Skip to content
DavidNiessenPublic

About

Advanced media editing and conversion tool for desktop(-first) and web

Resources

Contributing

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

277 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Frame

A cross-platform media editor and converter for images, video and audio. Edit a clip, layer audio, or convert an entire folder—all processed locally on your device.

Try the web app · Download desktop builds · Report a bug

Features

  • Images: a dedicated canvas with zoom, pan, before/after comparison, crop handles, rotation, text and editable brush layers. Export PNG, JPEG, WebP, BMP or TIFF; convert entire image folders.
  • Shapes & guides: editable rectangles, rounded rectangles, ellipses, lines and arrows for images and videos; on-canvas handles, rotation, color and stroke controls. Center, thirds and safe-area guides, with snapping for shapes, text and artwork.
  • Image layers: reorder, rename, hide, lock and group adjustments and frame objects. Nine blend modes, layer opacity, and editable rectangle, ellipse or brush masks with feathering and inversion.
  • Retouch & cutouts: editable clone/healing brushes and local background removal with soft edges and restore/remove refinement. Transparent PNG, WebP and TIFF exports.
  • Camera imports: HEIC/HEIF and common camera RAW formats through portable decoder workers. Original files remain available for recovery; RAW development produces 8-bit sRGB working images.
  • Image metadata: credits, copyright, descriptions and optional camera information in JPEG, PNG, WebP and TIFF. GPS location is excluded by default.
  • Multitrack: independent video/image layers with positioning, fit, opacity, color and curves; linked video/audio trimming, movement and splitting.
  • Animation: position, size, opacity and volume keyframes with linear, smooth and hold interpolation. Local subject tracking creates editable X/Y keyframes, up to 30 seconds per pass.
  • Color curves: RGB and separate red/green/blue tone curves, a frame histogram, draggable points, numerical controls and tonal presets. Also available as timed video effects.
  • Batch image sizes: resize by longest edge, dimensions or percentage; preserve proportions, crop to fill or stretch. Avoid enlargement and save custom size presets.
  • Visual adjustments: exposure, gamma, shadows, highlights, vibrance, hue, tint, vignette, blur, sharpen, black & white, sepia, invert, pixelate, posterize, film grain, sketch and emboss. These also work as timed video effects with live previews.
  • Edit: trim, split, reorder and join clips; insert videos from the library; replace or layer audio; extract audio from video.
  • Effects: crop, rotate, resize, speed, fades, color adjustments, LUTs, volume, EQ, noise reduction and loudness normalization. Timed effects have draggable timeline bars and live previews.
  • Playback: independent monitoring volume, five-second skips that keep playback running, precise frame stepping and an audio-reactive background.
  • Captions: local Whisper speech recognition for videos and audio, with a bundled model and language detection; independent text effects with on-frame editing, positioning and resizing; fonts, colors and styling; SRT import/export and burned-in subtitles.
  • Convert: files or recursive folders, a dedicated batch queue, presets, custom filenames, metadata and artwork, plus advanced encoding options.
  • URL imports: direct files on desktop/web; public YouTube, Instagram Reels, TikTok, SoundCloud, Vimeo, Dailymotion and Bandcamp links on desktop, with video/audio and quality choices.
  • Desktop downloads: an animated Install App button opens a platform picker, with automatic detection and an optional latest prerelease.
  • Updates: installed apps check GitHub at startup and every six hours, download verified updates in the background, and install on restart. Opt into prereleases in Settings; a settings badge highlights available updates.
  • Workspace: undo/redo, portable .frame.json projects, automatic local recovery, resizable panels, keyboard shortcuts and a command palette.
  • Appearance: system light/dark detection, Light, Dark and High contrast themes, responsive layouts and reduced-motion support.

Output formats: PNG, JPEG, WebP, BMP, TIFF, MP4, MOV, MKV, WebM, GIF, MP3, WAV, FLAC, OGG and M4A. AIFF audio inputs support both .aif and .aiff extensions. Input support depends on FFmpeg and the playback capabilities of your platform; desktop proxies help preview formats that the browser cannot play directly.

Desktop and web

Capability Windows / macOS / Linux Web
Editing, effects and conversion Bundled native FFmpeg Self-hosted FFmpeg WebAssembly
Folder conversion Native picker and destination folder Directory picker/drop where supported; ZIP download
Local file-manager actions Windows Explorer, Finder Quick Actions, Dolphin Unavailable
Compact context-menu converter Yes Unavailable
Direct media URLs Yes Requires the media host to allow CORS
Public video / music service URLs Bundled yt-dlp; no backend Unavailable
Installed system fonts Yes Bundled fonts only
Hardware encoding, proxies and watch folders Yes, subject to hardware/platform support Unavailable

Playback uses live EQ/gain, cached previews, lower-resolution visual processing and hardware encoding where available. See preview performance for measurements and limits.

Browser GIF conversion uses separate palette and encoding passes to avoid buffering an entire decoded movie. The browser has memory guards: 500 MB per source, 800 MB per library and 500 MB of retained conversion results. ZIP preparation uses additional memory. Use desktop for large collections.

Whisper Tiny model files are committed in vendor/speech-models/ with integrity checks. Desktop includes the model and works offline immediately. Web serves about 42 MB of model files with Frame on first use and caches them for subsequent offline inference. Audio is never uploaded. Review generated text and timing before export.

Media is processed on your device. URL imports connect to the source you choose. Desktop includes FFmpeg, FFprobe and yt-dlp; users do not need to install those tools separately. Web assets and fonts are served with the application, without third-party CDNs or API keys.

Frame is under active development. Automatic desktop builds are prereleases; maintainers can also publish full releases. Windows installers are currently unsigned. macOS apps are ad-hoc signed to verify bundle integrity, but are not Apple notarized; first launch may require approval in System Settings → Privacy & Security → Open Anyway. The Windows Explorer extension has a separate certificate-consent workflow. See installation, distribution and signing before redistributing builds.

Get started

In Editor, import media, select a source, then drag timeline edges to trim. Split and remove sections, add effects with + or a double-click, and click an effect bar to adjust it. Open Export to choose the output format.

In Convert, add files or a folder, choose the target format and start the batch. Advanced settings include bitrate, metadata, artwork and filename templates. Conversion uses complete original sources independently of your editor drafts.

Press Ctrl/Cmd + K for commands. Space toggles playback outside text fields. See the user guide for projects, audio replacement and file-manager actions.

Development

Use Node.js 24 (recommended), or Node.js 22.13+. npm is the reference package manager and package-lock.json is the canonical lockfile.

npm ci
npm run dev      # Web app
npm run desktop  # Electron app; requires a graphical session

pnpm 10.26+ / 11 / 12 is also supported:

pnpm import
pnpm install --frozen-lockfile
pnpm desktop

Install scripts prepare the local WebAssembly assets and checksum-verified vendored media tools. Electron's runtime is downloaded by its installer. If desktop startup reports an incomplete runtime, run npm run setup:desktop (or pnpm setup:desktop). See contributing for troubleshooting and cloud development for headless environments.

Checks and builds

npm run check            # Prettier, ESLint and TypeScript
npm test                 # Setup, core logic and native media regressions
npx playwright install chromium
npm run test:e2e         # Browser UI and actual WebAssembly processing
npm run test:desktop     # Build and run Electron regressions
npm run build            # Web and Electron bundles
npm run package          # Installer for the current OS and architecture

Speech inference uses the CPU runtime bundled by npm. On Linux, set ONNXRUNTIME_NODE_INSTALL_CUDA=skip when installing dependencies to avoid downloading unused CUDA libraries; CI already sets this.

Use npm run format and npm run lint:fix to apply automatic fixes. CI validates the web app, pnpm setup and native desktop builds on Windows, macOS Intel/Apple Silicon and Linux. Every successful master push publishes all four installers and SHA256SUMS.txt as a uniquely versioned prerelease. For a full release, open Actions → Validate and package → Run workflow, select master, choose release, and enter a version such as 1.0.0. Choose prerelease for a manual preview. Leaving the version blank continues the current release line. After v1.0.0, automatic previews use v1.0.1-build.N; a manually selected newer prerelease version becomes the preview line. Published tags determine versions without committing automated version bumps. Full releases are never overwritten.

Project structure

src/core/        Framework-independent edit model, validation and FFmpeg planners
src/engines/     Shared media-engine API with web and desktop adapters
src/hooks/       Application state and editing lifecycles
src/components/ UI, conversion workspace and settings
src/styles/     Theme tokens and responsive surfaces
electron/      Main process, sandboxed preload and native integrations
native/         Windows Explorer extension sources and verified public assets
scripts/        Setup, builds, tool verification, signing and release preparation
tests/          Core, browser, Electron and setup regressions
vendor/         Pinned media tools with upstream notices and checksums

See architecture, design, and contributing when extending Frame.

Hosting

npm run build produces the web app in dist/. Serve it over HTTPS with the isolation headers in the nginx example. GitHub Pages uses the included isolation service worker; the full-release workflow builds and tests that mode before deploying. GitHub Pages updates only on full releases, so automatic prereleases do not change the public web app. No media-processing backend is needed.

Licensing

Frame's application code does not yet have a chosen open-source license. An explicit license is required before reuse or redistribution of that code is permitted. Bundled tools, fonts and dependencies retain their upstream licenses; see distribution notices and vendored tools.

About

Advanced media editing and conversion tool for desktop(-first) and web

Resources

Contributing

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages