A modern .xlsx workbook is a ZIP package, and Excel saves every embedded picture inside that package as a separate image file under a folder called xl/media. Extracting those pictures means recovering the original image bytes Excel itself stored under xl/media — without taking a screenshot of the sheet, redrawing the picture, or sending the workbook to a remote server. The Extract Images from Excel tool reads the .xlsx package locally in your browser, lists the supported media entries it finds, and lets you download each picture on its own or grab every recovered file in a single ZIP archive. Because the output is the original payload from the package rather than a redrawn canvas, transparency, color depth, and pixel dimensions stay exactly as the workbook author saved them. This approach works for product photos, logos, scans, header images, and any other picture that was placed into the workbook as a media file.
How Excel Stores Pictures Inside an .xlsx File
A modern Excel file with the .xlsx extension is an Office Open XML package — a ZIP archive with a specific folder layout defined by the ECMA-376 Office Open XML standard. Inside that archive, Excel keeps worksheets as XML parts, drawings as separate XML parts, shared styles, calculation chains, and — whenever a picture is embedded — one or more image files under a folder named xl/media.
A photo dropped onto a sheet, a logo placed in a header, a scanned signature, or a chart exported as a picture each becomes its own file in that folder, typically named with a common image-file extension and given a numbered identifier such as image1.png. The worksheet's drawing XML references the picture by relationship ID, but the picture itself sits in the package as an ordinary standalone file. That separation is what makes a faithful recovery possible — the visual is not generated on the fly from cell values; it is shipped inside the workbook as data.
This structure, also described in the Microsoft Open XML documentation, is the reason an image can be recovered without launching Excel or relying on a server-side converter. Reading the package, finding the entries that live under xl/media, and writing those entries back out to disk produces an exact copy of what the workbook author placed there in the first place.
Why a Screenshot Loses What the Original Has
A screenshot of a worksheet can feel like the fastest option, especially for a single picture, but it captures the rendered grid along with the image. Cell borders, grid lines, frozen panes, header text, and the current zoom level all become part of the picture. Transparency is flattened onto whatever background color the screen showed at that moment. If the image was larger than the visible window, the screenshot also crops it, and anything placed partially off-screen disappears entirely. Resizing the screenshot to fix those problems means recompressing pixels that were already compressed once.
Extracting the original payload avoids all of that. A PNG keeps its alpha channel, a JPG keeps its original compression settings, and pixel dimensions match what the workbook author saved. The extractor does not redraw the image on a canvas and does not convert it to another format — it copies the bytes from xl/media to your downloads folder unchanged. The downloaded file is the image payload from the package, not a screenshot, a re-export, or a converted copy.
For worksheets with many pictures, the gap gets larger. Right-clicking each image and choosing Save as Picture works for the first few, but it is slow, the resulting files often pick up Excel's compression artifacts, and any picture that has been cropped, recolored, or layered with shapes loses fidelity. Going straight to the package skips every one of those steps and produces files identical to what was placed into the workbook.
How to Extract Images From an Excel Workbook
The Extract Images from Excel tool follows the three-step Office Open XML extraction path described above. Here is the exact procedure:
- Open the tool and choose one .xlsx workbook from your device using the file picker.
- Wait while the browser opens the package locally, validates the ZIP directory, and lists the supported image files stored under xl/media.
- Download each original image individually for selective recovery, or use the download-all option to receive every detected picture in a single ZIP archive.
There is no account, no upload step, and no option to send the workbook anywhere. The browser reads the archive's central directory, then walks the matching entries under xl/media and offers them as downloads. JSZip is loaded only after you choose a file, so the page itself stays light until the workbook arrives. If the package contains no supported embedded media, the tool reports that clearly rather than inventing an image from a chart or a cell range.
What the Extractor Pulls Out and What It Leaves Behind
Office packaging does not make every visual object an image file. The extractor looks specifically at entries stored under xl/media, so what you receive depends on how each visual element was originally saved. The table below summarizes which common workbook visuals are recoverable as standalone files and which stay as XML instructions inside the package.
| Visual element | How it lives inside the .xlsx | Recovered as an original image file |
|---|---|---|
| Embedded photo (PNG, JPG, BMP, GIF) | Standalone file under xl/media | Yes |
| Embedded vector picture (EMF, WMF) | Standalone file under xl/media | Yes, original bytes preserved |
| Picture placed in a header or footer | Standalone file under xl/media, referenced from header XML | Yes |
| Image placed on a sheet but later removed | Standalone file still present in xl/media | Yes, if the media entry is still present |
| Native Excel chart | Drawing XML and chart XML parts, no media entry | No |
| Inserted shape, SmartArt, or text box | Drawing XML describing vector geometry | No |
| Conditional-format icon set | Style rules in the worksheet XML | No |
| Cell fill, pattern, or theme color | Style table referenced by the worksheet | No |
| External image linked by URL | Only a URL in the drawing XML, no media entry | No |
| Chart saved as a picture (exported from another app) | Standalone file under xl/media | Yes |
The honest rule is short: if the picture was saved into the package as a media entry, the tool can return the bytes. If the visual was described by drawing XML, kept only as a URL, or stored as a style rule, it does not appear in the extracted set, and the tool reports that instead of guessing. The extractor reports only the supported files it finds; it does not claim that the list proves current placement, reading order, image dimensions in a worksheet, or visible appearance after Excel applies cropping and effects. A package can also retain a media entry that is no longer visibly placed on a sheet, so the recovered set may include files that no longer correspond to anything on the worksheet surface.
Local Processing and Safety Boundaries
Because the workbook stays on your device, the only files that leave it are the ones you choose to download. The browser reads the .xlsx as a ZIP archive, checks the central directory before decompressing any entry, and then walks through the supported files under xl/media. The extractor enforces finite package limits on the number of directory entries and the amount of expanded data, so a workbook that is malformed, uses unsupported ZIP64 directory fields, or simply contains no supported media is rejected with an explanation rather than being partially unpacked.
That behavior protects against two common failure modes. The first is a damaged archive: a truncated or corrupted .xlsx can decompress into misleading file names or zero-byte entries, and the central-directory check is what catches that before any payload is written to disk. The second is unsafe inputs: archives designed to abuse path traversal, zip bombs, or oversized directories are stopped at the same gate. A normal workbook should not need thousands of directory entries or tens of megabytes of expanded data merely to recover a modest set of pictures.
The tool also draws a clear line on which workbooks it accepts. Modern Excel .xlsx files are supported. Legacy binary .xls files, password-protected packages, and workbooks whose contents are entirely XML instructions or external links are not converted or guessed at — they are reported as outside scope, with no partial output. The scope is intentionally narrow so the recovery remains honest about what it produced.
Working With the Recovered Images
Once the original media files are on your device, the next step usually depends on what the pictures were used for in the workbook. For a quick size sanity check before reuse, Image Compressor will shrink a JPG, PNG, or WebP in the browser and show the actual output size. If a specific pixel dimension is needed for a web page or thumbnail, Image Resizer sets exact dimensions without uploading the file.
For format conversion, WebP Converter switches between JPG, PNG, and WebP, and Image to PDF collects the recovered pictures into a single PDF with one image per page. To estimate how large a recovered picture will print, DPI Converter turns the pixel dimensions and a chosen DPI into inch and centimeter values without altering the file.
A simple worked example makes the ZIP estimate straightforward. Suppose a workbook contains 12 product photos, each averaging 250 KB on disk. The image data alone is 12 × 250 KB = 3,000 KB. A standard ZIP store-mode archive adds roughly 1 KB of local-header overhead per entry, plus a small central directory of about 1 KB total. Adding that overhead gives 3,000 KB + 12 KB + 1 KB ≈ 3,013 KB for the final archive — close to the sum of the individual files because PNG and JPG are already compressed and do not shrink further inside the ZIP. Downloading the images individually instead produces 12 files totaling exactly 3,000 KB with no archive wrapper.
One housekeeping point is worth keeping in mind. The extractor preserves image files, not their positions, captions, or relationships to the worksheet. Keep the source workbook if the original layout matters, and compare each recovered filename against the source sheet before deleting or republishing anything. For a careful handoff, download the originals, leave the workbook untouched, and verify that the picture count matches what was expected to recover.