Analysis of PDF Compression Principles: Guide to Size Changes and Compression Strategies

PDF size consists of four parts: images, embedded fonts, document structure, and metadata. The proportion of each part varies greatly among different types of PDF. For scanned PDFs, images usually account for more than 90% of the total size, while the other three parts account for less than 10%.

For plain text PDFs or PDFs containing vector elements, embedded fonts can account for 20% to 60% of the total size. The proportion of document structure and metadata is usually within 5%, and vector content itself accounts for an extremely low proportion of the total size.

Re-rendering compression, also called bitmap compression, works by converting all vector and text content in the PDF to bitmaps uniformly, then saving the file according to the set DPI and quality parameters. This method has a significant compression effect on scanned PDFs, but it changes the original content attributes.

Lossless optimization compression does not change the clarity of the original content, and reduces file size only through three steps: object stream merging, font subsetting, and removing redundant metadata. Object stream merging integrates scattered objects to reduce redundant storage space.

For on-screen reading scenarios, when DPI is reduced from 300 to 72, the change in visual clarity between 300 DPI and 150 DPI is almost imperceptible, while the file size can be reduced by approximately 50%. When DPI is reduced from 150 to 72, the blurring of text edges becomes significantly more noticeable.

When JPEG quality is reduced from 100 to 50, the compression rate for the 50 to 80 quality range reaches 40% to 60%, and visual artifacts are almost invisible. When the quality drops below 50, color banding and edge blurring intensify rapidly, and the marginal gain of compression decreases.

Plain text PDFs are vector-based documents. The original size of a single page of text is usually between 1KB and 5KB, and the total size of an entire book is mostly controlled within a few megabytes.

If re-rendering bitmap compression is applied to this type of PDF, a bitmap corresponding to the page size will be generated for each page. The size of a single page bitmap can reach tens to hundreds of kilobytes, and the total size of the entire book will increase from a few megabytes to hundreds of megabytes.

Identifying the PDF type is the first step in selecting a compression strategy. First distinguish whether the document is a scanned document or an editable electronic document. Scanned documents consist entirely of image content, while editable electronic documents contain text and vector elements.

If the document is a scanned document, you can select re-rendering compression, and set DPI to 150 and JPEG quality to 60. This setting balances file size and clarity. If the document is an editable electronic document, prioritize lossless optimization compression to preserve the original clarity.

If the file size still does not meet requirements after lossless optimization, you can compress only the embedded images in the document separately, and retain the vector properties of text and vector content to avoid full-document bitmapping.