Sewlore / technical reference / version 1.0.0

Check both axes.
Then check the paper.

A reproducible PDF geometry kit for understanding scale changes in sewing pattern print workflows. Start with known A4 and Letter references, then compare the geometry of an actual saved output.

This release includes 4 actual Chrome HTML-to-PDF measurements and six generated reference/control files. Native PDF-viewer print-setting experiments: 0. Physical printer measurements: 0. Generated controls remain separate from browser output.

Download a known reference

A4

210 x 297 mm, portrait, one page.

A4 reference PDF

100 mm square, 2 in x 1 in rectangle, 180 mm line and a 5 mm inset border.

US Letter

215.9 x 279.4 mm, portrait, one page.

Letter reference PDF

The same measurement paths on a different page size. The source file is an input, not a printer result.

Measure independent directions

A square can shrink equally in two directions, or change width and height by different amounts. Extract both axes rather than inferring one from the other.

Scale ratio = output length / reference length.
Signed error (%) = 100 x (output - reference) / reference.

Measure the path centre line. Line thickness, screen zoom and screenshot pixels are different quantities. Use a physical ruler to check a real print.

Uniform and unequal scale changesIllustrative diagram comparing a square with a 95 percent uniformly reduced square and a rectangle with 96 percent width and 104 percent height.100 x 10095 x 9596 x 104ReferenceUniformUnequal axesIllustration only; this diagram is not a calibration ruler.

Actual Chrome HTML-to-PDF measurements

These are saved browser-engine outputs from original HTML/SVG fixtures. Each case retains its exact developer print parameters, browser version, source hash and output hash. They do not test the PDF viewer's Actual size/Fit controls or a physical printer.

Returned PDF producer metadata: Skia/PDF m154. The full browser version was not exposed by the tab-scoped capture capability; the producer label is not a complete Chrome build identifier.

What this captured run shows

Output page dimensions are reported from the actual MediaBox, not rounded to the requested paper name. For example, the A4 CSS-size output can have small numeric differences from 210 x 297 mm. These observations characterize this controlled rendering path only.

Observed Page.printToPDF output geometry; all three calibration paths extracted
Actual outputRecorded parametersPDF page sizeSquare width / heightWidth / height error
A4 CSS size, scale 1scale=1; preferCSSPageSize=true; paper=8.26772 x 11.69291 in209.88865 x 297.01066 mm99.99832 / 99.99832 mm-0.00168% / -0.00168%
A4 CSS size, scale 0.95scale=0.95; preferCSSPageSize=true; paper=8.26772 x 11.69291 in209.88865 x 297.01066 mm94.99841 / 94.99841 mm-5.00159% / -5.00159%
A4 source, Letter paper, CSS priority offscale=1; preferCSSPageSize=false; paper=8.50000 x 11.00000 in215.90000 x 279.40000 mm94.07284 / 94.07284 mm-5.92716% / -5.92716%
Letter CSS size, scale 1scale=1; preferCSSPageSize=true; paper=8.50000 x 11.00000 in215.90000 x 279.40000 mm100.00000 / 99.99999 mm-0.00000% / -0.00001%

Download browser measurement CSV · Exact parameters, hashes and full measured geometry JSON

Inputs: A4 HTML/SVG and Letter HTML/SVG. Downloaded output PDFs preserve the actual returned bytes. Small numerical differences belong to this rendering path and should not be interpreted as physical printer tolerance.

Measured detector controls

These dimensions were extracted from generated vector paths and checked against their explicitly specified transforms. Numerical tolerance: 0.005 mm. That tolerance is not a physical printer accuracy claim.

100 mm square: generated inputs and known-transform controls
FileEvidence classExtracted widthExtracted heightWidth / height error
A4 referenceReference input100.00001 mm100.00001 mm+0.00001% / +0.00001%
Letter referenceReference input100.00001 mm100.00001 mm+0.00001% / +0.00001%
A4: uniform 95% controlKnown software transform95.00001 mm95.00001 mm-4.99999% / -4.99999%
Letter: uniform 95% controlKnown software transform95.00001 mm95.00001 mm-4.99999% / -4.99999%
A4: 96% width / 104% height controlKnown software transform96.00001 mm104.00001 mm-3.99999% / +4.00001%
A4 to Letter: whole-paper fit modelKnown software transform94.07408 mm94.07408 mm-5.92592% / -5.92592%

Download extracted geometry CSV · Expected geometry CSV · Full measurement JSON · Versioned file hashes and units

What the A4-to-Letter model assumes

The whole-paper model uses min(215.9 / 210, 279.4 / 297), with no margins or printer restrictions. Its scale is about 94.0741%. It is a mathematical control, not an observed print setting result. A printable-area fit can differ.

Run an actual software comparison

  1. Open the PDF reference in the application you want to test.
  2. Open the visible print UI. Record the exact application, destination, paper, orientation and scale labels. Capture a settings screenshot.
  3. Save the actual output. Keep the original PDF and screenshot, including SHA-256 hashes.
  4. Inspect width and height independently, then visually review clipping and page boundaries.

Destination matters. Chromium documents a source-size path for Save as PDF. It can retain the PDF's original page size, which is different from printing to a PDF printer destination. If the UI does not expose paper or scale controls, record that fact instead of inventing applied settings.

Suggested cases: matched A4 at Actual size; matched Letter at Actual size; A4 to Letter at Fit to paper; Letter to A4 at Fit to paper. Use only controls that actually exist in the captured destination.

No native outcome is listed until a real output and settings receipt have been measured.

Reproduce the geometry checks

The generator and analyzer are downloadable Python scripts. Their control results should not be mistaken for an application benchmark.

python build_atlas.py
python inspect_pdf_geometry.py --all-generated
python inspect_pdf_geometry.py --native-receipt incoming/your-case.receipt.json

Reference/control generator · Vector geometry analyzer · Full method and limitations · Preparation environment

The analyzer reports missing or rasterized markers as unextractable. Raw path extents do not prove visibility after clipping. It does not guess missing dimensions or characterize a physical printer.

Read the methods behind the settings

For sewing patterns, check one actual test page before printing the full file. Confirm the pattern's own instructions and measure your printed page.