Point it at the SD card and walk away. Every frame is scored for focus, exposure, noise and composition, near-identical bursts are grouped, and your picks land in Lightroom or Bridge as stars and colour labels — from Stock Keyworder, the AI keywording tool photographers already run on their shoots.
The lens decides what counts as a keeper. Change it per batch — the same images graded under a wedding lens and a stock lens give you different picks, on purpose.
Eyes open, real emotion, decisive moments
Speakers, reactions, coverage variety
Flattering pose and expression, eye sharpness
Peak action, sharp subject, storytelling
Light, depth, level horizon, no people penalties
Sharp animal eye, behaviour, clean background
Sense of place, story frames, mixed subjects
Straight verticals, clean lines, window detail
Isolation, even light, edge-to-edge sharpness
Agency triage, marketability, saturation checks
Newsworthy, unstaged, context over polish
Mixed first-pass triage
The card goes in, the shoot comes out graded and tagged. Nothing to install, and no separate export step to remember.
Pick it in the browser; the whole scan runs in managed chunks
Matched by filename stem so you never grade the same frame twice
Reconnect a card mid-scan and carry on where you stopped
One .xmp per frame with stars, colour label, title, description, keywords
JPEG, PNG and TIFF get the metadata written into the file itself
Shipped in the export zip to push metadata into RAW originals
Most culling tools score frames. The differences that matter show up on your second shoot, not your first.
Every reject explains itself instead of showing a number
Grouping that never merges two different moments
Bokeh in a portrait isn't read as blur
You choose the stars and colour for each tier, once
XMP sidecars Lightroom, Bridge, darktable and digiKam read, plus Capture One with sidecar XMP on
An old laptop grades a wedding as fast as a workstation
Full-resolution originals and RAW never leave your device
Credits instead of a subscription you keep paying between shoots
Same engine, different rules per shoot. Start with the page closest to your work.
Blinks, soft frames and near-identical bursts shortlisted
Peak-action keeper surfaced per burst sequence
Brackets and repeat setups clustered and scored
One grading standard across every shooter
Stars and labels into Lightroom, Bridge and Capture One
Keywording, QC and agency CSV before you submit
Cull is the fast first pass over a whole shoot; the Quality Checker is the deeper delivery gate over what survived. Most photographers run them in that order, after keywording.
Titles, descriptions and keywords written straight to XMP
Score, group and cull a full shoot by genre
Verdict, critique and integrity checks on your keepers
Software that grades every frame of a shoot automatically — sharpness, exposure, noise and composition — then groups near-identical frames so you accept one keeper per burst instead of comparing them by hand. A 5,000-frame shoot becomes your strongest few hundred in minutes.
Yes. Point the browser at the card or the shoot folder and the whole thing is scanned — subfolder scanning is optional, RAW+JPG pairs are matched by filename stem, and huge cards run in managed chunks you can pause, stop, reconnect and resume. Only a 1280px preview is uploaded; the originals stay on the card.
No. You pick a lens per batch — wedding, event, portrait, sport, landscape, wildlife, travel, architecture, product, stock, editorial or general — and the grading changes with it. A wedding lens cares about open eyes and real expressions; a wildlife lens cares about a sharp animal eye and a clean background. The stock rules only apply when you choose a stock or editorial lens.
Yes. Near-identical frames are clustered with a 128-bit perceptual hash plus colour, chroma and edge-orientation maps, and EXIF timestamps pair rapid-fire bursts so two different scenes never end up in one group. Each cluster gets a suggested keeper you can override.
Yes. Folder scans run in managed chunks so a multi-thousand-frame shoot uploads and grades continuously, and you can pause, stop and resume without redoing graded frames.
Yes. Every tier maps to a star rating and a colour label you set once in Settings, written into XMP sidecars — so Lightroom Classic and Bridge read your picks as-is, Capture One does once "Prefer Sidecar XMP" is enabled, and darktable and digiKam get colour labels in their own fields too. JPEG, PNG and TIFF also get the fields embedded directly, and the export zip ships an ExifTool script for burning metadata into RAW originals.
No, and a tool that pretends it does will lose you frames you wanted. Treat it as a first pass that removes the obvious misses and organises the rest. Every pick is overridable and every reject explains itself, so you can disagree in one click.
Desktop apps use your own CPU and GPU and work offline, but need installing, updating and licensing per machine. This runs in the browser on any machine you sit down at, including a borrowed laptop on location — RAW previews are extracted locally, so full-resolution originals never upload.
Run one real shoot, not a demo set. Check three things: does it group your bursts without mixing different scenes, does it explain its rejects, and can you override a pick and have the change reach your editor. Then check the pricing model — per-image credits suit occasional shoots, subscriptions suit constant volume.
Very much. A shallow depth-of-field portrait is a soft-focus reject to a naive sharpness gate, and market-saturation scoring is meaningless on a wedding. The grading lens changes per shoot so the rules match what you shot.
Yes. Lenient, Standard, Strict and Ruthless presets cover most workflows, and Custom mode exposes focus tolerance, noise tolerance and how many near-duplicates to keep per cluster. Individual checks — market saturation, composition, anatomy, text, face integrity, megapixels — can each be switched off.
Yes — Canon CR2/CR3, Nikon NEF, Sony ARW, Adobe DNG, Fujifilm RAF, Olympus ORF and Hasselblad 3FR. The embedded preview is extracted in your browser; only a 1280px JPEG (~150–300 KB) is uploaded.
Yes. Cull-only uploads skip keywording entirely; you can run keywording later on any batch from the Vault if you want it.
Culling is 1 credit per image on any plan, with a free monthly allowance. Re-runs cost 1 credit per image; cached passes don't. There is no per-seat desktop licence.