RAW Cull Processor
RAW Cull Delivery · Local Windows App

An 1,800-frame event folder. Culled before the edit starts.

Private, local-first RAW culling and delivery support for high-volume event work. Your clients' images never leave your computer. No cloud upload. No remote image processing. No source-master edits.

See how it runs Why it is different
RAW Cull Processor running a RAW delivery cull with progress, valid image count, folder controls, and activity log.
Actual desktop build: RAW Delivery, preserve originals, profile control, valid RAW count, elapsed time, bucket totals, and live log.
The before-editing tax

You cannot start editing until you have waded through the pile.

TOO MANY FRAMESA full event runs into the thousands. The keepers are buried in there somewhere.
TOO MANY NEAR-DUPESBurst after burst of the same moment, and you still have to pick the one by hand.
TOO MUCH SORTINGHours disappear into triage before a single adjustment slider gets touched.
DJ working headphones and mixer under a red event tent.
Event folders rarely contain one kind of subject.
What it does

It runs the first pass, so you start editing at the keepers.

Point it at a folder. It scans, groups bursts, scores focus and subject evidence, routes every frame into a clean bucket, and writes a report you can read.

RAW Cull Processor progress panel showing delivered, culled, rejected, pending, elapsed time, and activity log output.
The useful part is visible while it runs: totals, elapsed time, routing decisions, and a readable activity log.

Scan folder

Reads and validates RAW files. Video, DAT, and unrelated files are ignored.

Cluster bursts

Groups near-identical frames by time, sequence, and folder order.

Score focus

Measures sharpness where the subject appears, not blindly on background bokeh.

Route frames

Delivered, Culled, Rejected, Pending. No mystery pile.

Export reports

Structured output, timing, decisions, and transparent run data.

Real folders, not demo-perfect images

Event work is messy, varied, and fast. The processor has to respect that.

A useful cull cannot assume every important frame is a centered portrait. It has to handle action, stage light, group coverage, details, faces, bodies, DJs, objects, and partial compositions.

Stage fitness competitor holding a dynamic strength pose under event lighting.
Action, hard light, motion, skin tone, and selective focus. This is the kind of frame the first pass has to preserve.
Matte-black sports car at a curbside car show with onlookers.
Car show: subject is the car, faces in the crowd are secondary.
Festivalgoer biting a giant spiral potato on a stick at a food vendor stall.
Food festival: tight frame, off-center subject, mixed depth.
Group of Ghostbusters cosplayers posing on a closed street at a community festival.
Group coverage: shared focus, multiple faces, no single hero.
Blue classic Mini Cooper with the hood open, engine bay visible.
Detail: mechanical subject, no face anywhere in the frame.
App proof

A real screen run beats a mockup.

The landing page should show the app behaving like production software: a folder selected, RAW files counted, the run active, and the delivery buckets updating.

Short screen capture of the current Windows build.
The wedge

Your clients' faces never touch a server.

Everything runs on your computer. The folder you point it at is the only place your photos ever live. No image upload. No cloud processing. No API call carrying a client's face to someone else's machine.

Beta reports contain logs, counts, scores, timing, errors, and decisions. Never client images. The work stays yours, start to finish.

Local-first No cloud image processing No upload Masters never modified
Vocalist in a white suit and hat singing into a microphone, the kind of client-recognizable face the processor handles without touching a server.
Private event images stay local. Reports carry decisions, not client photos.
Straight about the build

What the beta copy does today, and what lands during the beta.

This is an honest beta. Here is exactly where the line sits, so nothing on this page is a promise the installer cannot keep.

Live now

  • Folder scan and RAW validation
  • Burst and near-duplicate clustering
  • Focus and exposure scoring
  • Four-bucket routing
  • RAW Delivery and JPEG Delivery
  • Structured output and reports
  • Fully local processing

Landing during beta

  • Deeper face and eye awareness
  • Blink and eyes-open detection
  • Profile-specific judgement per shoot
  • Event subject modes for cars, aircraft, products, and mixed scenes
  • Preference learning from final decisions

Built in the open, shipped to testers as it lands.

Female fitness competitor posing under stage lighting.
Stage detail and profile-specific judgement.
Profiles

Different work, judged differently.

Each profile tunes cull strictness, duplicate pressure, review tolerance, and subject priorities. You set Camera, Light, Profile, Subject, Delivery, and Kelvin right on the run panel.

Camera and Scene Light panel: Camera Auto, Light Manual K, Profile Event Coverage, Subject Auto, Delivery RAW Delivery, Kelvin 5000, with Output, Beta, Reports, Train, and Review buttons.
The run panel: profile, subject mode, delivery path, and Kelvin. Set once per shoot, applied to every frame.
Client

Tight, defensible deliverable. Precision first.

Wedding

Eyes, expressions, and moments matter at the same time.

Stage

The moment in hard light. More forgiving on the technical edge.

Event

Broad coverage where subjects vary from people to objects.

Marketing

Content harvest with more tolerance for partials and story frames.

Brown short-coated dog with tongue out at a street event, illustrating that subject priority is not always a centered face on a human.
Some frames win on expression, and the subject is not always a person.
Control and context

Two delivery paths. One subject that actually matters.

RAW Delivery vs JPEG Delivery

RAW Delivery copies selected native RAW files into a structured set, ready for your editor. Preserve original filenames, or rename on copy.

JPEG Delivery exports finished JPEGs at your chosen size for fast turnaround.

Either way, original masters are never modified. The app works on copies and exports, never the source.

Subject is not always a face

Fitness performer in an inverted stage pose, showing a frame where body position and full-shape composition matter more than a centered face.

A stage frame is not always a face. Sometimes the keeper is a body line, a handstand, a detail crop, a group sequence, or a subject pushed to the edge of the frame.

It has to locate subject priority instead of blindly scoring the center or discarding the frame because the face is not dominant.

Common questions

Honest answers to the questions a working photographer asks first.

Do my client photos ever leave my computer?

No. Every scan, score, and routing decision runs locally on your Windows machine. There is no image upload, no cloud processing, and no API call that carries a client photo to another service. The folder you point the app at is the only place your photos live.

How fast is a typical event folder?

Roughly 1,800 forty-five megapixel RAW files in about an hour on an Intel i7. Speed scales with cores and storage I/O. SSD or NVMe storage helps materially, since the bottleneck is reading and validating thousands of RAWs.

Which RAW formats are supported?

Nikon NEF, Canon CR3, Sony ARW, and other common professional RAW formats. The first pass reads, validates, clusters, and scores. Original masters are never modified, renamed, or rewritten.

How is this different from Lightroom or Photo Mechanic?

Those are editors and ingesters. RAW Cull Processor sits before them. Its job is the first pass: turn an unsorted event folder into a clean set of buckets so editing starts at the keepers. Output is RAW files copied into a structured delivery set, or finished JPEGs, ready for whatever editor you already use.

What hardware do I need to run the beta?

Windows 11 and an Intel i7 or comparable processor. More cores and faster storage shorten run times, but the app does not require a discrete GPU. Everything that matters runs on CPU and local disk.

What does the 30-day beta actually involve?

You install a local Windows app and run it on real folders. Nothing uploads. At the end of the window you send back logs, the report, and an honest assessment. No image sharing required. Wedding photographers willing to test real wedding folders are especially welcome.

Is there a macOS build?

Not yet. A macOS build is on the roadmap once Windows beta feedback is locked. Mention Mac in your beta request notes and it moves up the queue.

Beta access

Looking for a few event and wedding shooters to run it.

  • You run Windows 11 and shoot RAW.
  • A 30-day window. Run it on real folders.
  • You install a local app. Nothing uploads, ever.
  • What comes back: logs, report, and honest assessment.
On a Mac?

A macOS build is on the table when demand is clear. Mention Mac in the notes and it moves up the queue.

Request beta access Human checked by Cloudflare
Tester fit

No images are uploaded through this form. Requests include IP, approximate location, browser, and OS details observed by Cloudflare for beta screening and abuse prevention.