Both formats place marks at absolute coordinates on a fixed page, so a conversion is a change of units and a flip of the vertical axis, not a re-layout. Done properly the text stays text. Done the easy way each page is photographed into an image, and everything searchable about the document is lost.
Where these files come from
For about a decade Windows installed a printer called Microsoft XPS Document Writer, and it sat in the printer list directly above or below Microsoft Print to PDF. People picked the wrong one. That is the origin of most XPS files in the world: someone meant to make a PDF, and made this instead.
The format itself is a zip. Inside are XML files describing each page, the fonts used, and any images. OXPS is the later version standardised by Ecma, with the same ideas and slightly different part names. Both open here.
Why the conversion is exact, not approximate
The two formats agree about far more than they disagree. Both fix the page size up front, both position every mark absolutely, both describe shapes as vector paths and text as runs of glyphs at a given point. So there is no reflowing, no guessing about where a paragraph should break, none of the uncertainty that makes converting a word processor document hard.
What differs is only bookkeeping, and it is exact bookkeeping:
The shortcut, and what it costs
Redrawing a page properly means reading each text run, finding the font it used, unscrambling that font, embedding it, and placing the glyphs. Rasterising means rendering the page to a bitmap and dropping the bitmap into the PDF. The second is far less work and looks identical in a thumbnail.
Rasterised pages
No searching, no selecting, no copying. Text softens the moment you zoom, files run several times larger, and screen readers get nothing at all. The damage is permanent, because the text is no longer in the file.
Redrawn as text and vectors
Search and copy work, edges stay sharp at any zoom, the file stays small, and the document remains accessible. This is what a conversion between two fixed-layout formats should produce.
The font trick, and how it is undone
Fonts inside an XPS package have the extension .odttf and will not open in any font tool. The first thirty-two bytes have been exclusive-ored with the identifier that forms the filename, which stops a font from being lifted straight out of a document and installed.
It is obfuscation rather than encryption, the key is written on the outside of the box, and the operation is its own inverse. Reading the identifier out of the part name and applying it again restores the original font, which is then embedded into the PDF so the text renders exactly as it did. Where an XPS only names a font that was never embedded, there is nothing to recover and a standard face is substituted; the tool reports that rather than letting you discover it later.
The parts inside the zip can be named anything, and page order lives in a separate document part that lists them. A converter that walks the archive instead of reading that list produces a correctly rendered document with its pages shuffled, which is easy to miss on a long file.
What comes across, and what does not
Text with its embedded fonts, colour, size and position. Vector paths with fills and strokes. Images placed on the page. Nested canvases with their transforms, including rotation. Page sizes exactly. That covers effectively everything in a document that came from a print driver, which is where most XPS files come from.
Gradients, soft masks, transparency groups and clipping paths are drawn without those effects, so a heavily designed page can come out flatter than the original. Documents like that are rare in this format for the same reason: print drivers make plain pages.
Questions people actually ask
What people ask after printing to the wrong virtual printer.
01What is the difference between XPS and OXPS?
OXPS is the standardised later version. Windows 8 onwards writes OXPS, earlier versions write XPS, and confusingly older XPS viewers will not open OXPS. Both work here, which is often the quickest way out of that particular trap.
02Can I go back, PDF to XPS?
Not here, and it is rarely what anyone wants. PDF opens everywhere and XPS does not, so the traffic runs one way.
03My file came out with the wrong fonts.
The document referenced fonts without embedding them, which print drivers sometimes do for fonts they assume are installed. Nothing in the file can supply them. The layout and spacing are still correct because the positions are absolute.
04Is the result searchable?
Yes, that is the point of doing it this way. Try selecting a line in the result to confirm before you archive anything.
05Does it handle multi-document packages?
Yes. A package can hold several fixed documents and they are converted in sequence into one PDF.