It's common knowledge among graphic designers that they can rename a PDF file and import it into Illustrator and access the "layers" within the document, revealing hidden information.

I don't know enough about the binary image formats (jpeg, bmp, gifs), or the variety of applications that exist to edit or manipulate such images, but I think it's feasible for another image format (or tooling) to include historic edits.

One example where this is unwanted is when a user posts a screenshot and blacks-out confidential information such as server names, or IP addresses.

  1. Under what conditions can an anonymous viewer of "blacked out information" reveal private information?

  2. What is the correct way to "black out" information that will be posted on a public forum such as Stack Exchange?


5 Answers 5


Layers and history are not the same thing.

PDF is a vector format. Picture elements are geometric figures which are to be "painted" successively in a precise order, so that newer figures hide parts of older ones. This is not per se a trace of how the image was edited, only an artefact of the way things are drawn in PDF. It is not equivalent to typing Ctrl-Z on a Word document to perform some "undo" commands. But it still means that in the scenario you allude to (trying to "black out" some part of the picture), the naive editor may believe that he destroyed the information while he merely hid it.

(Anecdote: I did have, at some point, the job of "anonymizing" PostScript documents submitted by researchers in a review process. The submission guidelines specified quite clearly that manuscripts were not allowed to include any identifying element, but the grasp of mundane realities by researchers being what it is, quite a few of the manuscripts featured the name and institution address of the author. The solution involved adding an extra white rectangle over the offending area and editing the image so that the author name read as a bunch of 'X' even if the rectangle was removed.)

In bitmap formats, the picture is made of elementary pixels which are juxtaposed, and which do not hide each other. When you modify a bitmap picture by applying a black rectangle on them, the colour previously held by the pixels is forgotten, not merely hidden. There are a few bitmap formats which handle "layers" and thus could potentially have the issue you evoke (depending on how the black painting is done), but the common bitmap formats (including JPEG, PNG, BMP and GIF) do not have layers, and when you paint some pixels in black, information is gone for good.

With the GIMP, a free wannabe replacement for Photoshop, you can see the image layers, and when you save to a format which does not support multiple layers, you get an explicit warning that layers have to be "merged" -- which means that hidden pixels are eradicated.

  • Is PDF a graphic format? I thought it was a document format using vector, raster, text and other formats. By the same logic, we can discuss any document, file or data storage formats Oct 31, 2011 at 1:17
  • PDF has grown a bit. Historically, it was for "documents" as in "a bunch of pages that can be displayed and printed". It was not initially meant to be edited (as opposed to, say, a Word file). Nowadays, PDF can have editable fields and scripts. But so can SVG, with a 'G' as 'graphic'. I pretend that PDF is a "graphic format" as long as people use it as such, and they do (for instance, when I include a figure in a document I type with LaTeX, the figure is usually a PDF file).
    – Tom Leek
    Oct 31, 2011 at 13:28

Which graphic image formats contain revision history?

All graphic image formats contain information about the actual image. Most image formats have space in the format for additional information. Most of the space is for information related to displaying the image: image height, image width, colour palette, etc. In some formats there is space for information not essential to displaying the image, such as Operating System, time and date, or comment. Given the ability to store arbitrary text with the image, some programs may take advantage of the free space and use it to record a revision number, but generally most graphic programs do not do revision tracking in the standard formats (BMP, GIF, JPEG, PNG). JPEG and TIFF are compatible with the Exchangable Image File (Exif) specification which stores a lot of information with the image. The Exif format mostly specifies information on camera settings, but has been used to store geographic information.

Does the image editing software being used matter?

Yes. Image editors may use the 'free text' parts of the format for non-standard uses. For images with Exif metadata some programs my update the data transparently and some programs may give you the ability to edit the metadata directly.

1.Under what conditions can an anonymous viewer of "blacked out information" reveal private information?

The most common case of the black out part failing is in portable document format (PDF) files. See the UK's MoD recent blunder here. PDF documents are cable of having multiple layers. Instead of replacing the intended text with a black box, a inexperienced user may create a black box in a layer above the text, which will obscure the text when viewed, but does not remove the text from the document.

What is the correct way to "black out" information that will be posted on a public forum such as Stack Exchange?

The best way is to make a paper copy and physically remove the portion you want redacted (removed). i.e. cut the text or graphic out with scissors. Shred the cut out portion and scan the remainder of the document into an electronic format. PNG seems a good choice as it is not compatible with Exif. You can scan the image file with a string tool to see if any surrupticious references have been embedded. Have a trusted individual double check your work to be certain you remove all relevant information.

  • "Shred the cut out portion and scan the remainter of the document into an electronic format" which will add your personal or other tracking info? Oct 31, 2011 at 1:25
  • 1
    Thanks for pointing out that omission. I made a small addition. Unfortunatly there is no guaranteed way to ensure anonymity. As the question seemed to focus on the unintentional disclosure of information included in an image,I focused my response on that aspect.
    – this.josh
    Nov 1, 2011 at 6:57
  1. Under what conditions can an anonymous viewer of "blacked out information" reveal private information?

As others pointed out, one should not use vector-formats or complex formats with history/layer information, such as the ones that Photoshop/GIMP use to save their working state. JPEG/PNG/GIF etc. are fine for posting online.

  1. What is the correct way to "black out" information that will be posted on a public forum such as Stack Exchange?

"Blacking out" should really happen by making pixels black in a simple pixel-based file format such as PNG/JPEG. Do not use artistic effects like spatial distortion or (possibly slightly transparent) shadows. Then the image's meta-data should be purged(EXIF/XMP). Fancy image manipulation tools may add all kinds of information such as thumbnails, date and time or possibly even user/hostname. Publishing tools like Office/Adobe are famous for that. Also check for that when only printing a bitmap to pdf for "more convenient distribution".

Note that when distributing a disk, thumb drive or (encrypted) filesystem container to other people, additional data may be stored in Windows' hidden thumbnail and Trashbin folders, deleted content on disk or filesystem journal.


“Simple” graphic formats (like gif, jpg and so) that are usually used for screenshots don’t store any change history, so I don't see any problem, just make sure that you use solid color, no transparency.

  • 3
    But they can have thumbnail metadata that contains a copy of the unmodified image.
    – Steve
    Oct 26, 2011 at 17:43
  • Unless you disable thumbnails
    – this.josh
    Oct 26, 2011 at 23:30
  • What are "complex" graphic formats? Oct 31, 2011 at 1:05

Any graphic image format can and is used to contain any info. Check wiki's article on steganography

There were a lot of reports of robberies, blackmailing, stalking after people did put on the internet their photos, for example, of their cars in their garages made by mobile phones. Such photos contain embedded GPS coordinates that were easily used by criminals
I saw such discussions on twitter. Unfortunately twitter search is very lousy to find there something after a couple of weeks

Here is a first search results:
"Unknown dangers of sharing photos online and how to get protected" site

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .