Culling in photography means reviewing a shoot and selecting the images worth editing or delivering. You separate strong photographs from technical failures, duplicates and weaker alternatives before spending time processing them. Culling is selection—not necessarily permanent deletion—and it works best with a simple, repeatable system.
01Why cull before editing?
Every unnecessary edit costs time and adds another decision to the delivery process. A gallery with six near-identical expressions can feel less intentional than one with the strongest frame and a genuinely different alternative.
Imagine skipping 100 redundant images that would each take two minutes to edit: that saves 200 minutes. This is planning arithmetic, not a productivity guarantee.
Good culling also preserves coverage: the strongest individual portraits cannot replace missing photographs of important people or promised products.
02First, protect and organise the originals
Copy the memory cards to your working drive and create a second verified copy on a separate device before deleting anything. Open representative files and check that the copied folders contain the expected images.
Keep cards intact until your backup process is confirmed.
Use a clear folder such as 2026-10-29-client-event. Synchronise camera clocks where necessary so moments appear together.
A Lightroom catalogue backup is not a backup of the original photographs; protect both the image files and the catalogue or selection metadata.
03Pass one: reject only obvious failures
Work chronologically at a fit-to-screen view. Mark accidental exposures, completely unusable focus, blocked views and unneeded test frames as rejects.
Leave uncertain images unflagged instead of spending a minute defending each one.
In Lightroom Classic, P marks a pick, X a reject, and U clears a flag. A reject flag is a label, not immediate deletion.
Avoid rejecting unusual expressions automatically: a fleeting laugh or imperfectly framed reunion may matter more than technical polish.
04Pass two: choose the strongest frame per moment
Compare related frames together. Look first for expression, gesture and storytelling, then confirm critical focus at 100% when needed.
For groups, inspect every face rather than only the person nearest the centre.
Pick the strongest photograph from each repeated setup, retaining an alternative only when it adds something useful. Lightroom Classic’s Compare and Survey views help evaluate candidates together.
Do not begin detailed colour work here: selecting and developing are different jobs, and switching repeatedly makes both harder.
05Pass three: audit the story and apply simple ratings
Filter to picks and review the gallery as a sequence. Check required people, wide views, details and transitions against the brief.
Remove remaining repetition without making the coverage incomplete.
Use flags for delivery selection and reserve stars for another purpose: three stars for portfolio candidates and five for exceptional favourites, for example. Lightroom supports one to five stars.
The meaning is yours; write it down. A pick should not mean “maybe” in one folder and “deliver” in another.
Worked example
06Worked example: 2,000 wedding frames to 600
Here is an illustrative workflow, not a wedding-delivery standard. Pass one removes 500 clear failures and tests, leaving 1,500 candidates.
Pass two removes 750 duplicates and weaker alternatives, leaving 750 picks. The coverage audit removes another 150 redundant frames, producing 600 deliverables.
The final selection rate is 600 ÷ 2,000 = 30%. At an assumed two minutes of editing per image, editing 600 rather than all 2,000 saves 2,800 minutes—46 hours 40 minutes—before culling time.
Real editing speeds and batch-processing savings vary considerably.
07Tools: Lightroom, Photo Mechanic or FastRawViewer?
Choose Lightroom Classic if you already use it for editing: flags, ratings, filters and comparison views keep selection inside your catalogue. Photo Mechanic is a dedicated browser and metadata tool; it supports ingest and fast review using embedded previews rather than replacing a full RAW editor.
FastRawViewer is useful when RAW exposure assessment matters: its RAW histogram and clipping statistics help distinguish recoverable files from genuinely lost detail. Its XMP sidecars can store ratings and other metadata.
Test a small folder before transferring selections between applications; do not assume every flag behaves identically.
08How many photographs should you deliver?
Agree deliverables in the brief rather than chasing a universal keeper percentage. For planning only, a portrait session might promise 15–30 edited images; a headshot job might deliver 2–5 per person.
A product brief might require exactly five specified views per item.
A full-day wedding might be scoped at 400–800 images, but hours, traditions and client expectations change that substantially. A landscape portfolio may retain just 1–5 photographs from an outing.
These are example package sizes, not verified industry averages or reasons to pad a gallery.
09What a first serious cull can look like
Illustrative scenario: a new portrait photographer imports 400 photographs and starts editing the first attractive face. After an hour, several edited frames look almost identical, while the best expression remains unseen farther down the sequence.
They restart with flags, select one or two images from each pose, and finish with 24 varied portraits. The improvement comes from comparing the entire shoot before committing editing time—not from an automatic rule that 24 is always the right number.
10In-camera versus computer culling
Use in-camera review to check focus, exposure and whether a scene needs reshooting. Avoid mass deletion on a tiny screen during an important assignment: subtle expressions and edge details are easier to miss, and deletion interrupts shooting.
Computer culling offers larger previews, side-by-side comparisons, reversible flags and backup options. If storage becomes tight, swap to a spare card where possible rather than rushing irreversible decisions.
Reserve permanent deletion for a deliberate retention policy after backup and any contractual obligations are considered.
11If culling feels slow or unreliable
If every frame becomes a maybe, the cause is usually unclear selection criteria. Write three priorities, such as expression, eye sharpness and varied coverage.
If previews lag, check preview generation, drive performance and software settings before blaming your judgement.
If ratings disappear between applications, inspect metadata and sidecar settings using a small test folder. If RAW files look different from embedded previews, remember that rendering differs.
If automated suggestions reject an emotionally important image, override them: technical scoring cannot know the client’s relationships or your intended story.
FAQ
12FAQ and a 15-minute practice workflow
Should you delete every reject? No. Flag first, then apply a documented retention policy after backups and delivery checks. Should AI make the final choice? Treat suggestions as assistance, especially for expressions, accessibility differences and meaningful imperfect moments.
For practice, choose 100 backed-up images. Spend four minutes rejecting obvious failures, six picking the strongest alternatives, and five auditing variety and checking focus. Export a contact sheet of your picks. Record the count and one reason for each borderline decision; use that evidence to simplify your next cull.
Sources & further reading
These references support the technical explanations. Exercises and starting settings are illustrative and should be adapted to your equipment and conditions.
- Adobe: Flags, labels and ratings in Lightroom Classic
- Adobe: Library module workflow and comparison views
- Camera Bits: Photo Mechanic features
- FastRawViewer: RAW exposure assessment
- FastRawViewer: XMP ratings and metadata
- Adobe: Assisted Culling
Free resources from LensSuite
Was this guide helpful?
