Compares the PDF approved by the customer with the production file prepared by the printer: a single label or an imposition. In an imposition every label is checked one by one; if one of eight differs, it is named by position.
No complex setup. Two PDFs are uploaded; the result arrives in seconds.
A single label, an imposition, or a multi-page file with several designs. The die line is found automatically; every design in the approved file is recognised, and every production label is matched to the design with the lowest pixel difference. Page order and page count do not matter; there is no AI in this matching, only measurement.
Each label is aligned with the approved file. Difference regions are found; shift is measured in direction and millimetres, colour difference as ΔE, and the colour’s definition is read from the file: spot name, tint, CMYK values. A shifted block is moved back and its content checked again.
A one-sentence explanation for every measured region: what changed, where and how. Regions with no difference are reported too; not counted, but kept in the report.
If all eight labels share the same difference, it is written once. If only one differs: “row 1, label 4 differs from the other 7” is stated and that label is shown separately.
It works where the human eye tires: a single letter, a single dot, a tenth of a millimetre.
A missing letter, a deleted colon, a changed digit, a lost ™; language independent.
Barcode, text block, logo: areas shifting in the same direction as one block, to 0.1 mm.
The colour’s name is read from the file: PANTONE 2995 C 80% → 100%, or a hand-defined colour in place of the spot. Not “a bit darker” — the definition.
Every label in the imposition is checked; the odd one is flagged by position, and orientation differences are stated separately.
Six designs in the approved file and five in production: the sixth is reported as “not in production”. A production design absent from the approved file is reported as “not in original”. The report is split per design: “Design 2 · 36 copies · CLEAN”.
If the files cannot be compared safely it does not say “clean”; it says uncertain and no credit is used.
The system recognises the structure of the uploaded files itself; you do not pick a mode.
Pool matching and every measurement are done without AI; the AI only explains the difference that was measured.
One-to-one comparison: content, shift, colour, objects.
Every copy is checked separately; the odd one is named by position. A size difference is compensated for flexo shrinkage.
Pool: designs are clustered, production labels are matched. Result per design: “Design N · K copies”.
Missing or extra designs are reported as separate warnings; there is no “page counts do not match” error.
It says “uncertain”; it does not count as clean and leaves the check to you.
One page: the imposition card and the whole label on top, a card for every difference below.
Red frame for a content difference, orange for a shift. The space bar switches original ↔ production; the eye checks too.
Original and production crops side by side, the measured value, the AI explanation. Under the shift card an accordion: the content checked after moving the block back.
0.25 mm left · PANTONE 2995 C 80% → 100% · 1/8 labels. The whole report downloads as a single HTML file.
Send us a file; we will run the check together. No installation, no download; it runs in the browser.