Remove Images from PDF creates a no-upload copy of your PDF where eligible embedded JPG image objects are replaced by a 1-by-1 white JPEG, so you can strip a photograph from a PDF in Adobe Reader without uploading it to Adobe's servers or paying for Acrobat Pro. Adobe Reader (the free viewer) does not expose an image-delete control at all, and the Edit Images panel inside Acrobat Pro is gated behind a subscription that also routes your file through Adobe's cloud when you use the web version. If the document is a draft, an internal report, a portfolio page, or anything containing a face, an address, or a contract, that upload is the part most readers want to skip. The catch is precision: image deletion in a PDF is not the same as deleting a picture in a Word document. A PDF image is an indirect object inside the page's resource dictionary, called by name from the page content stream, and removing it the wrong way can leave the next viewer reporting a damaged file. Remove Images from PDF avoids that by keeping the reference alive and swapping only the image stream, which is why it works as an Adobe-Reader-friendly substitute for the Edit Images panel. Below is exactly how to run it, what it will and will not remove, and how to confirm the output is safe to share.
Why Adobe Reader Alone Cannot Remove a PDF Image
Adobe Reader can open, sign, comment on, and fill a PDF, but its editing toolbar does not expose an image-deletion control. Adobe reserves the full Edit > Edit Images workflow for Acrobat Standard or Pro, and the Acrobat web version uploads your document to Adobe's servers before any change is saved. That upload is the part most readers want to avoid when the document is a draft, an internal report, a portfolio page, or anything containing a face, an address, or a contract.
A PDF image is not stored the way an image sits in a Word file. A picture embedded in a Word document is a single object the editor can delete with one command. A picture embedded in a PDF is an indirect object inside the page's resource dictionary, called by name from the page content stream. If you delete the object but leave the drawing command in place, the next time the file is opened the PDF viewer reports a missing resource and the page may render incorrectly, show an error toast, or refuse to open at all. That is why naive PDF image editors either silently corrupt the file or warn that the document is now damaged. The Remove Images from PDF tool is built around that hazard and refuses the job unless it can replace every image safely in one pass.
How the Tool Replaces Images in the PDF
The Remove Images from PDF tool takes a different path from most editors. Instead of trying to surgically remove one image from a multi-image page, it inspects every embedded image object in your document and checks whether each one meets a strict safety rule. The rule is published in the tool's contract: every embedded image must be an unmasked 8-bit DCTDecode JPEG using DeviceRGB or DeviceGray, with no DecodeParms dictionary. DCTDecode is the standard JPEG encoding used by most phone cameras, scanners, and basic image editors when the picture is placed into a PDF.
If even one image uses another filter, a mask, an indexed palette, CMYK colour, JPX (JPEG 2000), JBIG2, LZW, or another unsupported form, the tool stops before saving anything. This all-or-nothing rule matters because it prevents the worst kind of bug: a download button that says images were removed while a complex picture stays on the page. When every image passes the check, the tool generates a 1-by-1 white JPEG in your browser, walks every eligible image object, and assigns a replacement PDFRawStream to each existing reference. The page drawing command and its resource reference stay exactly where they were, so the page count, text layer, links, and any vector artwork remain valid. The PDFRawStream handling in pdf-lib is the underlying mechanism that lets the tool keep the reference valid while swapping only the bytes.
Remove an Image From a PDF in Adobe Reader Locally
Use the three-step procedure below inside any modern browser. You do not need Adobe Reader open while the tool runs, but you will want it (or any PDF viewer) to inspect the output before you save.
- Choose one local PDF up to 25 MB on the Remove Images from PDF page.
- Select Remove eligible images to start the all-or-nothing compatibility check on every embedded image object.
- Open the resulting file in Adobe Reader or another viewer, scroll every page to confirm the white replacements are where you want them, then download the cleaned file.
If the tool reports that one image could not be handled safely, no output is produced. Treat that as a feature: the alternative would be a partial file with a misleading "images removed" message. Re-encode the unsupported image as a plain 8-bit DeviceRGB or DeviceGray JPEG without a mask in a basic editor, replace it inside the source PDF, and try again.
Which Image Types the Tool Will and Will Not Replace
The compatibility table below is the source of truth for what the tool accepts. If your PDF contains an image outside this list, the whole job is refused.
| Image property | Eligible (will be replaced) | Refused (whole job stops) |
|---|---|---|
| Encoding filter | DCTDecode (JPEG) | JPX (JPEG 2000), JBIG2, LZW, Flate, CCITT, RunLength, any non-DCTDecode filter |
| Colour space | DeviceRGB, DeviceGray | DeviceCMYK, Indexed (palette), ICCBased, CalRGB with attached profile |
| Bit depth | 8 bits per channel | 1, 2, 4, 12, or 16 bits per channel |
| Masks | None | Soft mask, stencil mask, SMask, or any explicit /Mask entry |
| DecodeParms | Absent | Any DecodeParms dictionary |
| Placement | Regular XObject image referenced from the page | Inline images drawn directly in the content stream |
What Stays in the Output and What Goes Away
The table below summarises how the rest of the document is treated. The file is rewritten, not edited in place, so the structure is preserved but the bytes are new.
| Property | Behaviour in the output |
|---|---|
| Page count | Same as the source |
| Page geometry (size, rotation) | Same as the source |
| Text layer (selectable text) | Same as the source, not rasterised |
| Links and annotations | Same as the source |
| Vector shapes (paths, lines, fills) | Same as the source |
| Embedded JPG image streams | Replaced with a 1-by-1 white JPEG at the same object reference |
| PDF document metadata (Title, Author, Producer) | Carried over from the source |
| Original image payload in the file bytes | No longer used for visible rendering; bytes may remain on disk until the file is overwritten |
| Password protection, digital signatures | Rejected on input; the tool does not bypass a password or preserve a signature |
Inspecting the Result and Verifying the Output
After the tool finishes, it does two structural checks before exposing the download. It reopens the saved file with pdf-lib to confirm the page count has not changed, and it opens the file with the same PDF.js viewer used elsewhere on the site to confirm that the expected number of image objects still exist as valid replacements. Neither check is a complete validator, so you still need to do the human pass.
Open the result in Adobe Reader and flip through every page, not just the first one. Look at each former image position: if a white rectangle now sits where the picture used to be, the replacement took effect. Click any text that surrounded the image and confirm it is still selectable, not flattened into pixels. Click any link that used to sit on or near the image and confirm it still navigates. If your PDF uses non-Latin text, verify a sample glyphs renders the same as in the source, since the tool does not touch font programs. If any of these checks fails, the safest move is to keep the original file and try a different cleanup path.
When a White Replacement Is Not Enough
The tool is not a redaction product. A white rectangle can be covered by overlapping page content such as a sticky-note annotation, a watermark layer, or a form-field background. The original image bytes may still be recoverable from the file with a forensic tool until the file is securely overwritten, and PDF metadata that named the original author, the scanner, or the software that placed the image is not automatically scrubbed. Encrypted, malformed, signed, or unusual PDFs are rejected on input rather than silently coerced, so do not expect the tool to bypass a password or certify a legal-grade clean.
If your goal is court-grade redaction or compliance-grade privacy, use a dedicated redaction workflow that paints over the image with a real opaque object, flattens the result, and strips the metadata in the same pass. Keep the source file, work on a copy, and treat the white replacement only as a fast scrub when the result is going to an audience that does not need forensic certainty. For a deeper walk-through of how the same tool handles tricky layouts, see the guide on removing a picture from a PDF locally.
For a deeper look, see How to Remove a PDF Password on iPhone Without an App.