The practical answer
Open the Word document tool, choose Extract DOCX contents and download docx-content.zip. document.txt comes from the main document XML; images contains files stored under word/media. The output does not recreate pages or automatically apply Word crops, overlays and layout. Review each complete image before sharing it.
One picture can lead to three different deliverables
Microsoft’s WordprocessingML structure reference separates the main document from headers, footers, comments and other parts. This tool reads main-document text and collects packaged media separately. Neither output is a complete rendering of the Word page. Missing header text, comments or footnotes in document.txt does not prove that those parts are absent.
Extracted media also lacks an automatic map to figure numbers, paragraph positions or intended uses. A document may reuse one stored picture in several places. A stored media file may also be outside the body area you reviewed. Do not use the number of files as the number of visible picture placements.
| Output | What it retains | What still needs review |
|---|---|---|
| Packaged media | Files stored inside the DOCX | Compression and cropped-out regions |
| Visible page composition | The intended crop and page relationship | Separate controlled rendering or screenshot |
| Plain body text | Text from the main document XML | Complex order, tables and other document parts |
Why a square picture can become a wide extracted file
Consider an original size example, not a customer document. A 2400×1600 photograph is placed in Word and centrally cropped to a visible 1600×1600 square. The 400px on each side is excluded from that display. Microsoft’s SourceRectangle reference describes how a source region can be specified separately from the media itself.
If the full photograph remains in the package, extracting it returns 2400×1600 rather than the square. However, the author may already have deleted cropped areas or compressed the picture. A packaged file is therefore not guaranteed to be the camera original. Open the actual output rather than inferring its content from the page.
Separate extraction from the publication crop
Assume that a square picture in Word means the extracted file contains only that square.
Open the extracted wide image, review all its edges, then make the approved square publication version separately.
Extract into an empty folder and build a usage map
- Keep the original DOCX and work from a copy. Renaming an old .doc file or scanned PDF does not make it a DOCX.
- Choose Extract DOCX contents in the Word tool, select one supported file and process it. Confirm that the downloaded ZIP opens.
- Review document.txt and images in a new folder. Record each media filename, dimensions, whether it opens and its intended use.
- Compare each selected asset with the original Word page. Create a separate publication version when the visible crop must be retained.
- Send only approved assets. Do not treat the entire extracted ZIP as an automatically safe external handoff.
A small inventory makes the pictures usable
For an illustrative two-page product note, you might record image1.png as a header mark, image2.jpg as the cover photograph and image3.png as a process diagram. That mapping is invented for this example; the tool does not infer it. A person must check each placement. A name such as image2 does not establish that it belongs on page two.
If the report came from a Word template filling workflow, check template identifiers separately from business data. Then use a file handoff manifest to identify final versions, recipients and review results. Keep the unfiltered working archive separate from the publication package.
Four acceptance conditions
- Every selected asset opens and its actual format and pixels have been checked.
- The source placement, intended use and new filename can be matched.
- Cropped-out edges, backgrounds and additional packaged assets have been reviewed.
- Only approved assets are delivered; the source and working archive stay internal.
Text, drawing formats and counts have boundaries
The current text extraction converts main-document XML to plain text. It does not guarantee complete reading order, complex table structure or revision interpretation. Vector media may retain its original format; failure to open in a basic photo viewer does not automatically mean corruption. Externally linked pictures may not exist under word/media and are not fetched by this operation.
This operation processes the input in the browser, but you still control the download folder, browser storage and later sharing destinations. If extraction fails, check whether a trusted Word application opens the file normally. Do not treat unknown macro-enabled or protected files as ordinary, interchangeable media packages.
Common questions
Is the tool enlarging an image that looks bigger after extraction?
Not necessarily. Word may change only the displayed size or crop while retaining a larger stored file. Compare actual pixel dimensions separately from page appearance.
Why does the media count differ from the visible picture count?
One file can be reused, and packaged media can belong to headers or other locations. This tool counts stored media rather than every visible placement.
Does missing comment text prove there are no comments?
No. document.txt is derived from the main document XML. Review other document parts in a suitable Word application when completeness matters.
Further reading & sources
Put the guide into practice
Open the relevant tool, preview the result and check your exported file.
