How images get from a Markdown link into a Word document
A Markdown image is just a reference:  says where a picture lives, not what it contains. Word, on the other hand, expects pictures to be inside the .docx, stored as files under word/media/ and referenced through a relationship id from a drawing element in the text. Word can technically link to external pictures, but linked images break the moment the file is emailed or the URL changes, so a good conversion embeds every image. That is what MDtoDOCX does, and the details of fetching, sizing and accessible descriptions are where converters differ.
Fetching remote images
Images at https URLs are fetched from your browser. When the host sends permissive CORS headers, the bytes are read directly. When it does not, which is common for image hosts and wikis, the request goes through a small proxy on our server that only accepts image content types, caps each file at 5 MB and keeps no copy. The Markdown text itself is never sent; only the image URL is. Private URLs that need a login cannot be fetched by either route, so download those files and drop them onto the converter instead.
Base64 and local paths
Data URIs are decoded in place with no network request at all, which is why they work offline and are common in exported notebooks and AI chat exports. Relative paths such as ./figures/plot.png cannot currently be resolved by the web converter. Use a data URI or a public HTTPS image URL for those images.
Sizing to the page
Each image's dimensions are read from the file header. The printable width is the paper width minus both margins. An image wider than that is scaled down proportionally; a smaller image stays at its natural size, so icons remain small and screenshots remain legible. Each picture sits in its own centred paragraph and can still be resized in Word.
Alt text stays accessible
Markdown alt text is stored in the drawing's title and description fields, where screen readers and Word's accessibility checker can find it. Non-empty alt text is also placed in a Word Caption-styled paragraph beneath the image. Automatic figure numbers and a generated table of figures are not included yet.
Formats
PNG, JPEG, GIF, BMP and SVG files are recognised. Raster images are embedded with their measured dimensions. SVG stays vector in current desktop versions of Word; older and third-party viewers may not display it correctly. WebP and AVIF conversion is not available yet.
Free and Pro limits
The free converter embeds up to ten images per document. Pro removes the limit and raises the overall file cap to 50 MB, which matters for reports full of screenshots. Sizing, centring and captions are identical on both tiers. Images inside Word tables are scaled to the cell width instead of the page width. For the rest of the document's inline styles, see Markdown to Word with formatting; if you convert with the command line, note that Pandoc embeds images at their pixel size and often overflows the page.