🎥 Make videos programmatically with React
import React from 'react';
import {AbsoluteFill, HtmlInCanvas, IFrame, Img} from 'remotion';
// Demonstrates the "Privacy-preserving painting" rules from
// https://github.com/WICG/html-in-canvas#privacy-preserving-painting
//
// Each row contains content that the spec says must NOT leak into the
// painted snapshot:
// • Cross-origin iframe content
// • Cross-origin <img> without CORS
// • :visited link styling
// • Spellcheck markers
// • System colors (forced-colors / theme)
//
// Right column (HtmlInCanvas paint) should show those bits redacted/blank;
// left column (plain DOM) shows the full content for comparison.
const Section: React.FC<{
readonly title: string;
readonly children: React.ReactNode;
}> = ({title, children}) => (
<div style={{marginBottom: 14}}>
<div
style={{
fontSize: 12,
color: '#666',
marginBottom: 4,
fontFamily: 'sans-serif',
}}
>
{title}
</div>
{children}
</div>
);
const VISITED_LINK_CSS = `
.privacy-link { color: #0000ee; }
.privacy-link:visited { color: #ff00ff; }
`;
const SensitiveContent: React.FC = () => (
<div
style={{
fontFamily: 'sans-serif',
fontSize: 16,
color: 'black',
padding: 20,
backgroundColor: 'white',
width: 560,
}}
>
<style>{VISITED_LINK_CSS}</style>
<Section title="Cross-origin <iframe> (should paint blank)">
<IFrame
title="cross-origin"
src="https://example.com"
width={500}
height={120}
style={{border: '1px solid #888'}}
/>
</Section>
<Section title="Cross-origin <img> without CORS (should paint blank)">
<Img
src="https://upload.wikimedia.org/wikipedia/commons/thumb/4/47/PNG_transparency_demonstration_1.png/280px-PNG_transparency_demonstration_1.png"
alt="cross-origin"
width={200}
/>
</Section>
<Section title=":visited link (visited color must not leak)">
<a className="privacy-link" href="https://www.google.com">
google.com
</a>{' '}
·{' '}
<a
className="privacy-link"
href="https://this-is-almost-certainly-unvisited.example/abc123"
>
unvisited.example
</a>
</Section>
<Section title="Spellcheck markers (no red squigglies in canvas)">
<input
type="text"
defaultValue="Speeling errrors should not show in canvas"
spellCheck
style={{width: 460, padding: 6, fontSize: 14}}
/>
</Section>
<Section title="System color (color: ButtonText)">
<span style={{color: 'ButtonText', fontWeight: 'bold'}}>
Rendered with the SystemColor keyword
</span>
</Section>
</div>
);
export const HtmlInCanvasPrivacy: React.FC = () => {
return (
<AbsoluteFill
style={{
backgroundColor: '#eee',
padding: 30,
flexDirection: 'row',
gap: 24,
fontFamily: 'sans-serif',
color: 'black',
}}
>
<div style={{flex: 1}}>
<h2 style={{margin: '0 0 12px 0'}}>Live DOM (reference)</h2>
<SensitiveContent />
</div>
<div style={{flex: 1}}>
<h2 style={{margin: '0 0 12px 0'}}>HtmlInCanvas paint (redacted)</h2>
<HtmlInCanvas
width={600}
height={760}
style={{border: '1px dashed #888'}}
onPaint={({canvas, element, elementImage}) => {
const ctx = canvas.getContext('2d');
if (!ctx) {
throw new Error('Failed to acquire 2D context');
}
ctx.reset();
ctx.drawElementImage(elementImage, 0, 0);
// Push the live DOM off-screen so the canvas pixels (which
// have the redactions applied) are visible. CSS transforms
// on canvas children don't re-trigger the `paint` event, so
// this is safe to do inside the handler.
element.style.transform = 'translate(-99999px, 0)';
}}
>
<SensitiveContent />
</HtmlInCanvas>
</div>
</AbsoluteFill>
);
};