VideoHighlighter Pro
Get a license →
00:00:00:00 — INTRO

Your footage never
leaves your machine.

Find and explain the moments that matter in footage you won't upload — then export a cut. Drop a video, press Analyze: the app scores on your hardware, writes a highlight reel plus separate clips, and shows why each moment was picked. No upload. No subscription.

SIGNAL_TIMELINE LIVE ANALYSIS
signal timeline — review what it found, then cut demo → · Free v0.10.0
00:00:40:00 — HOW IT STARTS

One button if you want it. Every knob if you don't.

You do not have to learn the scoring model to get a cut. Analyze uses built-in defaults. Detailed settings stay one click away and are never rewritten by that run.

01 — DROP

Point it at footage

Drop an mp4, mov, mkv, or avi. Nothing is uploaded. The file stays on the disk you chose.

02 — LENGTH

Pick a reel length

Short (~1–2 min), medium (~4 min), or longer (~7 min). That sets how much of the source makes the highlight.

03 — ANALYZE

Review, then cut

Motion peaks and loudness are scored locally. Open the timeline, read the report, or ask why a moment was picked — then take the highlight reel and the separate clips folder.

00:01:12:00 — WHY OFFLINE

Not a shorts factory. A local cut you can defend.

Cloud highlight tools need a subscription because your footage pays for their GPU. VideoHighlighter runs detection, transcription, and composition on your machine — video, transcript, and models stay on disk. The timeline and report are the product: you see why a moment scored before you export.

01

Buy once, own it

A single license, not a monthly charge. The tool keeps working whether or not you renew anything.

02

Nothing leaves your disk

No upload step. Client footage, unreleased cuts, and anything under NDA never leave the machine you chose. Analysis does not call a cloud model.

03

Runs on real hardware

OpenVINO acceleration on Intel Arc, or any modern GPU. No render farm required.

00:03:40:00 — UNDER THE HOOD

The same local pipeline, six stages, still zero uploads

Simple Analyze is the front door. This is what runs underneath when you want objects, faces, speech, or search — still on your machine, in order.

DETECT

01Object & motion detection

YOLOX object detection running on OpenVINO, combined with optical flow, finds people, movement, and action across every frame — the raw signal everything downstream is built on.

IDENTIFY

02Face recognition & identity avoid

YuNet + SFace (OpenCV Zoo) tracks identity across a cut so the same person stays tagged scene to scene — find every moment someone appears in, or mark a person to avoid so clips containing them stay out of the edit.

TRANSCRIBE

03Speech transcription

Whisper transcribes dialogue and narration locally, giving you a searchable, timestamped script tied directly to the timeline.

COMPOSE

04Composition Engine

Reads spatial relationships between detected subjects to flag actual events — not just motion, but who's doing what, where, relative to whom.

SEARCH

05Semantic search

CLIP-based search lets you find a moment by describing it — "the handshake," "the empty stage" — instead of scrubbing the whole timeline by eye.

EDIT

06Signal & edit timelines

A PySide6 editor separates the raw detection signal from your final cut, so you can review what the pipeline found before deciding what makes the edit.

00:07:20:00 — IN USE

You decide what counts as a highlight

This is not a shorts factory that uploads your file and returns a black-box cut. The pipeline surfaces reviewable signals — teach it what to look for, open the timeline and report, then export a reel and separate clips. Every build ships free and open-source on GitHub first.

45
Releases shipped since the first public build
0
Uploads required to analyze your footage
2
Developers — the people who built it answer your support questions
00:08:30:00 — FREE VS PRO

Free or Pro?

The full analysis pipeline is free and open-source. Pro adds real-time detection, query-by-example search, a commercial license, and priority support.

FeatureFreePro
Scene, motion & audio detection
Object detection & action recognition
Transcription & subtitles (Whisper)
Natural-language visual search
Face recognition & identity avoid
VR side-by-side mode & VR-safe rendering
Fully offline · prebuilt app
Live object & action overlay — during playback
Teach a category by pointing — no training
Search by example — “find more like this”
Open-vocabulary detection — type any object
Counter / scoreboard detection
Commercial license (no AGPL)
Priority support & updates
PriceFree · AGPL$49 · one-time

The free edition is open-source on GitHub. Pro is a proprietary build for commercial use.

00:08:50:00 — PRICING

One license. No renewal.

Your key arrives with the receipt. Activate it once, and from then on it is checked entirely on your machine — no account, no subscription, nothing phoning home.

Studio

Contact us
For teams and multi-seat studio deployments
  • Everything in Individual
  • Multiple seats under one agreement
  • Priority support
  • Volume pricing
Get in touch

Need a detector trained on your own domain — a custom YOLOX/OpenVINO pack, invoiced separately? Email us. Not sold through checkout.

00:09:40:00 — FAQ

Before you download or buy

Short answers. If yours is missing, Discord or email is the same two people who built it.

Does any footage leave my computer?

No. Video files, transcripts, detections, and local models stay on disk. Analysis, search, and export never upload footage. The only network calls Pro makes are a one-time license activation with Lemon Squeezy, and an optional anonymous check for a newer build (a GET of a public JSON file — no key, no machine id).

Windows only?

The current prebuilt app is Windows. macOS is not a supported download yet. You can run from source on other platforms if you know the stack; we do not sell that as a product.

Do I need a GPU?

No. A GPU (including Intel Arc via OpenVINO) makes detection faster. A current CPU still runs the pipeline. There is no cloud render farm and no upload fallback.

How big is the download?

Free is a Windows Setup zip from GitHub Releases. Pro is a single .7z of about 2 GB from your Lemon Squeezy order — extract with 7-Zip or WinRAR, then run VideoHighlighter.exe. Models ship inside the build; you do not fetch them from us at runtime.

What is the difference between Free and Pro?

The analysis pipeline is in both. Pro adds live overlays during playback, teach-by-point categories, search-by-example, open-vocabulary detection, counter/scoreboard detection, a commercial license (no AGPL), and priority support. See the table above. Price is $49, once.

How does the Pro license work?

Your receipt includes a UUID key. Paste it once while online; after that the app checks the license on disk and does not need the internet. Up to 5 computers. Replacing a machine: email and the slot is freed. No account, no subscription.

I need a detector for a specific domain.

That is a services job — custom YOLOX/OpenVINO training, invoiced by bank transfer, not sold through Lemon Squeezy. Email with what you need to find in the footage.

Where do I get help?

Discord or przkreft@gmail.com. After a Pro purchase, the same addresses — there is no ticket portal.

Windows 10/11 · GPU optional · internet once to activate Pro · footage never uploaded

00:10:30:00 — OUTRO

Analyze the next cut without sending it anywhere.

Free and open-source at the core. Pro adds live tools, a commercial license, and support.