinco paneles:
 | Panel | Qué controla |
 |---|---|
 | **Composición** | Una sección de fondo (imagen, color sólido o degradado detrás de tu grabación; sube tu propia imagen o elige un preajuste), con una fila **Animación** (Ninguno, Deriva, Aurora, Ondas) que mueve por igual degradados e imágenes, y un desenfoque de fondo de 0 a 100 %. Luego el cuadro: sombra (Ninguna, Suave, Media, Fuerte), relleno, redondez de las esquinas y desenfoque de movimiento. Su fila **Formato** define la forma de salida para la vista previa y la exportación: **Auto** (la predeterminada en los proyectos nuevos), que ajusta el cuadro alrededor de tu grabación y de la disposición de cámara con un borde de relleno uniforme, las formas propias de tus clips en **Original**, más 16:9, 9:16, 1:1, 4:3, 4:5, 16:10 y 10:16. |
-| **Disposición de cámara** | Composición de la cámara web: imagen en imagen, apilado vertical, marco dual o sin cámara. Reflejo y "reducir al ampliar". En imagen en imagen: forma de la cámara (rectángulo o cuadrado), redondez (del todo redonda, una cámara cuadrada es un círculo), tamaño y posición: uno de ocho puntos junto al borde, siempre a la misma distancia de él. Arrastra la cámara web sobre el lienzo y se ajusta al punto más cercano. |
+| **Disposición de cámara** | Composición de la cámara web: imagen en imagen, apilado vertical, marco dual o sin cámara. Reflejo y "reducir al ampliar". En imagen en imagen: forma de la cámara (cuadrada, u original: las proporciones propias de la cámara, así que una cámara vertical sigue vertical), redondez (del todo redonda, una cámara cuadrada es un círculo), tamaño y posición: uno de ocho puntos junto al borde, siempre a la misma distancia de él. Arrastra la cámara web sobre el lienzo y se ajusta al punto más cercano. |
 | **Audio** | El nivel de salida, que se aplica igual en la vista previa y en la exportación. |
 | **Cursor** | Solo tiene sentido en grabaciones hechas en el modo de cursor editable, en Windows, macOS o Linux. Mostrar/ocultar y ocultar automáticamente, el estilo del cursor con su interruptor **Cursor 3D**, los tipos de cursor (cada tipo que muestra el video, dibujado tal como se grabó o como la flecha), controles deslizantes de tamaño (hasta cuatro veces el predeterminado), suavizado y desenfoque de movimiento, el rebote al clic (Ninguno, Suave, Marcado) e **Impacto del clic**, que empuja la pantalla hacia atrás en cada clic con cualquier cámara. |
 | **Transcripción** | La transcripción conjunta de todos los clips, editable: consulta [Edición de la transcripción](./captions.md#transcript-editing). Su botón **Subtítulos** activa los subtítulos, les da estilo y los traduce: consulta [Subtítulos y transcripción](./captions.md#captions). |
diff --git a/website/i18n/es/docusaurus-plugin-content-docs/current/guides/product-demo-video.md b/website/i18n/es/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
index f33b19f13..70db3dab5 100644
--- a/website/i18n/es/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
+++ b/website/i18n/es/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
@@ -52,7 +52,7 @@ La cámara web se graba en su propio archivo, así que su ubicación es una deci
 
 - **Imagen en imagen**, **Apilado vertical**, **Marco dual** o **Sin cámara**.
 - En todas las disposiciones: reflejo y un encuadre de la imagen de la cámara.
-- Solo en **Imagen en imagen**: **Forma de cámara** (Rectángulo o Cuadrado), **Redondez** (al 100 % una cámara cuadrada es un círculo), un tamaño del 15 al 50 % (25 % por defecto), **Posición** (una esquina o el centro de un borde, abajo a la derecha por defecto) y **Reducir al ampliar**, activado por defecto, que hace más pequeña la cámara mientras se reproduce un zoom para que no tape el detalle. Arrastra la cámara sobre el lienzo y se ajusta a la posición más cercana.
+- Solo en **Imagen en imagen**: **Forma de cámara** (Cuadrado por defecto, u Original, que conserva las proporciones propias de la cámara), **Redondez** (70 % para Cuadrado y 40 % para Original por defecto; al 100 % una cámara cuadrada es un círculo), un tamaño del 15 al 50 % (25 % por defecto), **Posición** (una esquina o el centro de un borde, abajo a la derecha por defecto) y **Reducir al ampliar**, activado por defecto, que hace más pequeña la cámara mientras se reproduce un zoom para que no tape el detalle. Arrastra la cámara sobre el lienzo y se ajusta a la posición más cercana.
 - **Fondo de la cámara**: Original, Desenfocado, Recortado o Personalizado. Recortado quita el fondo sin pantalla verde, con un modelo de segmentación que se ejecuta en tu CPU. Esta sección solo aparece cuando el entorno de ejecución de segmentación se carga en tu equipo.
 
 Para una introducción o un cierre, presiona `C` para agregar un segmento de **Cámara a pantalla completa**: la cámara llena todo el cuadro durante ese tramo.
diff --git a/website/i18n/fr/docusaurus-plugin-content-docs/current/editing-timeline.md b/website/i18n/fr/docusaurus-plugin-content-docs/current/editing-timeline.md
index f918df713..8b95ebdf0 100644
--- a/website/i18n/fr/docusaurus-plugin-content-docs/current/editing-timeline.md
+++ b/website/i18n/fr/docusaurus-plugin-content-docs/current/editing-timeline.md
@@ -31,7 +31,7 @@ Une barre d'icônes flottante se superpose à l'aperçu. Elle donne accès à ci
 | Onglet | Ce qu'il contrôle |
 |---|---|
 | **Composition** | Une section d'arrière-plan (image, couleur unie ou dégradé derrière votre enregistrement ; importez votre propre image ou choisissez parmi les préréglages), avec une ligne **Animation** (Aucun, Dérive, Aurore, Vagues) qui anime aussi bien les dégradés que les images, et un flou d'arrière-plan de 0 à 100 %. Puis le cadre : l'ombre (Aucune, Légère, Moyenne, Forte), la marge, l'arrondi des coins et le flou de mouvement. Sa ligne **Format** définit la forme de sortie pour l'aperçu et l'export : **Auto** (par défaut pour les nouveaux projets), qui ajuste le cadre autour de votre enregistrement et de la disposition caméra avec une marge égale sur chaque bord, les formes propres à vos clips sous **Original**, plus 16:9, 9:16, 1:1, 4:3, 4:5, 16:10 et 10:16. |
-| **Disposition caméra** | La composition de la webcam : incrustation d'image, empilement vertical, double cadre ou sans webcam. Effet miroir et « réduire au zoom ». En incrustation : forme de la caméra (rectangle ou carré), arrondi (tout à fait ronde, une caméra carrée devient un cercle), taille et position : l'un des huit emplacements le long du bord, toujours à la même distance de celui-ci. Faites glisser la webcam sur le canevas : elle se cale sur l'emplacement le plus proche. |
+| **Disposition caméra** | La composition de la webcam : incrustation d'image, empilement vertical, double cadre ou sans webcam. Effet miroir et « réduire au zoom ». En incrustation : forme de la caméra (carrée, ou originale : les proportions de la caméra elle-même, une caméra en portrait reste en portrait), arrondi (tout à fait ronde, une caméra carrée devient un cercle), taille et position : l'un des huit emplacements le long du bord, toujours à la même distance de celui-ci. Faites glisser la webcam sur le canevas : elle se cale sur l'emplacement le plus proche. |
 | **Audio** | Le niveau de sortie, appliqué de la même façon dans l'aperçu et à l'export. |
 | **Curseur** | Utile uniquement pour les enregistrements faits en mode curseur éditable, sous Windows, macOS ou Linux. Afficher/masquer et le masquage automatique, le style du curseur avec son interrupteur **Curseur 3D**, les types de curseur (chaque type que montre la vidéo, dessiné tel qu'enregistré ou comme la flèche), des glissières pour la taille (jusqu'à quatre fois la taille par défaut), le lissage et le flou de mouvement, le rebond au clic (Aucun, Léger, Marqué), et **Impact du clic**, qui repousse l'écran à chaque clic, avec toutes les caméras. |
 | **Transcription** | La transcription agrégée de tous les clips, modifiable : voir [Montage par la transcription](./captions.md#transcript-editing). Son bouton **Sous-titres** active les sous-titres, les met en forme et les traduit : voir [Sous-titres et transcription](./captions.md#captions). |
diff --git a/website/i18n/fr/docusaurus-plugin-content-docs/current/guides/product-demo-video.md b/website/i18n/fr/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
index c46859f41..3917f1fc3 100644
--- a/website/i18n/fr/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
+++ b/website/i18n/fr/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
@@ -52,7 +52,7 @@ La webcam est enregistrée dans son propre fichier : son placement est donc une
 
 - **Incrustation d'image**, **Empilement vertical**, **Double cadre** ou **Sans webcam**.
 - Pour toutes les dispositions : miroir, et recadrage de l'image de la caméra.
-- Pour **Incrustation d'image** uniquement : **Forme de la caméra** (Rectangle ou Carré), **Arrondi** (à 100 % une caméra carrée devient un cercle), une taille de 15 à 50 % (25 % par défaut), **Position** (un coin ou le milieu d'un bord, en bas à droite par défaut), et **Réduire au zoom**, activé par défaut, qui réduit la caméra pendant un zoom pour qu'elle ne cache pas le détail. Faites glisser la caméra sur le canevas : elle se cale sur la position la plus proche.
+- Pour **Incrustation d'image** uniquement : **Forme de la caméra** (Carré par défaut, ou Original, qui garde les proportions de la caméra), **Arrondi** (70 % pour Carré et 40 % pour Original par défaut ; à 100 % une caméra carrée devient un cercle), une taille de 15 à 50 % (25 % par défaut), **Position** (un coin ou le milieu d'un bord, en bas à droite par défaut), et **Réduire au zoom**, activé par défaut, qui réduit la caméra pendant un zoom pour qu'elle ne cache pas le détail. Faites glisser la caméra sur le canevas : elle se cale sur la position la plus proche.
 - **Arrière-plan de la caméra** : Original, Flouté, Détouré ou Personnalisé. Détouré retire l'arrière-plan sans fond vert, grâce à un modèle de segmentation qui tourne sur votre CPU. Cette section n'apparaît que si le moteur de segmentation se charge sur votre machine.
 
 Pour une introduction ou une conclusion, appuyez sur `C` pour ajouter un segment **Caméra plein écran** : la caméra remplit tout le cadre pendant ce passage.
diff --git a/website/i18n/ja/docusaurus-plugin-content-docs/current/editing-timeline.md b/website/i18n/ja/docusaurus-plugin-content-docs/current/editing-timeline.md
index 140a11b12..21c517724 100644
--- a/website/i18n/ja/docusaurus-plugin-content-docs/current/editing-timeline.md
+++ b/website/i18n/ja/docusaurus-plugin-content-docs/current/editing-timeline.md
@@ -31,7 +31,7 @@ keywords:
 | パネル | 設定できる内容 |
 |---|---|
 | **コンポジション** | 背景のセクション（録画の背景に画像、単色、グラデーションを配置。自分の画像を読み込むか、プリセットから選択）。グラデーションにも画像にも同じように動きを付ける**アニメーション**の行（なし、ドリフト、オーロラ、ウェーブ）と、0〜100% の背景のぼかし。続いてフレーム：影（なし、弱、中、強）、余白、角の丸み、モーションブラー。**フォーマット**の行で、プレビューとエクスポートの出力形状を設定します。**自動**（新規プロジェクトの既定）は、録画とカメラレイアウトを均一な余白で囲むようにフレームを合わせます。**元のサイズ**にはクリップ自体の形状が並び、ほかに 16:9、9:16、1:1、4:3、4:5、16:10、10:16 を選べます。 |
-| **カメラレイアウト** | ウェブカメラの合成：ピクチャーインピクチャ、縦並び、デュアルフレーム、Webカメラなし。左右反転と「ズーム時に縮小」。ピクチャーインピクチャでは、カメラの形状（長方形/正方形）、丸み（最大にすると正方形のカメラは円になります）、サイズ、位置（縁に沿った 8 か所のいずれかで、縁からの距離は常に同じ）を設定できます。キャンバス上でウェブカメラをドラッグすると、いちばん近い位置に吸着します。 |
+| **カメラレイアウト** | ウェブカメラの合成：ピクチャーインピクチャ、縦並び、デュアルフレーム、Webカメラなし。左右反転と「ズーム時に縮小」。ピクチャーインピクチャでは、カメラの形状（正方形、またはオリジナル。オリジナルはカメラ本来の縦横比で、縦向きのカメラは縦向きのまま）、丸み（最大にすると正方形のカメラは円になります）、サイズ、位置（縁に沿った 8 か所のいずれかで、縁からの距離は常に同じ）を設定できます。キャンバス上でウェブカメラをドラッグすると、いちばん近い位置に吸着します。 |
 | **オーディオ** | 出力レベル。プレビューとエクスポートに同じように適用されます。 |
 | **カーソル** | Windows、macOS、Linux で、編集可能なカーソルのモードで録画した場合にのみ意味があります。表示/非表示と自動非表示、**3D カーソル**スイッチを含むカーソルのスタイル、カーソルの種類（動画に現れる種類ごとに、記録どおりに描くか矢印として描くか）、サイズ（既定の最大 4 倍）・スムージング・モーションブラーのスライダー、クリックバウンス（なし、弱、強）、そして**クリックの押し込み**。これはどのカメラでも、クリックのたびに画面を奥へ押し込みます。 |
 | **文字起こし** | すべてのクリップをまとめた文字起こしで、編集できます（[文字起こしの編集](./captions.md#transcript-editing)を参照）。**字幕**ボタンで字幕をオンにし、スタイルを設定し、翻訳できます（[字幕と文字起こし](./captions.md#captions)を参照）。 |
diff --git a/website/i18n/ja/docusaurus-plugin-content-docs/current/guides/product-demo-video.md b/website/i18n/ja/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
index facb72a8e..3b73cab36 100644
--- a/website/i18n/ja/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
+++ b/website/i18n/ja/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
@@ -52,7 +52,7 @@ HUD のすべてのコントロールについては、[画面録画](../recordi
 
 - **ピクチャーインピクチャ**、**縦並び**、**デュアルフレーム**、**Webカメラなし**。
 - すべてのレイアウトに共通：左右反転と、カメラ映像のクロップ。
-- **ピクチャーインピクチャ**のみ：**カメラの形状**（長方形、正方形）、**丸み**（100% で正方形のカメラは円になります）、15〜50% のサイズ（既定は 25%）、**位置**（角または辺の中央。既定は右下）、既定でオンの**ズーム時に縮小**。ズーム時に縮小は、ズームの再生中にカメラを小さくして、細部を隠さないようにします。カメラをキャンバス上でドラッグすると、いちばん近い位置に吸着します。
+- **ピクチャーインピクチャ**のみ：**カメラの形状**（既定は正方形。オリジナルはカメラ本来の縦横比を保ちます）、**丸み**（既定は正方形で 70%、オリジナルで 40%。100% で正方形のカメラは円になります）、15〜50% のサイズ（既定は 25%）、**位置**（角または辺の中央。既定は右下）、既定でオンの**ズーム時に縮小**。ズーム時に縮小は、ズームの再生中にカメラを小さくして、細部を隠さないようにします。カメラをキャンバス上でドラッグすると、いちばん近い位置に吸着します。
 - **カメラ背景**：オリジナル、ぼかし、切り抜き、カスタム。切り抜きは、CPU 上で動くセグメンテーションモデルを使い、グリーンバックなしで背景を取り除きます。このセクションは、お使いのパソコンでセグメンテーションのランタイムを読み込めた場合にのみ表示されます。
 
 イントロやアウトロには、`C` を押して**フルスクリーンカメラ**のセグメントを追加します。その区間では、カメラがフレーム全体を占めます。
diff --git a/website/i18n/pt-BR/docusaurus-plugin-content-docs/current/editing-timeline.md b/website/i18n/pt-BR/docusaurus-plugin-content-docs/current/editing-timeline.md
index 56feaee3a..0571478e0 100644
--- a/website/i18n/pt-BR/docusaurus-plugin-content-docs/current/editing-timeline.md
+++ b/website/i18n/pt-BR/docusaurus-plugin-content-docs/current/editing-timeline.md
@@ -31,7 +31,7 @@ Uma barra flutuante de ícones fica sobre a pré-visualização. São cinco abas
 | Aba | O que controla |
 |---|---|
 | **Composição** | Uma seção de fundo (imagem, cor sólida ou gradiente atrás da gravação; envie sua própria imagem ou escolha uma predefinição), com uma linha **Animação** (Nenhum, Deriva, Aurora, Ondas) que movimenta tanto gradientes quanto imagens, e um desfoque do fundo de 0 a 100%. Depois, o quadro: sombra (Nenhuma, Leve, Média, Forte), espaçamento, arredondamento dos cantos e desfoque de movimento. A linha **Formato** define a proporção de saída da pré-visualização e da exportação: **Auto** (o padrão para projetos novos), que ajusta o quadro em volta da gravação e do layout da câmera com uma borda de espaçamento uniforme, as proporções dos seus clipes em **Original**, além de 16:9, 9:16, 1:1, 4:3, 4:5, 16:10 e 10:16. |
-| **Layout da câmera** | Composição da webcam: picture-in-picture, empilhamento vertical, quadro duplo ou sem webcam. Espelhamento e "encolher ao ampliar". No picture-in-picture: formato da câmera (retângulo ou quadrado), arredondamento (toda arredondada, uma câmera quadrada vira um círculo), tamanho e posição: um de oito pontos junto à borda, sempre à mesma distância dela. Arraste a webcam no canvas e ela se encaixa no ponto mais próximo. |
+| **Layout da câmera** | Composição da webcam: picture-in-picture, empilhamento vertical, quadro duplo ou sem webcam. Espelhamento e "encolher ao ampliar". No picture-in-picture: formato da câmera (quadrado, ou original: as proporções da própria câmera, então uma câmera vertical continua vertical), arredondamento (toda arredondada, uma câmera quadrada vira um círculo), tamanho e posição: um de oito pontos junto à borda, sempre à mesma distância dela. Arraste a webcam no canvas e ela se encaixa no ponto mais próximo. |
 | **Áudio** | O nível de saída, aplicado da mesma forma na pré-visualização e na exportação. |
 | **Cursor** | Só faz sentido para gravações feitas no modo de cursor editável, no Windows, no macOS ou no Linux. Mostrar/ocultar e ocultar automaticamente, o estilo do cursor com a chave **Cursor 3D**, os tipos de cursor (cada tipo que o vídeo mostra, desenhado como foi gravado ou como a seta), controles deslizantes de tamanho (até quatro vezes o padrão), suavização e desfoque de movimento, o rebote ao clicar (Nenhum, Leve, Forte) e **Impacto do clique**, que empurra a tela para trás a cada clique, com qualquer câmera. |
 | **Transcrição** | A transcrição agregada de todos os clipes, editável — veja [Edição da transcrição](./captions.md#transcript-editing). O botão **Legendas** ativa as legendas, define o estilo delas e as traduz — veja [Legendas e transcrição](./captions.md#captions). |
diff --git a/website/i18n/pt-BR/docusaurus-plugin-content-docs/current/guides/product-demo-video.md b/website/i18n/pt-BR/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
index 0222913d4..a4f157e1c 100644
--- a/website/i18n/pt-BR/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
+++ b/website/i18n/pt-BR/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
@@ -52,7 +52,7 @@ A webcam é gravada em um arquivo próprio, então a posição dela é uma decis
 
 - **Picture in Picture**, **Empilhamento Vertical**, **Quadro Duplo** ou **Sem Webcam**.
 - Em todos os layouts: espelhamento e um enquadramento da imagem da câmera.
-- Só em **Picture in Picture**: **Formato da Câmera** (Retângulo ou Quadrado), **Arredondamento** (a 100% uma câmera quadrada vira um círculo), um tamanho de 15 a 50% (25% por padrão), **Posição** (um canto ou o meio de uma borda, embaixo à direita por padrão) e **Encolher ao ampliar**, ativado por padrão, que deixa a câmera menor enquanto um zoom é exibido, para que ela não cubra o detalhe. Arraste a câmera no canvas e ela se encaixa na posição mais próxima.
+- Só em **Picture in Picture**: **Formato da Câmera** (Quadrado por padrão, ou Original, que mantém as proporções da própria câmera), **Arredondamento** (70% para Quadrado e 40% para Original por padrão; a 100% uma câmera quadrada vira um círculo), um tamanho de 15 a 50% (25% por padrão), **Posição** (um canto ou o meio de uma borda, embaixo à direita por padrão) e **Encolher ao ampliar**, ativado por padrão, que deixa a câmera menor enquanto um zoom é exibido, para que ela não cubra o detalhe. Arraste a câmera no canvas e ela se encaixa na posição mais próxima.
 - **Plano de fundo da câmera**: Original, Desfocado, Recorte ou Personalizado. Recorte remove o fundo sem tela verde, usando um modelo de segmentação que roda na sua CPU. Esta seção só aparece quando o runtime de segmentação carrega no seu computador.
 
 Para uma introdução ou um encerramento, pressione `C` para adicionar um segmento de **Câmera em Tela Cheia**: a câmera preenche o quadro inteiro nesse trecho.
diff --git a/website/i18n/zh-CN/docusaurus-plugin-content-docs/current/editing-timeline.md b/website/i18n/zh-CN/docusaurus-plugin-content-docs/current/editing-timeline.md
index db56d9dfc..2be37ae79 100644
--- a/website/i18n/zh-CN/docusaurus-plugin-content-docs/current/editing-timeline.md
+++ b/website/i18n/zh-CN/docusaurus-plugin-content-docs/current/editing-timeline.md
@@ -31,7 +31,7 @@ keywords:
 | 面板 | 控制内容 |
 |---|---|
 | **画面合成** | 先是背景部分（在录制画面后面放置图片、纯色或渐变；可以上传自己的图片，也可以从预设中选择），其中的**动画**一行（无、漂移、极光、波浪）让渐变和图片同样动起来，还有 0 到 100% 的背景模糊。然后是画框：阴影（无、轻、中、强）、内边距、圆角和运动模糊。其中的**格式**一行设置预览和导出的输出形状：**自动**（新项目的默认值）让画面包住录制内容和摄像头布局，四周留出均匀的内边距；**原始**下列出片段自身的形状，另外还有 16:9、9:16、1:1、4:3、4:5、16:10 和 10:16。 |
-| **摄像头布局** | 摄像头的合成方式：画中画、垂直堆叠、双画框或无摄像头。还有镜像和“缩放时缩小”。画中画下可设置摄像头形状（矩形或正方形）、圆角（调到最圆时，正方形摄像头就是圆形）、大小和位置：沿边缘的八个位置之一，与边缘的距离始终相同。在画布上拖动摄像头，它会吸附到最近的位置。 |
+| **摄像头布局** | 摄像头的合成方式：画中画、垂直堆叠、双画框或无摄像头。还有镜像和“缩放时缩小”。画中画下可设置摄像头形状（正方形，或原始：保持摄像头自身的比例，竖屏摄像头仍是竖屏）、圆角（调到最圆时，正方形摄像头就是圆形）、大小和位置：沿边缘的八个位置之一，与边缘的距离始终相同。在画布上拖动摄像头，它会吸附到最近的位置。 |
 | **音频** | 输出电平，在预览和导出中以相同方式应用。 |
 | **光标** | 仅对在 Windows、macOS 或 Linux 上以可编辑光标模式录制的内容有意义。包括显示/隐藏和闲置时自动隐藏，带有**3D 光标**开关的光标样式，光标类型（视频中出现的每种类型，可按录制原样绘制或绘制为箭头），大小（最大为默认值的四倍）、平滑和运动模糊的滑块，点击弹跳（无、轻、强），以及**点击按压**：在任何镜头下，每次点击都会把画面向后推。 |
 | **转录文本** | 汇总所有片段的转录文本，可以编辑，请参阅[编辑转录文本](./captions.md#transcript-editing)。其中的**字幕**按钮用于开启字幕、设置样式和翻译，请参阅[字幕与转录](./captions.md#captions)。 |
diff --git a/website/i18n/zh-CN/docusaurus-plugin-content-docs/current/guides/product-demo-video.md b/website/i18n/zh-CN/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
index 6e23985e2..25e5e0e60 100644
--- a/website/i18n/zh-CN/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
+++ b/website/i18n/zh-CN/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
@@ -52,7 +52,7 @@ HUD 的全部控件请参阅[录制](../recording.md)。
 
 - **画中画**、**垂直堆叠**、**双画框**或**无摄像头**。
 - 所有布局均可设置：镜像，以及裁剪摄像头画面。
-- 仅限**画中画**：**摄像头形状**（矩形或正方形）、**圆角**（100% 时正方形摄像头就是圆形）、15% 到 50% 的大小（默认 25%）、**位置**（某个角或某条边的中点，默认右下），以及默认开启的**缩放时缩小**，它会在播放缩放时把摄像头画面变小，以免挡住细节。在画布上拖动摄像头画面，它会吸附到最近的位置。
+- 仅限**画中画**：**摄像头形状**（默认正方形，或保持摄像头自身比例的原始）、**圆角**（正方形默认 70%，原始默认 40%；100% 时正方形摄像头就是圆形）、15% 到 50% 的大小（默认 25%）、**位置**（某个角或某条边的中点，默认右下），以及默认开启的**缩放时缩小**，它会在播放缩放时把摄像头画面变小，以免挡住细节。在画布上拖动摄像头画面，它会吸附到最近的位置。
 - **摄像头背景**：原画、模糊、抠图或自定义。抠图无需绿幕即可去除背景，使用的是在你的 CPU 上运行的分割模型。只有当分割运行时能在你的电脑上加载时，才会显示这一部分。
 
 如需片头或片尾，按 `C` 添加一个**全屏摄像头**区间：在这段时间内，摄像头画面会占满整个画面。
diff --git a/website/i18n/zh-TW/docusaurus-plugin-content-docs/current/editing-timeline.md b/website/i18n/zh-TW/docusaurus-plugin-content-docs/current/editing-timeline.md
index f65aaf2fa..ec9ef423f 100644
--- a/website/i18n/zh-TW/docusaurus-plugin-content-docs/current/editing-timeline.md
+++ b/website/i18n/zh-TW/docusaurus-plugin-content-docs/current/editing-timeline.md
@@ -31,7 +31,7 @@ keywords:
 | 面板 | 可控制的內容 |
 |---|---|
 | **畫面合成** | 背景區塊（在錄影後方放上圖片、純色或漸層；可以上傳自己的圖片或從預設中挑選），其中的**動畫**列（無、漂移、極光、波浪）讓漸層與圖片同樣動起來，還有 0 到 100% 的背景模糊。接著是畫框：陰影（無、輕、中、強）、內邊距、圓角與動態模糊。其中的**格式**列會設定預覽與匯出的輸出形狀：**自動**（新專案的預設值）會讓畫面包住錄影與攝影機版面，四周留出均勻的內邊距；**原始**底下列出片段本身的形狀，另外還有 16:9、9:16、1:1、4:3、4:5、16:10 與 10:16。 |
-| **攝影機版面** | 網路攝影機的合成方式：子母畫面、垂直堆疊、雙畫框或無網路攝影機。鏡像與「縮放時縮小」。子母畫面下可設定攝影機形狀（矩形或正方形）、圓角（調到最圓時，正方形攝影機就是圓形）、大小與位置：沿邊緣的八個位置之一，與邊緣的距離始終相同。在畫布上拖曳網路攝影機，它會吸附到最近的位置。 |
+| **攝影機版面** | 網路攝影機的合成方式：子母畫面、垂直堆疊、雙畫框或無網路攝影機。鏡像與「縮放時縮小」。子母畫面下可設定攝影機形狀（正方形，或原始：保持攝影機本身的比例，直向攝影機仍是直向）、圓角（調到最圓時，正方形攝影機就是圓形）、大小與位置：沿邊緣的八個位置之一，與邊緣的距離始終相同。在畫布上拖曳網路攝影機，它會吸附到最近的位置。 |
 | **音訊** | 輸出音量，在預覽與匯出中以相同方式套用。 |
 | **游標** | 只對在 Windows、macOS 或 Linux 上以可編輯游標模式錄製的錄影有意義。顯示／隱藏與閒置時自動隱藏，附有**3D 游標**開關的游標樣式，游標類型（影片中出現的每種類型，可依錄製原樣繪製或繪製為箭頭），大小（最大為預設值的四倍）、平滑與動態模糊的滑桿，點擊彈跳（無、輕、強），以及**點擊按壓**：在任何鏡頭下，每次點擊都會把畫面往後推。 |
 | **逐字稿** | 彙整所有片段的逐字稿，可以直接編輯，請參閱[逐字稿編輯](./captions.md#transcript-editing)。其中的**字幕**按鈕可以開啟字幕、設定樣式並翻譯，請參閱[字幕與逐字稿](./captions.md#captions)。 |
diff --git a/website/i18n/zh-TW/docusaurus-plugin-content-docs/current/guides/product-demo-video.md b/website/i18n/zh-TW/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
index c6004a075..111fe38d2 100644
--- a/website/i18n/zh-TW/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
+++ b/website/i18n/zh-TW/docusaurus-plugin-content-docs/current/guides/product-demo-video.md
@@ -52,7 +52,7 @@ OpenScreen 以 60 fps 為擷取目標，在 Windows 與 macOS 上最高可達 38
 
 - **子母畫面**、**垂直堆疊**、**雙畫框**或**無網路攝影機**。
 - 所有版面都可以設定：鏡像，以及裁切攝影機畫面。
-- 僅限**子母畫面**：**攝影機形狀**（矩形或正方形）、**圓角**（100% 時正方形攝影機就是圓形）、15% 到 50% 的大小（預設 25%）、**位置**（某個角或某條邊的中點，預設右下），以及預設開啟的**縮放時縮小**，它會在縮放播放時把攝影機畫面縮小，以免遮住細節。在畫布上拖曳攝影機畫面，它會吸附到最近的位置。
+- 僅限**子母畫面**：**攝影機形狀**（預設正方形，或保持攝影機本身比例的原始）、**圓角**（正方形預設 70%，原始預設 40%；100% 時正方形攝影機就是圓形）、15% 到 50% 的大小（預設 25%）、**位置**（某個角或某條邊的中點，預設右下），以及預設開啟的**縮放時縮小**，它會在縮放播放時把攝影機畫面縮小，以免遮住細節。在畫布上拖曳攝影機畫面，它會吸附到最近的位置。
 - **攝影機背景**：原畫、模糊、去背或自訂。去背不需要綠幕就能移除背景，使用的是在你的 CPU 上執行的分割模型。只有在你的電腦能載入分割模型的執行環境時，才會出現這個區塊。
 
 如果要做開場或結尾，請按 `C` 加入**全螢幕攝影機**片段：在這段範圍內，攝影機畫面會佔滿整個畫面。

From f46dc4089731ed9dd722d67d76e2c8b52a5b616b Mon Sep 17 00:00:00 2001
From: EtienneLescot <etiennelescot@gmail.com>
Date: Thu, 1 Oct 2026 16:33:48 +0200
Subject: [PATCH 2/2] fix(camera): fall back to the given shape's own default
 roundness

---
 src/lib/compositeLayout.ts | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)

diff --git a/src/lib/compositeLayout.ts b/src/lib/compositeLayout.ts
index 044270ce9..891361213 100644
--- a/src/lib/compositeLayout.ts
+++ b/src/lib/compositeLayout.ts
@@ -429,7 +429,9 @@ export function computeCompositeLayout(params: {
 		webcamSizePreset = 25,
 		webcamAnchor = "bottom-right",
 		webcamMaskShape = "rectangle",
-		webcamRoundness = DEFAULT_WEBCAM_ROUNDNESS.rectangle,
+		webcamRoundness = DEFAULT_WEBCAM_ROUNDNESS[
+			webcamMaskShape === "square" || webcamMaskShape === "circle" ? "square" : "rectangle"
+		],
 		frame = "none",
 	} = params;
 	const { width: canvasWidth, height: canvasHeight } = canvasSize;