Processed locally
Your input stays in this browser.
What the checker measures
The browser reads the selected image into a canvas, tries its local QR decoder, and samples dark and light pixels around the detected image. It reports the decoded text, a luminance contrast estimate and whether a pale border is visibly present at the edges. Nothing is uploaded. The result is evidence about this raster file, not a print-quality certificate for every camera, substrate or lighting condition.
Contrast uses luminance, not colour names
Relative luminance is approximated from linear red, green and blue components: 0.2126R + 0.7152G + 0.0722B. Black on white gives a very large ratio; dark blue on pale yellow can look strong to a designer but have less separation after a phone camera compresses it. The checker displays sampled values so a result is not reduced to a mysterious red or green badge.
A concrete sample
If a dark sample has luminance 0.05 and a light sample 0.95, the contrast ratio is (0.95 + 0.05) ÷ (0.05 + 0.05) = 10:1. If a coloured background lowers the light area to 0.45 while dark modules are 0.15, the ratio is 0.50 ÷ 0.20 = 2.5:1. Both images may contain the same modules, yet the second gives the camera much less separation.
The quiet zone is geometry
A QR detector expects a clear field around the three finder patterns. Four modules are a common minimum design target, but an uploaded bitmap does not announce its module scale reliably after cropping or resampling. This tool therefore reports visible blank edges and warns when dark pixels touch the border. It cannot reconstruct a margin that a screenshot has already clipped.
Decode first, then diagnose
A successful decode says the current file contains a recoverable payload. It does not make a thin quiet zone safe on glossy packaging. A failed decode can result from blur, rotation, a small image, unusual inversion or an actual code defect; it is not proof that contrast alone caused the problem. Keep a plain black-on-white export with the same text as a control and compare it with the decorated file.
Why photographs need caution
A photograph includes shadows, reflection, perspective and camera white balance. Sampling the corner of a glossy label may report a bright highlight even though a scanner sees glare across a finder pattern. Photograph a printed proof square-on in even light, then also test the original export. If export decodes and the photograph does not, the evidence points to reproduction or capture rather than the encoded bytes.
Typical corrective action
Restore a plain light background, keep dark modules uniform, add margin outside the matrix, and enlarge the final physical size if cells are soft. Do not add a logo halo or crop the white edge merely to make a layout look balanced. Change one condition at a time and retain the failed image, control export and revised proof; that comparison isolates the cause faster than a sequence of unrelated redesigns.
What a decoded image cannot disclose
A camera decoder returns recovered payload data, not the original segment plan, printer settings, or the condition of the physical label. Several valid QR encodings can carry the same text. The inspection result is consequently limited to this image's pixels and its current decode attempt; it cannot reconstruct how an unknown symbol was authored or certify another copy.
Keep the sampled file identifiable
When a check result informs a print decision, retain the exact image that was loaded, not a later screenshot of the browser result. Note its pixel dimensions, file format, export settings and the decoded payload. A PNG that has been copied through a messaging application may have different compression or cropping from the approved source, and a JPEG photograph may contain perspective that was absent from the artwork. Re-running the checker on both files is useful precisely because it preserves that distinction. A high contrast ratio on an original export should not be used to dismiss a failed label photograph; it says only that the clean digital source had separation before printing and capture introduced new variables.
Compare two explicit samples
Take the same QR artwork in black on white and in the proposed brand colours. Load both files separately and record decode result, dimensions and contrast estimate. If black on white decodes while the brand version does not, preserve the payload and change only colour or surrounding artwork. If neither decodes, inspect matrix size, cropping and export process before discussing colour. This turns a subjective claim about contrast into two files and two observable results.
Inspect pixels first, then inspect the physical label
Upload the final exported image rather than a design mock-up. A checker can measure the darkest and lightest sampled areas and look for a pale border around the QR matrix, so it can reveal a background photo, translucent logo panel or crop that has entered the quiet zone. It cannot measure reflection from a metallic packet, focus at a shop counter, or ink gain after production. ISO/IEC 18004:2024 covers QR Code symbol characteristics and production quality requirements: https://www.iso.org/standard/83389.html . Treat a passing image report as evidence about that exact file, not as a verdict on every future PDF export or print run.
Diagnose a failure by its layer
If an untouched PNG passes contrast and margin inspection but the final label fails, compare a scan of the flat control with one of the finished substrate. A control pass plus a substrate failure points toward print, gloss, curvature, scale or placement. If the uploaded final raster already lacks a pale border, correct the artwork crop before changing printers. If the image decodes but opens the wrong address, contrast is irrelevant: inspect the stored payload with the content parser or reader. Write down image filename, measured result, final dimensions and device result. That sequence prevents a common false repair in which someone darkens the modules to compensate for a missing quiet zone, or changes the payload because a laminate produced glare. This tool evaluates what pixels contain; a real scan test decides whether the physical code remains usable.
Use the checker after every image conversion
A design can pass as SVG, fail after export to JPEG, then pass again after an unrelated crop is removed. Run the checker on the exact raster file sent to print or embedded in the final PDF, naming that file in the release record. It is an efficient preflight because it catches a damaged margin before materials are produced, but it must not replace a physical sample scan.
Compare margin loss with the source export
When the checker reports a weak or missing quiet zone, open the original SVG beside the uploaded image. A correct source with a failed raster points to an export or crop stage; a failed source points to the generator or design choice. This comparison gives the production team a precise repair target instead of a vague instruction to make the QR code darker.