Efficient Forged Atomic Parallel

Manual Image duplicates & series
Contents

Image duplicates & series

The image comparison finds not only exactly identical images but also similar ones: the same shot at a different resolution, lightly edited variants or entire photo series. It also rates the image quality and automatically proposes the best image of each group to keep — ideal for pulling the one good photo out of a series and sorting out the duplicates.

10.1 Opening

The header bar holds the drop-down image module switch (icon with a picture frame and “▾”, tooltip “Open image modules — image comparison or metadata (ShotScope)”). It offers two entries: 🖼 Image comparison and 🏷 ShotScope (chapter 12). The two views are mutually exclusive; the switch is only usable with an active job.

10.2 Two modes: pixel comparison or photo series

On the left of the control bar sits the mode drop-down. The button always names the mode currently active; a click opens the list containing both options, with a tick on the active one:

ModeFinds…
🖼 Pixel comparisonidentical and perceptually similar images — across the whole folder, regardless of when they were taken
🎞 Photo seriesshots that belong together: images are bundled into series by EXIF capture time and camera, then refined within by similarity

Switching regroups immediately, without another scan. The first time you switch to 🎞 Photo series, the EXIF capture data is read in — this takes a moment and is shown.

The Threshold: slider (the “−”/“+” buttons change it in 5 % steps) determines how strictly “similar” is meant. At the very top, 100 %, only byte-identical images count — the comparison then works like a pure image duplicate finder. The lower you set the threshold, the more variants end up in one group. In 🎞 Photo series mode the Time gap slider (in minutes) is added: it defines the pause between shots at which a new series begins.

Screenshot 45: Control bar of the image comparison: “🖼 Pixel comparison” mode drop-down, threshold slider, zoom and the hit statistics “🖼 Image duplicates — N groups · M images detected”
Fig. 45 — Control bar of the image comparison: “🖼 Pixel comparison” mode drop-down, threshold slider, zoom and the hit statistics “🖼 Image duplicates — N groups · M images detected”

10.3 Result, quality and selection

The findings appear as group cards whose header names the kind: “≡ Identical”, “≈ Similar” (with a percentage) or “🎞 Series”. Every image is a tile with a preview, a Keep tick, the file name, information (size, resolution) and a quality rating in stars (★ to ★★★★★). The stars combine seven measurements — among them sharpness, resolution, contrast, exposure, colour, freedom from noise and level of detail; a tile’s tooltip shows the individual values. A double-click opens the image in the original view. If you want to see these seven values for two images side by side, use the quality bar in the image comparison viewer (section 10.4).

In every group the image with the best quality is automatically pre-selected to be kept (green frame), the rest are marked for removal (red frame). Two buttons per group change that:

ButtonEffect
✓ Keep bestkeeps only the best image (default)
Keep allkeeps all images of the group
⇆ Compareopens the image comparison viewer for this group (section 10.4)

Individual images can additionally be ticked or unticked by hand.

Screenshot 46: “Similar” group with the best image framed in green and duplicates marked red, star rating and the buttons “✓ Keep best” / “Keep all”
Fig. 46 — “Similar” group with the best image framed in green and duplicates marked red, star rating and the buttons “✓ Keep best” / “Keep all”
ℹ️ Limits of the scan
Image files between 4 KB and 50 MB are included (formats include JPG, PNG, BMP, GIF, TIFF, WEBP as well as — depending on the installed Windows codecs — HEIC/HEIF/AVIF/ICO), at most 10,000 images per scan. With more images the app asks you to choose a smaller folder.

Reviewing all remaining images: “Show unassigned”

A group needs at least two images. A shot with no similar counterpart therefore does not appear anywhere — correct for duplicate hunting, but useless when reviewing an entire shoot: exactly the largest part of the yield is missing there, and you can neither view nor compare nor select those images.

The Show unassigned box in the toolbar closes this gap. When switched on, further cards headed “◇ Unassigned #1, #2 …” appear below the findings, containing all remaining images. They behave like any other group: tiles with stars, double-click for the original view, ⇆ Compare opens the image comparison viewer.

PropertyWhy it is so
All images start red (deselected)This reverses the direction of work: you pick what should stay instead of sorting out what should go — the usual approach when reviewing a shoot. Whatever stays red, however, moves to the cache when you apply the selection; see the warning below.
At most 10 images per cardA card with hundreds of tiles cannot be taken in at a glance.
Sorted by capture time, split at time gaps and camera changesThis keeps together what was created together — otherwise they would be arbitrary groups of ten, and a comparison inside would compare random things. The time gap takes priority: better a card with six images of one situation than ten spread across two.
Cards start collapsedA shoot easily produces hundreds of them; the findings should stay at the top.
Do not count as findingsThe header (“N groups · M images detected”) still refers only to genuine similarity findings.
⚠️ Whatever stays red is cleared away when you apply
The cards start fully deselected. If you click ✓ Apply selection without setting anything green first, all unassigned images move to the recycle-bin cache — with a shoot that can be the largest part of the pictures. Nothing is lost in the process: the images sit in the cache and come back in full with ↩ Undo. They are only deleted for good with ✓ Finish — so check the result view before you confirm there.
💡 The usual route for photographers
After the scan, first go through the found groups — that is where the genuine duplicates and series are. Then switch on Show unassigned, expand the cards and click green everything you want to keep. Only then apply. If you want to keep a whole block, Keep all in the card header does the work for you.
ℹ️ Side effect: images that were invisible until now
These cards also show images whose content could not be evaluated — damaged files or formats for which no codec is installed on this PC. Until now they silently dropped out of every consideration. If you spot a file there that you actually consider to be fine, it is worth a look: usually Windows is missing the matching codec.
Screenshot 47: The “Show unassigned” box switched on; below the found groups the collapsed cards “◇ Unassigned #1, #2 …”
Fig. 47 — The “Show unassigned” box switched on; below the found groups the collapsed cards “◇ Unassigned #1, #2 …”

10.4 Comparing two images directly (image comparison viewer)

With similar shots a detail often decides: is the left image really sharper? Did someone blink? For that purpose every group card carries the ⇆ Compare button next to the keep buttons. It opens the image comparison viewer for exactly that group — there you place two images side by side, examine them in four display modes and mark the good ones green. The viewer changes only the keep markings; nothing is deleted or moved.

Layout

The window has two comparison containers: Comparison 1 (left) and Comparison 2 (right). At the bottom there is a thumbnail bar with all images of the group. This is how you assign images:

Action on the thumbnailEffect
Button 1 or 2load the image specifically into Comparison 1 or Comparison 2
Left-click on the tileload the image into the active container
Right-click on the tileload the image into the other container (the active one stays as it is)
Tick at the top right of the tilekeep the image (green) or discard it (red)

The active container is the one you are currently working with — recognisable by the coloured line beneath its header. You can switch it at any time with a click on the header of the container you want; in Side by side mode a click into the respective image is enough as well. That way you can carry on working on the right, for example, without the thumbnail bar overwriting the left image every time.

The orange marker or on the tile shows at all times which image is in which container. The tile’s frame remains the familiar traffic light: green = kept, red = will be sorted out — the same marking as in the group card behind it.

Each container’s header carries the keep tick right next to the file name, beside it the decision in words (“✓ will be kept” / “✕ will be discarded”), and with unequal quality the marker “▲ better” on the higher-rated image — plus the Apply image button on the right. Below that come size, resolution, date and star rating.

Screenshot 48: Image comparison viewer in “Side by side” mode: toolbar, the headers of “Comparison 1” and “Comparison 2” with a tick next to the file name, below the images the quality bar with the seven measurements per image, and beneath it the thumbnail bar with the markers ① / ② and the footer “N of M kept”
Fig. 48 — Image comparison viewer in “Side by side” mode: toolbar, the headers of “Comparison 1” and “Comparison 2” with a tick next to the file name, below the images the quality bar with the seven measurements per image, and beneath it the thumbnail bar with the markers ① / ② and the footer “N of M kept”

The quality bar beneath the images

Beneath the images there is a breakdown of the star rating for each container. On the left stands the container label with the overall score (e.g. “★★★★ 67 %”) in the colour of the stars; below it appear the seven measurements that make up this score — each with a percentage and a small bar. In Side by side mode each breakdown therefore sits directly beneath its image.

MeasurementWhat it judgesShare of the score
Sharpnessedge definition — the most important indication of whether a shot is spot on or slightly blurred25 %
Resolutionimage size in pixels20 %
Contrasttonal range between the brightest and darkest areas15 %
Exposuredistribution of brightness values; blown highlights and crushed shadows cost points10 %
Coloursaturation and colour variety10 %
Noisefreedom from noise — the higher the value, the cleaner the image10 %
Detaillevel of detail in relation to the file size; low values indicate heavy JPEG compression10 %

If the two containers hold different images, the better of the two is highlighted in green for each measurement. The “▲ better” marker in the header tells you which image is rated better overall — the quality bar tells you why. If a container is empty or the same image is in both, the highlighting is omitted.

💡 When the stars are equal, the bar decides
Two shots from the same series often get the same number of stars. The breakdown still shows differences: frequently one image is sharper, the other better exposed. As sharpness carries the largest share at 25 %, it is usually the stronger argument — exposure can be corrected afterwards, an out-of-focus shot cannot. The tooltip of each measurement states its share of the overall score once more.

The bar stays visible in all four display modes — the values belong to the two containers, not to the image arrangement. In the modes with a shared image area (Overlay, Slider, Difference) you recognise the assignment by the container label in the same colour as in the header above. ⇄ Swap swaps the breakdowns along with the images.

The four display modes

ModeShows…Own slider
Side by sideboth images separately next to each other
Overlaycomparison 2 lies on top of comparison 1 — ideal for spotting shifts and blurOpacity 0–100 %
Slidera vertical dividing line: comparison 1 on the left, comparison 2 on the rightPosition 0–100 % or drag the handle directly in the image
Differenceonly the deviation between the two images: black = identical, bright areas = differenceAmplification 1–10×

In Difference mode the footer additionally gives figures: “Ø deviation … % · maximum … %”. A value close to zero means the two images are practically the same. In the other modes the metadata comparison appears there: “Comparison 2 versus 1 — resolution … · size … · quality …”.

Screenshot 49: “Slider” mode: vertical dividing line with the handle in the middle of the image, comparison 1 on the left, comparison 2 on the right
Fig. 49 — “Slider” mode: vertical dividing line with the handle in the middle of the image, comparison 1 on the left, comparison 2 on the right
Screenshot 50: “Difference” mode with the “Amplification” slider and the figure “Ø deviation … % · maximum … %” in the footer
Fig. 50 — “Difference” mode with the “Amplification” slider and the figure “Ø deviation … % · maximum … %” in the footer

Tools

ToolFunction
⇄ Synczooming and panning act on both images at once (default: on). Switched off, the zoom applies only to the active container — marked by an orange line beneath its header.
Blinkalternates continuously between the two images. Even the smallest differences in sharpness or composition become visible as movement.
Griddisplays a rule-of-thirds grid — helps when comparing composition and horizon alignment.
⇄ Swapswaps the contents of comparison 1 and 2.
Zoom, 1:1, Fitslider up to 800 %; the mouse wheel zooms at the cursor position, dragging with the left mouse button pans, a double-click fits the image.

How to work through a group

  1. Click ⇆ Compare in the group card.
  2. Assign two images — using 1 and 2 in the thumbnail bar. After that a left-click on a tile is enough to reload the active container.
  3. Choose a display mode and examine (with nearly identical images, Difference or Blink leads to a decision fastest).
  4. For every image that should stay, press Apply image — or set the tick next to the file name.
  5. Assign the next pair and carry on. The arrow keys / page through the group in the active container.
  6. Return to the group view with ✓ Apply & close.
💡 Marking any number of images green per group
Apply image works additively: images of the group already marked green stay green. So you can go through a series pair by pair and collect as many images as you like. The button is greyed out as soon as this container’s image is already green. Undo individually via the tick; ↩ Reset restores the state as it was when the viewer was opened.
ℹ️ Keyboard shortcuts in the viewer
/ previous/next image in the active container · Space toggle keep · + / zoom · 0 fit · 1 original size · S sync · F blink · G grid · Esc close.
ℹ️ What happens technically when overlaying
In Overlay, Slider and Difference mode both images are stretched to the area of Comparison 1 so that they lie congruently on top of each other — this way images of different resolutions can be compared directly. For the sake of speed the difference is computed at a working width of at most 1,600 pixels; for judging deviations that is more than sufficient.

10.5 Removing — with a safety net

You trigger the actual sorting-out with ✓ Apply selection — the button sits in the module’s bar, to the right of 🔍 Scan. The deselected images are then not deleted immediately but moved to a temporary recycle-bin cache; afterwards the image result view opens (“✓ Result — N images remaining”). There you find:

ButtonFunction
🗑 Removemove further deselected images to the cache (“Only deleted for good when you finish”)
↩ Undobring all images back from the cache and return to the duplicate view
✓ Finishconfirm the clean-up — the cached images are now deleted for good
ℹ️ Nothing is lost before you finish
Removed images stay in the recycle-bin cache until ✓ Finish and can be restored at any time with ↩ Undo. Only finishing is final.

10.6 Merging series

With 📁 Copy selection you collect the kept (green) images of every group in a subfolder of their own beneath a target folder you choose — the originals are preserved (they are copied, not moved). SynQaiPro proposes the name of the series subfolder; you can adjust it per card. Should there be name collisions, the renaming kit (chapter 7) opens automatically so that all images get unique names.

Screenshot 51: Image result view with “🗑 Remove”, “↩ Undo” and “✓ Finish”; at the top the message “✓ Result — N images remaining”
Fig. 51 — Image result view with “🗑 Remove”, “↩ Undo” and “✓ Finish”; at the top the message “✓ Result — N images remaining”

10.7 In trial mode

🧪 Viewing free — deleting locked
In trial mode you can scan, group, change the threshold and the mode, view the quality and make your selection without restriction (image scanning is subject to no 10-file limit). Only the final removal is locked — the clean-up via ✓ Apply selection cannot be started (“Final deletion in the “image similarity” module is locked …”).
An unhandled error has occurred. Reload 🗙