Conference September 2012 📍 San Jose, California, USA

PDF Days North America 2012 — San Jose: PDF/A-2 Published and Enterprise Archiving

PDF Days North America 2012 took place in San Jose in September 2012, shortly after the publication of ISO 19005-2 (PDF/A-2). Discussions centred on migrating enterprise document archives to PDF/A-2, leveraging new features like optional content and embedded files, and planning for the PDF/A-3 standard.

Official event page →

Key takeaways

2012 was the most productive year for new PDF standard publications: PDF/A-2 (2011), PDF/UA-1 (2012), and PDF/A-3 (2012) all available at this conference.
PDF/A-2 key additions: JPEG2000 compression, transparency/layers permitted, OpenType fonts, embedded PDF/A-compliant files, new /u (Unicode) conformance level.
Open-source PDF ecosystem maturing: PDFBox, iText/AGPL, Poppler, Ghostscript — veraPDF in early development.
PDF/A-3 hybrid document pattern introduced: embedded files of any type enable PDF-as-data-container — ZUGFeRD (then in development) as the flagship use case.
Silicon Valley location brought stronger developer and open-source community engagement than DC editions.

PDF Days North America 2012 took place in San Jose, California in September 2012, organised by the PDF Association. The conference arrived at a specific moment of transition: ISO 19005-2 (PDF/A-2) had been published in 2011, and the early months of 2012 had seen the publication of both PDF/UA-1 (ISO 14289-1:2012) and PDF/A-3 (ISO 19005-3:2012) — making 2012 the most productive year for new PDF standard publications since PDF 1.4 (2001). San Jose brought Silicon Valley representation: a stronger technical developer presence than the Washington D.C. editions, with sessions addressing PDF library implementation, open-source tools, and the technical specifics of the new standards for software engineers building PDF tooling.

PDF/A-2: what the new standard added

ISO 19005-2:2011 (PDF/A-2) was the first update to the PDF/A archiving standard since the original PDF/A-1 (2005). It was based on PDF 1.7 (vs PDF 1.4 for PDF/A-1) and added several significant capabilities: JPEG2000 compression — PDF/A-1 restricted image compression to JPEG (DCT), Flate (lossless), and CCITT/JBIG2 (for monochrome). PDF/A-2 added JPEG2000 (JPX), which offers better compression at the same quality for photographic images. This was particularly relevant for scanned document archives — JPEG2000 at medium quality settings could reduce file size by 30–50% compared to JPEG at equivalent visual quality. Transparency permitted — PDF/A-1 prohibited transparency entirely. PDF/A-2 permitted transparency and layers (optional content groups), enabling modern graphic design documents (annual reports, marketing materials) to be archived as PDF/A-2 without flattening. This was a major enabler for design-heavy documents that had previously required significant processing before archiving. Embedded PDF/A-compliant files — PDF/A-2 permitted embedding of PDF/A-compliant files as attached files (but not arbitrary file types — that would come in PDF/A-3). This enabled embedding of supplementary documents, appendices, or referenced exhibits within a PDF/A archive package. OpenType fonts — PDF/A-1 required fully embedded PostScript Type 1 or TrueType fonts. PDF/A-2 added full OpenType font embedding support, enabling archiving of documents using contemporary OpenType typefaces. Conformance levels — PDF/A-2 added a /u (Unicode) conformance level alongside /a (accessible) and /b (basic). PDF/A-2u requires all fonts to have ToUnicode CMaps, enabling reliable text extraction, without requiring the full structural tagging of 2a.

Open-source PDF tooling in 2012

The 2012 San Jose conference had stronger open-source tool representation than previous editions. The open-source PDF tool ecosystem was maturing: Apache PDFBox — The primary Java open-source PDF library, maintained by the Apache Software Foundation. By 2012, PDFBox supported PDF creation, manipulation, text extraction, and rendering. PDF/A validation was being developed. Sessions discussed PDFBox's capabilities and upcoming releases. iText — Then available under AGPL (with commercial licensing). The most feature-complete Java PDF library, widely used in enterprise Java applications for programmatic PDF generation. PDF/A creation and digital signature support were strong suits. Poppler — The C++ PDF rendering library used in Linux desktop PDF viewers (Evince, Okular) and the pdftotext/pdfimages utilities. Strong text extraction and rendering; limited creation capabilities. Ghostscript — The open-source PostScript/PDF interpreter. The workhorse for PDF conversion, rasterisation, and manipulation in Unix-based workflows. The -dPDFA flag for PDF/A output was available but produced imperfect results — commercial tools were more reliable for compliant PDF/A generation. veraPDF (pre-release) — The PDF Association had announced the veraPDF project — a purpose-built open-source PDF/A validator. The project was in early development in 2012 and would be productively released in 2017. Sessions previewed the project goals and architecture. pdfminer (Python) — The primary Python PDF parsing library at the time. Text extraction with layout analysis, font inspection, and PDF object access. Slower than commercial alternatives but freely available.

PDF/A-3 and the hybrid document opportunity

PDF/A-3 had been published in 2012, extending PDF/A-2 by permitting embedded files of any type. The technical session on PDF/A-3 previewed the business case: The ZUGFeRD connection — Work on the ZUGFeRD standard (which would be published in 2014) was underway in Germany. ZUGFeRD's use of PDF/A-3 to embed invoice XML was the most prominent planned deployment. While ZUGFeRD itself was a European standard, North American attendees saw the pattern — PDF/A-3 as a hybrid format — as applicable to a wide range of use cases. North American e-invoicing context — The US did not have a ZUGFeRD equivalent in 2012, but the broader electronic invoicing market was active. EDI (Electronic Data Interchange) was the incumbent standard for large enterprise-to-enterprise invoicing. PDF/A-3 with embedded XML offered a more accessible path for mid-market and small business invoicing — the PDF provided human readability that EDI lacked. Scientific data publication — San Jose's technology community had representatives from scientific publishing. Embedding datasets in PDF papers was an emerging practice. PDF/A-3's formalisation of this pattern with the /AF associated file mechanism and standardised relationship types (/Data, /Alternative, /Supplement) provided a formal specification for what had been ad hoc. Implementation requirements — Sessions walked through the specific requirements for creating valid PDF/A-3 documents with embedded files: the /AF array in the document catalogue; the AFRelationship key; the /Params dictionary for file metadata (size, checksum, dates); and the XMP metadata requirements for the container PDF. See PDF/A for detailed technical requirements.

Frequently asked questions

What is the difference between JPEG and JPEG2000 compression in PDF/A?

JPEG (DCT compression in PDF) and JPEG2000 (JPX compression) are both lossy image compression formats — they reduce file size by discarding some image detail. JPEG2000 advantages: (1) Better compression ratio at equivalent visual quality — 30–50% smaller than JPEG for the same quality. (2) No blocking artefacts — JPEG produces visible 8×8 block patterns at high compression; JPEG2000 uses wavelet compression that produces smoother degradation. (3) Lossless mode available — JPEG2000 can be lossless (no data discarded) while still providing 2:1 compression for typical images; standard JPEG is always lossy. (4) Region of interest encoding — JPEG2000 can encode parts of an image at higher quality than the rest. In PDF/A: PDF/A-1 allows only JPEG (plus Flate and CCITT/JBIG2). PDF/A-2 and later allow JPEG2000 in addition. For new archives of photographic content, JPEG2000 provides better quality-to-size ratios.

What is PDF/A-2u (Unicode conformance) and when should I use it?

PDF/A-2u is a conformance level that requires all fonts in the document to have ToUnicode CMaps — ensuring that all text in the document can be reliably extracted, searched, and copied. Unlike PDF/A-2a (which requires full accessibility tagging), PDF/A-2u does not require a structure tree or reading order. It is positioned between PDF/A-2b (basic visual compliance) and PDF/A-2a (full accessibility). Use PDF/A-2u when: (1) Reliable text extraction and search are required (full-text indexing of an archive, search engine crawling, AI text processing), but full accessibility tagging is not currently feasible. (2) You want to prevent the garbled text extraction problems common in PDFs with incomplete ToUnicode data. (3) Regulatory requirements reference text extractability but not full accessibility structure. PDF/A-2u is the most appropriate conformance level for searchable document archives where accessibility tagging is being incrementally added.

Try it in your browser

The PDF tools and standards discussed at this event are available free — no account needed.

Edit PDF Compress PDF Annotate PDF Open editor