Everything clever in LightVision is a proven idea from machine learning — most of it running privately on your own machine. Here's what's actually going on, in plain English.
Every feature here is a well-understood idea — pointed at the one thing rivals can't copy: you.
Culling, grading and search run on your machine — privacy by design, not a promise.
Every call breaks down into reasons you can see, never one mystery verdict.
Edits are sliders on a non-destructive stack — anything the AI does, you can undo.
Seeded from your own keepers and edits, so it converges on your eye over time.
First it has to understand what it is looking at — quality, focus, content and who is in the frame.
Every frame gets seven readable scores — not one mystery number.
Instead of a single thumbs-up, LightVision scores each photo on clarity, focus, open eyes, emotion, composition, moment and exposure, then sorts it into Pick, Alternate or Review. You always see why a frame made the cut — and you can overrule it.
It catches the shots where the eye just missed focus.
◉ Sharp on the eye background falls soft A photo can look crisp in a thumbnail yet be soft on the eye — the one place that has to be sharp. LightVision measures exactly where the focus landed and flags the near-misses, so you never keep the wrong frame.
Search your shoot the way you'd describe it.
Type 'first kiss' or 'confetti toss' and the matching frames surface instantly — no tags, no filenames, no endless scrolling. It understands what a photo is about, not just what it's called.
The same person, gathered from every corner of the shoot.
LightVision recognises each face and groups every photo of the same person — shot hours apart, in totally different light. Duplicates of the bride collapse to one decision, while two different guests never get merged.
The part rivals can't copy: it stops judging photos in general and starts learning how you judge them.
Every keep quietly teaches it what you would choose.
You never write rules — you just cull. From the frames you keep over the ones you cut, LightVision learns your eye and floats the photos you'd pick to the top of the next shoot.
A first-pass edit in your style — tuned to each photo.
A preset slaps the same edit on everything. LightVision learns how you edit and predicts the right adjustment for each frame, so a bright outdoor shot and a candle-lit one each get the proper touch in your voice.
Already shaped like you, from day one.
Point LightVision at galleries you've already delivered and it reads the look you've spent years refining — shrugging off the odd dramatic edit and learning your typical style per scene. Your very first import already sounds like you.
Now it changes pixels — exactly, locally, and always reversibly.
Name what you want changed — and it's selected.
⌕ "the background" — selected Type the thing you want to adjust — the background, the sky, a dress, a face — and LightVision finds it and draws a clean, soft-edged selection for you. Local edits that used to mean minutes of careful masking now take seconds.
Plain words become real, adjustable sliders.
Say 'warm it up and lift the shadows' and LightVision moves the actual sliders — the same ones you'd drag by hand. Nothing is baked in: tweak it, stack on more, or undo it entirely.
Cleaner portraits that never become a stranger.
Restored Original Restoration sharpens eyes, clears noise and tidies skin while staying locked to the real person — so your subject looks like themselves, only better. No plastic, generic 'AI face.' Drag the slider to compare.
All of it rests on one choice: the intelligence runs where your photos already live.
Private by design — and still fast.
Culling, grading, style learning and search all run on your own computer. Your shoot never has to leave your machine — privacy isn't a promise here, it's how it's built — and it's quick enough for a real deadline.
LightVision is in beta — join free, no card required, and we'll help you get set up.