惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

WordPress大学
WordPress大学
A
About on SuperTechFans
量子位
B
Blog RSS Feed
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
博客园_首页
MongoDB | Blog
MongoDB | Blog
小众软件
小众软件
Blog — PlanetScale
Blog — PlanetScale
Microsoft Azure Blog
Microsoft Azure Blog
V
V2EX
Google DeepMind News
Google DeepMind News
Cyber Security Advisories - MS-ISAC
Cyber Security Advisories - MS-ISAC
H
Hackread – Cybersecurity News, Data Breaches, AI and More
G
Google Developers Blog
U
Unit 42
D
DataBreaches.Net
博客园 - Franky
D
Docker
宝玉的分享
宝玉的分享
Y
Y Combinator Blog
月光博客
月光博客
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
Hugging Face - Blog
Hugging Face - Blog

Hacker News - Newest: "AI"

AI can't read an investor deck AI as an attorney? Student uses ChatGPT, Gemini to sue UW over alleged racial discrimination Hacking MCP Servers in AI Systems – The Rug Pull: Tool Changes After Approval GitHub - GenAI-Gurus/awesome-eu-ai-act: Curated tools, official sources, OSS, templates, and guides for EU AI Act compliance. Can AI judge journalism? A Thiel-backed startup says yes, even if it risks chilling whistleblowers Coming soon: 10 Things That Matter in AI Right Now DARPA built an AI to fact-check enemy weapons claims What explains heterogeneity in AI adoption? When AI Meets Muscle: Context-Aware Electrical Stimulation Promises a New Way to Guide Human Movements - Department of Computer Science AI Changed How We Build. It Did Not Change What Matters. Linux rules on using AI-generated code - Copilot is OK, but humans must take 'full responsibility for the… Meta spins up AI version of Mark Zuckerberg to engage with employees Code Mode: Let Your AI Write Programs, Not Just Call Tools | TanStack Blog GitHub - Delavalom/graft: Go framework for building AI agents. Type-safe tools, multi-provider (OpenAI, Anthropic, Gemini, Bedrock), zero vendor SDKs. India's TCS tops estimates, says new AI models did not dent services demand Gen Z's fading AI hype Strong feeling: we are in a folded AI reality GitHub - machinarii/total-recall-catalog: A reference catalog of latest knowledge retrieval, memory & RAG systems GitHub - mensfeld/code-on-incus: Give each AI agent its own isolated machine with root, Docker, and systemd. Active defense detects and stops threats automatically.. Quantization, LoRA, and the 8% Problem: Benchmarking Local LLMs for Production AI Iran war: We spoke to the man making Lego-style AI videos that experts say are powerful propaganda Powell, Bessent discussed Anthropic's Mythos AI cyber threat with major U.S. banks GitHub - immartian/bellamem: Persistent belief-graph memory for AI agents. Retrieves decisive context by importance — not recency, not RAG, not /compact. recursive-mode: The Repo-Native Operating System for AI Engineering After the attack on Sam Altman's home, will AI CEO's go on the offensive? The biggest advance in AI since the LLM Opus 4.6 vs GPT 5.4 One Prompt Unity World Generation Test “AI polls” are fake polls Client Challenge Can AI be a 'child of God'? Inside Anthropic's meeting with Christian leaders
GitHub - MeepCastana/KubeezCut: Free Web based video editor
2026-04-16 · via Hacker News - Newest: "AI"

KubeezCut Logo

Professional video editing — entirely in your browser

No installation. No uploads. No subscriptions. Everything runs locally using WebGPU, WebCodecs, and the File System Access API.

License: MIT TypeScript React WebGPU Vite


KubeezCut — timeline editing


What is KubeezCut?

KubeezCut is a fully client-side, multi-track video editor that runs without a backend. Your media never leaves your machine — everything from decoding to rendering to export happens locally inside the browser tab, powered by modern web platform APIs.

  • GPU-accelerated — effects, compositing, and transitions run as WebGPU compute shaders
  • Frame-accurate — custom Clock engine with WebCodecs-based export
  • Non-destructive — full undo/redo, keyframe animation, and project bundling
  • Private by design — no server, no cloud, no account required

Media generationKubeez is the media generation platform for creating AI-generated assets you can bring into KubeezCut (see Generate media in Screenshots).


Screenshots

Landing images live in public/assets/landing/:

File What's shown
timeline.png Multi-track timeline (video + audio, tools, waveforms)
keyframe.png Keyframe editor — graph / sheet / split view
projects.png Video settings — resolution presets (YouTube, Shorts, etc.) and custom canvas
ai-generation.png Kubeez (Generate media) — model catalog and prompt panel
local-tts.png AI sidebar — Text to Speech (local WebGPU, Kitten TTS)

Export — Encoding runs in the browser (WebCodecs: MP4, WebM, MOV, MKV; see Export under Features). Output size matches your project canvas; the Video settings screenshot above is where preset resolutions and custom width/height are chosen.


Features

Timeline & Editing

  • Multi-track timeline — video, audio, text, image, and shape tracks
  • Track groups with mute / visible / locked gate propagation
  • Trim, split, join, ripple delete, and rate stretch tools
  • Rolling edit, ripple edit, slip, and slide tools
  • Per-track Close Gaps — removes dead space between clips
  • Filmstrip thumbnails and audio waveform visualization
  • Pre-compositions (nested sub-comps, 1 level deep)
  • Timeline markers for organizing long edits
  • Source monitor with mark in/out via playhead or skimmer, plus insert/overwrite edits
  • Configurable undo/redo history depth

GPU Effects

All visual effects run as WebGPU shaders — zero CPU overhead during preview.

Category Effects
Blur Gaussian, box, motion, radial, zoom
Color Brightness, contrast, exposure, hue shift, saturation, vibrance, temperature/tint, levels, curves, color wheels, grayscale, sepia, invert
Distortion Pixelate, RGB split, twirl, wave, bulge/pinch, kaleidoscope, mirror, fluted glass
Stylize Vignette, film grain, sharpen, posterize, glow, edge detect, scanlines, color glitch
Keying Chroma key (green/blue screen) with tolerance, softness, and spill suppression

Blend Modes

25 GPU-accelerated blend modes — normal, darken, multiply, color burn, linear burn, lighten, screen, color dodge, linear dodge, overlay, soft light, hard light, vivid light, linear light, pin light, hard mix, difference, exclusion, subtract, divide, hue, saturation, color, luminosity.

Transitions

  • All-GPU pipeline — dissolve, sparkles, glitch, light leak, pixelate, chromatic aberration, radial blur
  • Standard transitions — fade, wipe, slide, 3D flip, clock wipe, iris (with directional variants)
  • Adjustable duration and alignment per transition

Keyframe Animation

  • Bezier curve editor with preset easing functions
  • Easing presets: linear, ease-in/out, cubic-bezier, spring
  • Auto-keyframe mode for hands-free recording
  • Graph editor, dopesheet, and split view

Export

In-browser export via WebCodecs — no server required.

Options
Containers MP4, WebM, MOV, MKV
Video codecs H.264, H.265, VP8, VP9, AV1
Audio formats MP3, AAC, WAV (PCM)
Quality presets Low (2 Mbps) → Ultra (20 Mbps)

Media Import

Format Supported
Video MP4, WebM, MOV, MKV
Audio MP3, WAV, AAC, OGG, Opus
Image JPG, PNG, GIF (animated), WebP
Max file size 5 GB
  • Files are referenced via the File System Access API — never copied
  • OPFS proxy generation for smooth preview on high-bitrate sources
  • Media relinking for moved or missing files
  • Scene detection and optical flow analysis

Transcription

  • Browser-based speech-to-text via Whisper (runs in a Web Worker, fully local)
  • Models: Tiny, Base, Small, Large v3 Turbo
  • Auto-generate caption text items from transcripts
  • Multi-language support

More

  • Native SVG shapes — rectangle, circle, triangle, ellipse, star, polygon, heart
  • Text overlays with custom fonts, colors, and positioning
  • GPU scopes — waveform, vectorscope, histogram
  • Transform gizmo — drag, resize, rotate directly in the preview
  • Layer masks with keyframeable geometry transforms
  • Project bundles — export/import projects as ZIP files
  • IndexedDB content-addressable storage with auto-save
  • Customizable keyboard shortcuts with preset import/export

Quick Start

Prerequisites: Node.js 18+ and Chrome or Edge 113+

git clone https://github.com/MeepCastana/KubeezCut.git
cd KubeezCut
npm install
npm run dev

Open http://localhost:5173 in Chrome.

Workflow

  1. Create a project from the projects page
  2. Import media by dragging files into the media library
  3. Build your edit — drag clips to the timeline, trim, arrange, add effects and transitions
  4. Animate with the keyframe editor
  5. Preview in real time with GPU compositing
  6. Export directly from the browser — no upload, no wait

Browser Support

Chrome or Edge 113+ is required. KubeezCut relies on WebGPU, WebCodecs, OPFS, and the File System Access API — all of which need a modern Chromium browser.

Brave users — Brave disables the File System Access API by default. Navigate to brave://flags/#file-system-access-api, set it to Enabled, and relaunch.


Keyboard Shortcuts

Action Shortcut
Play / Pause Space
Previous / Next frame /
Previous / Next snap point /
Go to start / end Home / End
Split at playhead Ctrl+K
Split at cursor Shift+C
Join clips Shift+J
Delete selected Delete
Ripple delete Ctrl+Delete
Freeze frame Shift+F
Nudge item Shift+Arrow / Ctrl+Shift+Arrow
Undo / Redo Ctrl+Z / Ctrl+Shift+Z
Copy / Cut / Paste Ctrl+C / Ctrl+X / Ctrl+V
Selection tool V
Razor tool C
Rate stretch R
Rolling edit N
Ripple edit B
Slip Y
Slide U
Toggle snap S
Add / Remove marker M / Shift+M
Add keyframe A
Toggle keyframe editor Ctrl+Shift+A
Mark In / Out I / O
Insert / Overwrite edit , / .
Zoom in / out Ctrl+= / Ctrl+-
Zoom to fit \
Save Ctrl+S
Export Ctrl+Shift+E

See the in-app shortcuts dialog for the complete list.


Tech Stack

Layer Technology
UI React 19, TypeScript 5.9, Tailwind CSS 4, shadcn/ui
Routing TanStack Router (file-based, type-safe)
State Zustand 5 + Zundo (undo/redo)
GPU WebGPU — effects, compositing, transitions, scopes
Decoding Mediabunny (WASM-accelerated media decode)
Encoding WebCodecs — browser-native encode and export
Storage OPFS + IndexedDB via idb
Parallelism Web Workers for export, decode, transcription, scene detection
Build Vite 6 with manual chunk splitting

Development

npm run dev            # Dev server on port 5173
npm run dev:quiet      # Dev server without debug panel
npm run dev:compare    # Run dev + perf preview side-by-side
npm run build          # Production build
npm run lint           # ESLint
npm run test           # Vitest (watch mode)
npm run test:run       # Vitest (single run)
npm run test:coverage  # Coverage report
npm run routes         # Regenerate TanStack Router route tree
npm run verify         # Full CI gate: boundaries + lint + tests + build

Architecture checks (enforced pre-push)

npm run check:boundaries           # Feature isolation boundaries
npm run check:deps-contracts       # Deps contract seam routing
npm run check:legacy-lib-imports   # Block direct @/lib/* usage
npm run check:deps-wrapper-health  # Unused pass-through deps wrappers
npm run check:edge-budgets         # Feature seam coupling budgets

Performance workflow

npm run perf           # Build + serve a production-like local target (port 4173)
npm run dev:compare    # Compare dev (5173) vs prod-like (4173) side-by-side

Use npm run perf to distinguish real regressions from dev-mode noise (React overhead, HMR, debug instrumentation).


Repository Layout

src/
├── app/                     # App bootstrap, providers, debug panel
├── domain/                  # Framework-agnostic domain logic
│   └── timeline/            # Transitions engine, registry, and renderers
├── infrastructure/          # Browser/storage/GPU adapters
├── lib/
│   ├── gpu-effects/         # WebGPU effect pipeline + shader definitions
│   ├── gpu-transitions/     # WebGPU transition pipeline + shaders
│   ├── gpu-compositor/      # WebGPU blend mode compositor
│   ├── gpu-scopes/          # Waveform, vectorscope, histogram renderers
│   ├── masks/               # Mask texture management
│   ├── analysis/            # Optical flow and scene detection
│   ├── thumbnails/          # GPU-accelerated thumbnail renderer
│   ├── fonts/               # Font loading
│   ├── shapes/              # Shape path generators
│   └── migrations/          # Versioned data migration system
├── features/
│   ├── editor/              # Editor shell, toolbar, panels, stores
│   ├── timeline/            # Multi-track timeline, actions, services
│   ├── preview/             # Preview canvas, transform gizmo, GPU scopes
│   ├── player/              # Playback engine (Clock, composition)
│   ├── composition-runtime/ # Rendering sequences, items, audio, transitions
│   ├── export/              # WebCodecs export pipeline (Web Worker)
│   ├── effects/             # GPU effect system and UI panels
│   ├── keyframes/           # Keyframe animation, Bezier editor, easing
│   ├── media-library/       # Media import, metadata, OPFS proxies, transcription
│   ├── project-bundle/      # Project ZIP export/import
│   ├── projects/            # Project management
│   └── settings/            # App settings, keyboard shortcut editor
├── shared/                  # Shared UI, state, and utilities
├── components/ui/           # shadcn/ui base components
├── config/hotkeys.ts        # Keyboard shortcut definitions
├── routes/                  # TanStack Router (file-based, auto-generated)
└── types/                   # Shared TypeScript types

Built on / Inspired by

KubeezCut is a combined fork — built on top of FreeCut as its base, with selected features and UX patterns adapted from OpenCut, plus our own optimizations on top of both.

  • FreeCut — base architecture. Fully client-side, non-linear editing, project bundling, and the foundation for our multi-track timeline, preview engine, and export pipeline.
  • OpenCut — source of the smooth timeline drag-and-drop UX (drag line indicators, track hit-detection, drop zone behavior) and several UI patterns we adapted into our editor.

Our additions on top:

  • Kubeez AI integration — first-class image / video / music / TTS generation powered by the Kubeez API (model picker, variant resolution, credits tracking, auto-routing edit models for reference media).
  • WebGPU effects & transitions pipeline — all visual effects and transitions run as WebGPU compute shaders rather than CSS filters.
  • Performance optimizations — RAF-throttled drag-drop, stale-while-revalidate model caching, generation ID extraction robustness, smart music/dialogue polling (all-songs-ready gating), and GPU pipeline device reuse.
  • Professional workflow additions — keyframe Bezier editor, GPU scopes (waveform / vectorscope / histogram), pre-compositions, project bundle ZIP export/import, frame-accurate WebCodecs export.

If you like KubeezCut, please go star FreeCut and OpenCut too — this project would not exist without their work.


License

MIT — free to use, modify, and distribute.