# Actual English PDF compression checks

14 October 2026. The English Doc2X compression tool processed existing self-authored
synthetic English chart PDFs. These are digital teaching fixtures from October 10,
not photographs, physical scans or customer files. The text-and-chart input bytes
match the original in the already published compression proof ZIP.

Four real compression runs produced unchanged downloads:
- 4,379,754-byte text/chart input → Standard: 2,149,168 bytes, all three pages;
- same original → High Compression: 352,357 bytes, all three pages;
- same original → High Compression, range 2-3: 238,026 bytes, two selected pages;
- 4,378,912-byte image-only input → High Compression: 344,567 bytes, three pages.

Fast web view, Clear title/author information and Optimize Structure were enabled.
The UI presets are labelled Standard (75% quality) and High Compression (60% quality).
Those numbers are encoding parameters, not retained visual-quality percentages.
All runs started from the corresponding original, not a previous compressed file.
The two agent-created tool tabs were returned to Standard and All Pages afterward.
No privacy/browser setting was changed and no OCR task was created.

Text extraction from standard/high matches the original after whitespace is
normalized. The range output matches original pages 2 and 3, including their source
footer identifiers. The image-only input and compressed output both have zero
extractable text. Compression did not add an OCR text layer.

Source/Standard page images are 1800×2200 pixels; High/range/scan images are 900×1100.
First-page screenshots and detail crops are rendered from actual PDFs at the same
PDFium scale and coordinates. High compression visibly softens raster fine print.
The source texture was deliberately synthetic; this is not an accuracy benchmark
or a guarantee of readability on another document.

Original title/author metadata is absent from the outputs, while generated producer
and date metadata remains. This is not a claim of complete metadata sanitization.
The 94.57% size reduction for range 2–3 includes removing page 1 and must not be
compared as pure compression of the complete document. Standard/High full-document
reductions are 50.93% / 91.95% for this one fixture only.

No exact target-size control, universal compression ratio, physical print, forms,
signatures, source bookmarks or browser clipboard behavior is established here.
A smaller PDF is not automatically searchable, correctly extracted or readable.
The proof ZIP is locally assembled from two inputs, four native outputs and checks;
it is not one native tool export. UI size labels are rounded; compare exact bytes
with the receiving service’s stated limit.