VIDSCRIPT
BY KICKSICK
Turn any video into a script, scenes and AI prompts.
Upload. Analyze. Recreate. · FREE / PRIVATE / NO API KEY
Works with direct video links (.mp4 / .webm). YouTube, Facebook, TikTok etc. links can't be fetched by a browser tool — download the video and use Upload.
Drag & drop a video here
MP4, WebM, MOV, M4V (whatever your browser can decode)
🔒 Your video stays on your device. Processing happens locally in your browser whenever supported.
Analysis Settings
First use downloads open-source model files (from Hugging Face via jsDelivr CDN) into your browser cache. Only model files are downloaded — your video and audio are never uploaded anywhere.
Blueprint Report
No report yet.
Select a video and click Analyze Video. Your report appears here with 22 sections, copy buttons and local exports.
Features
Local speech-to-text
Whisper (Tiny/Base) runs in a Web Worker via Transformers.js with timestamps, language selection and auto-detect.
Frame sampling & scene cuts
Canvas frame sampling with color-histogram scene-change detection, thumbnails and timestamps.
Measured visuals
Brightness, contrast, saturation, color palette, color temperature and camera-motion estimates computed from pixels.
22-section blueprint
Summary, transcript, clean script, hook, structure, scenes, cinematography, prompts, editing guide, checklist and workflow.
AI prompts
Per-scene image and video generation prompts built from measured data plus your own editable scene descriptions.
Local exports
TXT, CSV (Excel/Sheets-safe), SRT, VTT and print — generated in your browser with Blob URLs.
Privacy
- No video upload. This site has no upload endpoint, no database and no server-side processing. Your file is opened with the browser File API and played from a local
blob:URL. - Model downloads only. When you enable transcription or captions, open-source model files are downloaded from Hugging Face (and the Transformers.js library from jsDelivr) into your browser cache. Nothing about your video is sent to them.
- No trackers, no analytics, no login, no API key.
- Nothing persisted. Videos, frames, audio, transcripts and reports are kept only in page memory — never in localStorage, IndexedDB or Cache Storage. (Your browser may cache the downloaded model files so they need not be re-downloaded.)
- Clear Session stops playback, revokes object URLs, removes the video source, discards frames, audio, transcript and report, and shuts down the AI worker to free memory. Closing the tab also discards everything.
- Honest output. The report labels which details were measured, transcribed, AI-captioned or merely inferred. Without the optional caption model, no AI "sees" your frames — visual descriptions come from pixel statistics and your own edits.