On-device AI · React SDK

The image & video editor that lives in your React app

A drop-in <CreativeEditor /> with on-device AI — captions, voiceover, background removal — running entirely in the browser. No backend, private by default.

Open the live demoRead the docs
$ npm i @lumen/react
Works withReactNext.jsViteRemix
Lumen
ImageVideo
On-device
1280 × 720Export
Media
Text
Shapes
AI
Background removed
Made in your browser
Properties Design
DesignAnimateAI
Brightness+12
Contrast+4
Video
Voiceover
Captions

Private by default

Every pixel and frame is processed in the browser. Nothing uploads — no servers to run, no user media to secure.

No per-use fees

AI runs on-device, so there's no metered inference bill — ever. One flat annual licence, not per-minute or per-credit pricing.

Live in minutes

One typed React component. npm install, render <CreativeEditor />, ship in an afternoon.

See it run — right here

The real editor, in your browser

Launch it and the actual <CreativeEditor /> runs right in this tab — image and video, captions and voiceover, all on-device. Nothing uploads.

app.lumeneditor.com

It loads on click to keep this page fast. Prefer full screen? Open the editor in a new tab.

On-device AI

The AI everyone meters — running in the browser

Captions, voiceover and background removal are usually cloud add-ons billed per credit. In Lumen they run on-device — private, and with no per-use fee.

Web Worker

Auto-captions

Whisper transcribes the clip on-device — 10 visual styles, 4 pacing modes, and a timecoded transcript editor.

Captions, on‑device
Kokoro

AI voiceover

Natural text-to-speech, 18 US/UK voices.

MODNet

Background removal

Image and video, one click.

And more, all on-device

Each model runs in its own Web Worker and is cached after first load. No server. No per-use fee.

Translate captionsDub to EnglishDenoise audioStabilise shake
Easy to configure

Yours by a single prop

Theme it to your brand, flip tools on and off, lock fonts and colours with a Brand Kit, white-label every string, and bind text to your data with template slots. The whole editor reshapes from one config object — no fork required.

Explore the config API
<CreativeEditor
  theme={{ preset: 'iris', accent: '#574BE0' }}
  tools={{ video: true, audio: true, camera: true }}
  brandKit={{ allowedFonts, colorPalette }}
  labels={{ export: 'Save to project' }}
  templateVariables={{ title: post.title }}
  onReady={(api) => api.exportImage()}
/>
themetoolsbrandKitlabelstemplateVariablesonChangeonExport
Everything in the box

One component. The whole studio.

The headline AI gets the spotlight above. Here's the rest — image, video, audio, text, export and the SDK surface — all driven by the same drop-in component.

  1. ImageKonva

    Interactive sub-region crop

    Drag any region, or snap to 1:1 · 4:5 · 9:16 · 16:9.

    • Brightness · contrast
    • Saturation
    • Blur · greyscale
    • Flip
    • LUT presets
    • Fit to artboard
  2. Video & timelineWebCodecs

    Multi-track timeline

    Trim, split and per-clip speed 0.25–4×, composited into one export.

    • Transitions
    • Multi-clip compose
    • Camera record
    • Broad formats
    • Multi-track
  3. On-device AIWeb Workers

    Captions, voiceover & BG removal

    Whisper, Kokoro and MODNet — every model runs in-browser.

    • Translate
    • Dub (EN)
    • Denoise
    • Stabilise
  4. AudioWeb Audio

    Waveform audio tracks

    Trim, fades and volume boost, mixed into the export.

    • Mic record
    • Trim · split
    • Multi-track mix
  5. TextFonts

    Auto-hugging boxed captions

    Padding and rounding derived from the font size.

    • Free font catalogue
    • Weight · italic
    • Outline / stroke
    • Tracking
    • Text styles
  6. ElementsVector

    Shapes, icons & arrows

    Plus a timeline-synced progress bar that redraws each frame.

    • Star · polygon
    • Tables / grids
    • Vector icons
    • Alpha picker
  7. Canvas & presetsPresets

    Social presets & safe zones

    Snapping, guides, align and 80-step undo history.

    • Custom sizes
    • Snapping + guides
    • Align
    • History panel
  8. Export & SDKMediabunny

    GPL-free MP4 / WebM export

    Hardware H.264 via WebCodecs — no FFmpeg, no x264.

    • Multi-track audio mix
    • Brand Kit · white-label
    • Theming · flags
    • Save / open + autosave
    • Touch shell
See the full feature table
CategoryCapabilityNote
ImageTrue interactive sub-region cropPlus fixed-ratio presets (1:1, 4:5, 9:16, 16:9, 4:3)
ImageAesthetic LUT presetsCinematic, Vintage, B&W Noir — one click, image & video
TextAuto-hugging boxed captionPadding/rounding derived from font size
ElementsTimeline-synced progress barFill recomputed every frame on playback
Content libraryRoyalty-free stickers, backgrounds, emoji, SFXCleared for commercial redistribution; offline-first
VideoPer-clip speed 0.25–4×Pitch-preserving (WSOLA time-stretch)
VideoMulti-clip / multi-video compositionComposited into the export
AIAuto-captionsWhisper, 10 styles + 4 pacing modes + transcript editor
AIVoiceover (TTS)Kokoro, 18 US/UK voices
AIBackground removalImage and video (MODNet)
AnimationPresets + keyframe trackEntrance/exit/emphasis on every layer; baked into export
ExportMP4/WebM via WebCodecsHardware H.264; GPL-free (no FFmpeg/x264)
SDKBrand Kit, white-label labels, template slotsRestrict fonts/colours, rename strings, bind to host data
SDKProject save/open + crash-recovery autosaveNamed projects + portable .lumen files
Coming soon

A hosted app for creators

The same editor, as a private, no-upload web app. Unlimited on-device AI, no credit meters. Join the list for early access.

Add a full editor to your app this afternoon

One React component. Image + video + on-device AI. Private by default, no per-use fees, a flat licence.

Open the live demoContact sales