Batch Dust Removal for Photo Scanners: The Complete Workflow

Why Batch Is the Whole Point

Removing dust from a single scan is a solved problem. You zoom in, you clone-stamp, you move on. Removing dust from five hundred scans is a different problem, and it is the one almost every scanning workflow gets stuck on.

A moderate scan-back — 10 rolls of 36, or one memory card of 4,000 family slides — is ten to fifty hours of manual spotting using Lightroom or Photoshop. That is the step that kills archival projects and the reason professional scanning labs have long had in-house proprietary batch-cleaning pipelines.

This article walks through how to build the same workflow on a modern laptop using DustMagic as the batch engine — a small, focused desktop tool that sits happily in front of your existing Lightroom or Photoshop workflow rather than trying to replace it. If you’re processing more than twenty images at a time, this is the workflow you want.

Before You Batch: Organise Your Source

The fastest batch-clean is the one with the fewest variables. Before you run a single image, spend ten minutes sorting your source folder.

Separate by film stock. A folder of mixed Portra, Ektar, HP5, and Provia will never run cleanly on a single preset. Each stock has different grain structure, base density, and typical defect size. Put them in different folders.
Separate slides from negatives. They use different detector settings. Do not mix them in one batch.
Separate by scanning method. DSLR-scanned images and flatbed scans have different dust characteristics. If you did both in a session, batch them separately.
Set consistent filenames. DustMagic processes by folder order. If you want your output in a predictable sequence, rename before you batch.
Back up first. DustMagic writes to a cleaned/ subfolder and never modifies originals, but your own process should include a fixed “master” copy you never edit.

A sensible folder structure for a typical archive job:

ClientName_2026-04/
├── 01_masters/                      ← untouched original scans
│   ├── portra_400_roll_01/
│   ├── portra_400_roll_02/
│   ├── xp2_roll_01/
│   └── provia_100_slides/
├── 02_dust_cleaned/                 ← DustMagic output (auto-created)
└── 03_graded/                       ← Lightroom colour work on cleaned files

The Workflow, Start to Finish

Step 1: Scan a Representative Image

Pick a single frame from each folder that represents the typical dust level of that stock and scanning session. This is the image you will use to tune your preset. Choose one that has a mix of sky (or smooth base for negatives), mid-tones, and some fine detail — a landscape with foliage, a portrait with clothing texture.

Step 2: Tune on the Representative Image

Open DustMagic, load the folder, and select the representative image. Then:

Set the correct Film Type (Negatives or Slides).
Start with the Balanced profile.
Switch the view to Detection Overlay. Orange pixels are what will be removed.
Scan the image for two failure modes: missed dust (visible defects without an overlay) and false positives (overlay on eyes, fabric weave, foliage, text).
If you see missed dust, lower Confidence in 0.05 steps until it is caught. If you see false positives, raise Confidence — or switch to Conservative.
Switch to the Cleaned view and check the result at 100 per cent zoom. Look for smooth patches where texture should sit — if you see any, nudge Confidence up a touch so the rebuild is only triggered where it’s really needed.
When the representative image looks clean with no collateral damage, save the settings as a named profile: Portra_DSLR_Balanced_0.45.

Step 3: Spot-Check Two More Images

Before running the full batch, apply the saved profile to two other images — ideally a very clean one and a very dusty one. If the profile handles both, you are good to batch. If the dusty image still has visible dust, re-tune Confidence down a touch and save a second profile (“dustier” variant).

This spot-check saves you from discovering, three hours into a batch of 1,000 images, that the profile was wrong for the second half of the set.

Step 4: Run the Batch

Click Process All Images. DustMagic detects your CPU core count and available RAM, spawns workers in parallel, and processes the batch. Typical speeds on a 6-core modern laptop, scanning 5300×3500 images at half-resolution detection:

10 images: ~20 seconds
100 images: ~3 minutes
500 images: ~15 minutes
1,000 images: ~30 minutes

Walk away. Make coffee. Come back to a cleaned folder.

Step 5: Quality-Control Pass

This is the step most people skip and later regret.

Open the cleaned/ folder in a fast viewer — Lightroom’s Library module in Survey view, Photo Mechanic, or FastRawViewer. Scroll through every image at a medium zoom level. You are looking for three things:

Over-processing: smooth patches where texture or detail should be. If you see this on more than ~5 per cent of images, your Confidence was too low. Re-run that batch with a slightly stricter preset.
Under-processing: visible dust that was not removed. Flag those images for a second pass or manual touch-up.
Rebuild artefacts: unnatural smoothness near detected defects. If you see this on more than a few images, tighten Confidence a notch and rerun, or finish them by hand with the Smart Brush Tool.

DustMagic keeps a detection mask for every image, viewable right inside the app — overlay it, flip to split view, and zoom up to 20× to see exactly what was detected and removed. This is the audit trail that makes the process trustworthy for archival and client work.

Step 6: Manual Touch-Up on Problem Images

For the small number of images that need a little more attention:

Reach for DustMagic’s built-in Smart Brush Tool — click a stubborn speck and it’s gone. No round-trip to another app.
For full-frame scratches, open the cleaned file in Lightroom and run a manual Heal pass with a straight-line mask.
For hero images going to print, the final five per cent is always a human pass at 100 per cent zoom. DustMagic isn’t trying to replace that step — it handles the first 95 per cent in three minutes so your eyes are fresh for the images that deserve them.

Step 7: Hand Off to Colour Grading

Import the cleaned/ folder into Lightroom for grading. Your scans are now dust-free before you have started the creative work, which is the correct order. Grading dusty scans leads to adjustments that look wrong once the dust is removed — contrast bumped to hide specks, shadows crushed to hide fibres.

Preset Library — The Payoff

After three or four batches you will have a library of saved profiles. A sensible naming convention:

STOCK_SCANMETHOD_DPI_PROFILE_CONF
e.g.
Portra400_DSLR_3200_Balanced_0.45
HP5_V850_3200_Conservative_0.60
Provia100_DSLR_3200_Slides_0.50

The next time a client drops off six rolls of Portra shot on a trip, you load the right preset and process the whole job in minutes. This is the compound value of a batch workflow: every job after the first is faster.

Volume Tips for Serious Scan-Backs

Batch size ceiling: somewhere around 300–500 images per batch on a typical 16GB-RAM laptop. Beyond that, RAM starts to matter more than core count. Split into chunks and run sequentially.
Detection resolution: Half (0.5x) is 4× faster than Full and visually identical on everything below print-at-A2. Reserve Full for hero images.
Close other applications. Parallel batch uses most of your CPU. Running Lightroom import alongside a 500-image batch doubles both jobs.
Disk speed matters. Reading and writing 500 TIFF files is I/O-heavy. SSDs outperform spinning disks by roughly 3× on large batches.
Temperature: a sustained 30-minute batch runs your CPU hot. On a laptop, use a stand and keep vents clear. Throttled cores halve throughput.

When Not to Batch

Batch processing is the wrong choice for:

Single hero images headed to print or client delivery — those deserve manual attention at 100 per cent zoom.
Mixed-condition batches where each image needs a different approach. Split into sub-folders by condition and batch each separately.
Heavily damaged images with tears, fading, or chemical stains. Those need Generative Fill or a restoration specialist, not dust removal.
Film you have not scanned yet. Clean the physical film first (anti-static brush, blower) — a clean scan is always faster to clean than a dirty scan.

The Time Math

A typical scanning lab operation for reference. A scan-back of 10 rolls (360 images) on DSLR scans:

Manual-only workflow: 3 minutes per image × 360 images = 18 hours of spotting.
Batch-first workflow: 12 minutes of DustMagic + 1 hour of QC and touch-ups = 1 hour 12 minutes.

That is a 15× time saving on a single job. For a lab doing two jobs like that a week, the payback on a DustMagic licence is inside the first week.