For Professional Retouchers • Client Images Stay Private

Retouching QA Tool
The Final Check Before Client Delivery

You've retouched it 40 times. Are you sure you got everything?

Photoshop layer toggling is not a QA method — by the time your eye gets back to the canvas, you've already forgotten what you were checking. Blink testing shows your raw and retouched at the same fixed location in rapid alternation. Missed spots, clone stamp artifacts, and frequency separation issues pop immediately. 30 seconds. Every deliverable.

30-second QA workflow Blink, diff, and split wipe NDA-safe for client work

Toggling layer visibility is not the same as comparing images.

Every retoucher uses this as their QA method. None of them talk about the cognitive problem it creates.

What happens when you toggle a Photoshop layer

Your eye tracks to the Layers panel. Your hand moves to click. Your eye travels back to the canvas. The image is already different — but you're now rebuilding your visual impression of it from scratch. You are comparing what you see now to what you remember from 600 milliseconds ago. That is not a comparison. That is a memory test.

What happens during blink testing

Your raw and retouched image alternate at the same pixel coordinates, at precise intervals, while your eyes stay still. You're not comparing a perception to a memory — you're comparing two perceptions directly. Every change, intended or accidental, appears as visible motion. Your visual system detects it without conscious effort.

Every comparison mode, applied to retouching.

Most retouchers only toggle layers. You have three comparison methods — each reveals different categories of error.

👁️

Blink test

Rapid alternation between raw and retouched. Every change appears as motion. Catches missed spots, residuals, and accidental edits instantly.

🔴

Pixel diff heatmap

Mathematical comparison showing exactly which pixels changed and by how much. Verifies edits are contained to intended regions. Boundary spill shows up immediately.

↔️

Split wipe

Drag a divider across the image. Inspect specific regions at full resolution. See raw and retouched side-by-side at any point.

🔒

NDA-safe

All processing runs locally on your machine. Safe for client images under NDA — your files stay on your device.

30 seconds. Every deliverable. No more sending work with missed spots.

Add this as the last step before export. It catches what an hour of pixel-peeping in Photoshop won't.

1

Export your final

Finish retouching. Export your final TIFF or JPG from Photoshop or Capture One. Don't close your editor yet.

2

Drop raw + final

Drag your original raw export (or unretouched JPEG) as anchor. Drag your final export as a state.

3

Blink at 300ms

Every change — intended or accidental — appears as visible motion. Your eye snaps to anything that's different.

4

Verify with diff

Switch to pixel diff to confirm edits are contained to the intended regions. Boundary spill shows up immediately.

Original artwork alongside its pixel diff heatmap — retouching QA showing every edit, missed spot, and boundary change as glowing regions

Original (left) vs. pixel diff heatmap (right) — every change glows. Missed spots stay dark.

The errors that survive hours of zoomed-in editing

🔍

Missed spots and residuals

A blemish you overlooked at 200%. A stray hair you thought you removed. A background distraction that survived. Blink comparison makes these obvious because your eye is drawn to the thing that doesn't change when everything around it does — a residual element that should have been removed jumps out as a stationary artifact in a blinking field.

🖌️

Clone stamp and healing brush artifacts

Repeated patterns from clone stamp work, smudged edges from the healing brush, texture inconsistencies from patch tool overreach — these are nearly invisible at 100% zoom. They pop immediately when you blink between raw and retouched, because the repetition or discontinuity appears as an anomalous motion pattern.

🎨

Frequency separation issues

Texture layer inconsistencies from frequency separation work — areas where the texture feels disconnected from the color underneath, or where blending on the color layer crept beyond the intended mask. The pixel diff heatmap shows color shifts across the entire image, revealing where your frequency separation work touched areas you didn't intend.

📐

Dodge and burn color shift

Did your dodge-and-burn work shift the overall color balance? Did a heavy burn create an unnatural shadow density? The blink test reveals overall tonal changes because the entire image appears to shift when you blink — not just the specific areas you were working on.

✂️

Boundary spill from masking

When you mask and adjust a region, did the adjustment spill beyond the mask edge? This is especially common with luminosity masks and frequency separation masks that seem precise at 100% zoom but have soft edges that spread further than intended. Pixel diff mode shows exactly where change occurred — and where it shouldn't have.

📏

Liquify and warp displacement

Liquify, warp, and perspective corrections can subtly displace adjacent areas. The blink test makes geometric displacement immediately obvious — especially important for fashion and beauty retouching where proportions, symmetry, and anatomy are under scrutiny.

Multi-pass review for complex retouching jobs

For beauty work, compositing, and extensive color grading — you're not just comparing "before" and "after." You're comparing multiple passes: the raw, the skin work, the color grade, the final export. The tool supports up to 50 states, so you can load each pass as a separate state and blink through the entire progression of your work.

This is especially useful when reviewing work from an outsourced retoucher or a junior in your studio. Load the original, their first-pass delivery, your revision notes, and their corrected second pass. Blink between any two to verify that each round of feedback was actually implemented — not approximately addressed, but actually done.

Building this into your Lightroom and Capture One delivery workflow

If you work in Lightroom or Capture One, the workflow is: make your edits, export a proof JPEG at full resolution, then drag both the original exported preview and the retouched version into the tool. Thirty seconds of blink testing. If you spot something, go back to your editor, fix it, re-export, and run the check again. The whole QA loop takes under two minutes per image — less time than it took to pixel-peep the last section of skin at 300%.

Comparing two different edit directions

Beyond raw-versus-retouched QA, the multi-state support lets you compare two different editing directions for the same image — useful when a client asks for both a "natural" and a "polished" version, or when you're deciding between two color grades. Load both exports, blink between them, and show the comparison directly to your client. The blink test often makes the better version obvious to clients who can't articulate why one edit feels stronger than another. See also: before/after comparison for client presentations →

Questions from retouchers

Layer toggling requires your eye to travel to the Layers panel, click, then travel back to the canvas. By that point you've lost the instantaneous comparison — you're comparing what you see now to what you approximately remember from half a second ago. Blink testing shows the change at a fixed position with no eye movement required. The difference between the two images is presented directly to your visual system, not reconstructed from memory. It's a different cognitive process entirely — and a more reliable one.
Yes. All processing runs locally on your machine — your images are never transmitted to a server. This makes it safe for client work under NDA, HIPAA-adjacent workflows, and any situation where images can't leave your device. You can verify this yourself: disconnect your internet after loading the page and continue using the tool normally. It works identically.
The tool works with rasterized formats — PNG, JPG, WebP. For RAW comparison, export your raw conversion as a PNG or JPG first (from Lightroom, Capture One, or Camera Raw), then load both versions into the tool. The quality of the rasterized export is sufficient for QA purposes — you're not evaluating the RAW data itself, you're evaluating what the final delivered image looks like.
For blink comparison to work properly, both images should have the same dimensions and alignment. If your retouching included a crop, compare the pre-crop versions, or export both at the same canvas size with the same positioning. The pixel diff heatmap requires matched dimensions to be meaningful — comparing different crops will produce diff data that reflects the crop rather than the retouch.
Yes, and this is one of the most useful applications. Export both edit versions, load them as separate states, and blink between them. You can also show this comparison directly to a client during review — the blink test often makes the stronger edit immediately obvious even to clients who can't articulate why one version looks better than another.

More from Compix

Pixel Diff Tool

Verify edits are contained to intended regions with a pixel-level heatmap. Open →

AI Image Comparison

Compare Stable Diffusion outputs, Midjourney variations, and Flux generations. Open →

Scene Compositor

Build client presentations, moodboards, and comparison layouts with layered compositions. Open →

Thirty seconds. Every deliverable. No more missed spots.

Raw in the anchor. Final as the state. Blink at 300ms. Done.

Check Your Work Now →