CLOUD NOTES: PDF READABILITY MINI LAB
Four synthetic one-page A4 PDFs. No private data, signatures or form fields.

Open all four in the same viewer at the same zoom. Compare the reference,
amount, 8 pt line, thin rules and colour swatches. Check selection/search too.
The three raster PDFs have no embedded OCR layer; a viewer may add its own recognition.
Smallest does not automatically mean acceptable. Do not use these as official records.

Read measurements.json for exact bytes and SHA256 hashes.
Compare results in comparison-sheet.txt.

Optional reproduction: Python 3, ReportLab, Pillow and Poppler's pdftoppm.
Run python reproduce.py in a separate folder containing that script.
It writes the four PDFs, a scan-source.png and measurements.json in its own folder.
No network requests or deletion operations are used.
Recorded environment: ReportLab 4.4.9, Pillow 12.3.0, Poppler 26.05.0.
Different library/rendering versions can produce different bytes and sizes.
JPEG quality values are Pillow settings, not portable presets for every app.

Method: generate native text/vector PDF; render it at 300 ppi RGB; embed
that raster losslessly; separately downsample to 150 ppi grayscale Q75
and 72 ppi grayscale Q25 JPEG, then place each on the same A4 page.
This is a synthetic illustration, not a benchmark of a named PDF compressor.
Author/publisher: Cloud Notes by Aain Ul Raza, 26 October 2026.
Guide: https://aainulraza.cloud/guides/reduce-pdf-file-size-without-losing-readability/
