Article to Video Studio — Gemini 4 Argon + FFmpeg Web

Session ID: generating...

📄 Step 1: Article Text File(s) or Content

Upload one or more .txt / .md files or paste text below. Multiple files automatically enable multi-article summarization.

📑

Tap or click to add text file(s)

Read directly in your browser (never saved on server)

Article Content

🖼️ Step 2: Upload Scene Images

0 images loaded

Add images (.jpg, .png, .webp, .gif, .bmp) for Gemini 4 Argon vision & Ken Burns animations.

🖼️

Tap or click to add images

Kept in browser memory & IndexedDB

🎛️ Step 3: Video & Gemini 4 Argon Options

Narration & Script Language (83 Languages)

Opening Splash Image (Optional)

Pipeline Execution Target

Multi-Speaker Narration (2 News Anchors) Uses 2 distinct universal Gemini voices taking natural turns Multi-Article Summarization Combines and transitions smoothly across multiple source articles Ken Burns Camera Animations Zoom, pan-to-focal-point, and slide effects guided by Gemini 4 Argon vision AI Quote Text Overlays Extracts key article quotes with animated transitions Burn-in SRT Subtitles Renders synchronized subtitles onto the output MP4 video in FFmpeg Web

Generated Video & Audio Player

No video loaded yet. Generate a video or click "Render MP4 in FFmpeg Web" if a storyboard is ready.

📊 Agent Token Analytics & Pipeline Inspector

Total Tokens

Input / Prompt

Output Tokens

Thinking / API Time

0 / 0s

Agent / Step	Model	Input	Output	Thinking	Total	Time
Run a task to view per-agent token consumption.

Generate a video to inspect the Gemini 4 Argon voiceover script.

Generate a video to inspect the Gemini 4 Argon scene schedule & bounding boxes.

Generate a video to inspect the SRT subtitle cues.

FFmpeg Web (ffmpeg.wasm) output logs will appear here during rendering...

🗂️ Previous Video Generations (Browser Cache)

🔒 Device-Only Credentials (Browser `localStorage`)

Your credentials stay 100% on your device inside your browser cache (localStorage). The Node.js server never stores your keys or files on disk.

Authentication & API Mode

Gemini API Key (Powers gemini-4-argon + Native Gemini Multi-Speaker TTS) *

✨ Architecture & Agent Token Tracking

1. 6 Specialized AI Agents Tracked Per Task

Every video task tracks input, output, and thinking tokens across all 6 agents:

1. Script Writer Agent (gemini-4-argon)
2a. Multi-Speaker TTS Agent (gemini-3.8-flash-tts / gemini-2.5-flash-preview-tts)
2b. Subtitle Alignment Agent (gemini-4-argon multimodal audio)
3a. Vision & Bounding-Box Agent (gemini-4-argon multimodal vision)
3b. Scene Storyboard Director Agent (gemini-4-argon)
3c & 3d. Quote Overlay & Sentiment Music Agents (gemini-4-argon)

2. Portable FFmpeg Web (`@ffmpeg/ffmpeg` WASM)

Final 720p video encoding, Ken Burns motion, and background music mixing run inside your browser via WebAssembly with native toBlobURL and fetchFile support.