What PDF conversion preserves — and what it cannot

A PDF usually stores positioned text, images, vector paths, fonts, and page geometry. It does not have to preserve the semantic structure of the document that created it. A heading may be represented as larger text at a coordinate rather than a true heading element; a table may be individual words drawn in columns.

That distinction explains why converting an office document to PDF is reliable for visual delivery, while reconstructing an editable office document from an arbitrary PDF is a much harder problem.

Text extraction works when the PDF contains a real text layer. The result can preserve words and lines, but layout, columns, tables, images, annotations, and typography will not become equivalent plain-text structures.

A scanned document may contain only page images. Convyx does not claim OCR support, so an image-only PDF will not turn into meaningful text without a separate OCR system.

Creating a PDF from text is deterministic because the input is simple. The converter can lay out lines on pages and create a portable document, but plain text does not carry fonts, headings, links, or rich layout to preserve.

Keep the source text alongside the PDF. Use the PDF for distribution or printing and the text for future edits.

Check the output against the original, especially page order, characters outside basic Latin, columns, tables, and line breaks. For legal, financial, or accessibility-sensitive documents, human review is part of a responsible workflow.

Convyx intentionally blocks PDF-to-office targets today rather than presenting text extraction as faithful document reconstruction.