Recovering Ami Pro Equations and WMF Diagrams

The silent-loss problem

The worst conversion failure is not the one that errors out. It is the one that succeeds, produces a clean-looking document, and quietly omits content. You do not find out at conversion time; you find out when someone reads a report that says "the relationship is given by the equation below" followed by a blank line.

Technical Ami Pro documents are particularly exposed to this. Engineering calculations, laboratory reports, university course notes, standards submissions, and patent drafts all lean on figures and formulas, and those are exactly the elements a text-oriented conversion cannot see.

Which means verifying diagram and equation recovery is not an optional refinement. For a technical archive it is the acceptance test.

What Ami Pro actually embedded

Ami Pro used the Windows Metafile family for drawings. WMF and its 32-bit successor EMF are vector formats: they store drawing commands — lines, arcs, filled polygons, text placement — rather than a grid of pixels. That is why an Ami Pro flowchart from 1994 can be re-rendered crisply at any size today, provided something can read it.

Equations came from Ami Pro’s built-in equation editor, which stored formulas as structured markup describing the mathematical layout: numerators and denominators, radicals, sub- and superscripts, sums and integrals with their limits. It is a description of mathematics rather than a picture of it, which turns out to be very useful.

Both object types live inside the .sam stream alongside the text. Because the format is text-based with bracketed markup, embedded binary and structured data is encoded within it — recoverable, but only by a parser that knows the encoding is there.

Why generic converters lose them

How recovery works in practice

Diagram recovery starts by locating embedded metafile data in the stream, reconstructing it into a valid WMF or EMF file, and then rendering it. From there the image is written out and referenced from the converted document — embedded in DOCX and PDF, or written as a linked image file for HTML and Markdown output.

Equation recovery runs two paths in parallel because different outputs need different things. The formula is rendered to an image so that DOCX, PDF, and RTF show the mathematics laid out correctly; and it is translated to LaTeX so that Markdown, HTML, and text outputs carry a representation a machine can read, search, and re-render.

That dual approach is what makes technical archives genuinely reusable. An image of an equation is fine to look at and useless to search. LaTeX is searchable, indexable, convertible to MathML, and can be pasted straight into a modern editor when someone needs to reuse the formula.

Choosing an output format for technical documents

• PDF — the reliable visual record; diagrams and equations render exactly as laid out
• DOCX — editable text with figures and equation images embedded; use the LaTeX to rebuild formulas you need to edit
• Markdown — best for search and AI pipelines, because equations arrive as LaTeX rather than pictures
• HTML — figures become linked image files and formulas can be rendered client-side; good for intranet publishing
• TXT — expect to lose all visual content; useful only for triage and full-text search

Verifying that nothing was dropped

Build your review sample around figures rather than around file size. Search the plain-text output of the archive for phrases like "Figure", "Table", "Equation", "shown below", and "see diagram", then check those specific documents in the converted output. Cross-references are a reliable map to where visual content should be.

For each one, confirm three things: the image is present rather than a placeholder, it renders at a sensible size and is legible, and it appears in the right position relative to the text that references it.

For equations, additionally confirm the mathematics is correct — subscripts and superscripts in the right places, radicals covering the right terms, limits attached to the right operators. A rendered equation that is subtly wrong is more dangerous than one that is missing, because it looks authoritative.

A note on documents that were designed, not just written

Ami Pro supported frame-based layout, which some authors used heavily for newsletters, brochures, and report covers. Those documents are less about recoverable content and more about composition, and their layout does not have a clean equivalent in DOCX.

For that subset, treat PDF as the authoritative output and DOCX as a convenience copy. You keep a faithful visual record of what the document looked like, plus editable text for anyone who needs to reuse the words, and you avoid spending days trying to rebuild 1990s page design in Word.