iArt.ai
AI motion graphics software — the After Effects alternative anyone can use. Create animations and explainer videos from a text prompt in minutes.
Original Developer: Ducky/Sight8218 AI Assistance: Developed with structural layout and scripting assistance from Artificial Intelligence.
MediaTypeScript · JavaScriptGitHub @Sight8218OnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
Loading the full README. The saved excerpt is shown below.
OPNduck is a local-first, open-source media processing suite. It brings powerful AI
upscaling, motion frame-interpolation, media downloads, conversions, and transcription
directly to your local GPU/CPU at zero cost — no subscriptions, no cloud, no spying. All
your media stays on your machine.Read the full README ↗OPNduck is a local-first, open-source media processing suite. It brings powerful AI upscaling, motion frame-interpolation, media downloads, conversions, and transcription directly to your local GPU/CPU at zero cost — no subscriptions, no cloud, no spying. All your media stays on your machine.
This is v0.4.0-Beta, the first Beta. It ships a much more capable Downloader (8K, subtitles,
dubbed audio tracks, whole channels and lists, sign-in for private content, Smart mode),
Subscriptions that auto-download new uploads while OPNduck is open, an AI Audio card
(Demucs: remove vocals, extract vocals, isolate a voice, echo and noise cleanup), Subtitles
(faster-whisper), Library management (search, filters, sorting, batch groups, JSON export/import),
a noticeably faster UI, and job queues, on top of the multi-format Converter, AI upscalers, frame
interpolation, theming, settings and the adaptive card dashboard. Beat Finder is still not wired in.
The Tauri migration is postponed (see Desktop Shell below); OPNduck ships on Electron.
AI Audio needs a Python venv with torch, demucs and soundfile; that install is large
(about 5.5 GB including the CUDA libraries).
OPNduck uses a flat, solid look: no blur, no translucency. (Earlier builds used a "Liquid Glass" style; that theme was flattened and renamed Blarange in v0.3.0.)
Switching happens in Settings → Appearance:
| Mode | Description |
|---|---|
| Blarange (default) | Flat black + orange, matching the OPNduck website's palette. |
| Monochrome | Strict high-contrast black & white for maximum readability and minimal distraction. |
| Synced with Omarchy | Only shown on an Omarchy desktop; follows its live theme colors. |
Themes are implemented as CSS custom-property tokens switched by a single
data-theme attribute on <html>. Components never change, only tokens do.
tauri-migration-phase1 branch that isn't published on GitHub), but is postponed, not shipping: on Linux with
an NVIDIA proprietary driver under Wayland, Tauri's webview (WebKitGTK) can't do
hardware-accelerated compositing — it either segfaults, or falls back to a
software-rendering path so slow the app is barely usable (confirmed against the
identical build running fine in a normal browser, so this is a WebKitGTK/NVIDIA/Wayland
platform bug, not an OPNduck bug). This is a known, long-standing upstream issue, not
something fixable from application code. Revisit once it's resolved upstream, or if
packaging specifically for Windows (WebView2, unaffected) or non-NVIDIA Linux.faster-whisper installed; set OPNDUCK_PYTHON to use a venv's interpreter).Wired in now (as system binaries on the PATH):
yt-dlp, FFmpeg, LibreOffice, Real-ESRGAN and RIFE, plus faster-whisper through your own Python. The Python sidecar and librosa are not wired in yet.
.SRT and .VTT tracks.npm install
npm run dev # dev server → http://localhost:1420
npm run build # type-check + production build
npm run preview # preview the production build
npm run lint # oxlint
The frontend runs in any browser today — it is the exact bundle the desktop shell hosts. The Electron wrapper, downloader, converter, upscalers and interpolation are wired in, as is Subtitles (faster-whisper); Beat Finder lands in a later update.
v0.4.0-Beta (most engines wired in; Beat Finder still stubbed — expect bugs!)This project is made completely by AI/vibe-coding with human input and testing. If you don't like that, sorry — I'm a single developer and can't afford to build all of this alone. If you want to contribute and work on this project alongside me, you're welcome, but don't expect to be paid anything.
Found a bug? Got a feature idea? Submit it — I'll review whether it belongs in OPNduck core or should be a community plugin. EXPECT BUGS. 🦆