blur text

[natalia rasem]

Is there any possibility to blur (deface) text. So that I can see later on, that there is something written but I cant't identify it...

[Will - Tracker Supp]

Hi Natalia,

Thanks for the post - this isn't something that is currently possible, but I will place a feature request for this. Do please note, however, that a feature request is not a promise to deliver, only a promise that the implementation of the feature will be given seriously consideration.

Cheers,

[skuffel]

Any update on whether this will ever be developed?

I create training documents and often have to blur out sensitive information on the screenshots I use in these documents.

[Dimitar - PDF-XChange]

Hello skuffel,

Welcome to our forum.


For that purpose you should use Mark for Redaction tool:

image.png

Regards.

[skuffel]

Thank you but redacting won't work. People need to see the area on the page my screenshot is referring to, but if there is a real name or address or PII there, it needs to be blurred out. Please consider this another request for a Blur or Smudge tool.

[Stefan - PDF-XChange]

Hello skuffel,

I am afraid that for the moment the complete removal of the text using the Redaction tool is the only solution we can offer.
Blurring a PDF is not as 'straight forward' as blurring an image - as there are multiple types of objects in multiple structures that can all be located in the same area of a PDF page, so blurring a section of a page is not really something that I see as being possible in the near future if at all.

Kind regards,
Stefan

[claude vidal]

Sorry for reviving this thread. I also often need to blur sensitive information.

I understood from Stefan's post that blurring the text itself would be too challenging.

PDFX already provides the ability to overlay a rectangle over text with color, transparency and blend mode parameters. Would it be possible to add to this a blend mode pattern that would make the underlying text unreadable?

[Daniel - PDF-XChange]

Hello, claude vidal

Unfortunately no, unless by unreadable you mean the text is deleted entirely, in which case, yes!
If you use the "redaction" tool to apply these rectangles, it will delete the content below it entirely, so it is not possible to recover, see, or lookup the text that used to be in that position ever again.

Kind regards,

[claude vidal]

I don't know the in's and out's of PDF, but I managed to create a PDFX stamp that, when applied over text, gives the following result.

PDFX sample blur.png

The pattern is crude and needs to be tweaked further, but I believe it proves the feasibility. The text underneath is still there, visible with a blur and accessible. Once the comments are flattened, it should be permanent.

Now, can this be done using the approach I suggested in my previous post:

"...PDFX already provides the ability to overlay a rectangle over text with color, transparency and blend mode parameters. Would it be possible to add to this a blend mode pattern that would make the underlying text unreadable? ..."

[Paul - PDF-XChange]

Hi Claude,

thanks for your added suggestion.

I must confess to being unclear on why we are being asked for blurring specifically rather than the redaction. If the objective is to prevent people reading the text that is "blurred" how is that different to redaction and removing the text? In the examples given the text its still available, it has a mask over it making it hard to read when rendered, but the text will still be there.

Can you explain why you want to obscure the text but not remove it? Obscuring it is easily overcome to access the content. I am failing to understand how it helps prevent people seeing the "blurred content" if they really want to see it.

Am I missing something?

[claude vidal]

... In the examples given the text its still available, it has a mask over it making it hard to read when rendered, but the text will still be there ...

After playing around a bit more, I realize now that the text can still be 'lifted' behind flattened comments when using 'Edit', an additional required step would be to convert the page to an image ... getting a bit complex.

... I must confess to being unclear on why we are being asked for blurring specifically rather than the redaction. If the objective is to prevent people reading the text that is "blurred" how is that different to redaction and removing the text? ...

'Blurred text' gives the appearance that it's the real text being blurred. 'Redacted text' is simply a black rectangle. It's all about visual perception honestly. I realize now that redacting actually deletes the text and replaces it with a black rectangle image. What if the redaction process could pass the text to a routine that would apply a blur to the text and return an image that the caller could apply, instead of the black rectangle?
_____________________________________________________________________

Finding - I often want to 'make non visible' a column of dollar figures. I found that I can redact an entire column of whatever in one operation by first pressing Ctrl while dragging the cursor, basically defining an area instead of doing it line by line.

[Paul - PDF-XChange]

Hi Claude,

I took this to the development team leader and he has agreed to a feature request to provide a "blur text" option to the Redaction Tool. The idea would be similar to how right now you can choose a colour for the block that goes where the content was, you would be able to choose "Blur content" (or whatever it ends up being called) and it would rasterize the text, blur it, then apply that image to the block.

I have raised a formal feature request around this, however please appreciate that this is a lot of work to do for a purely aesthetic/visual result and as such it is considered low priority. The ticket is for internal use only, however referring to RT#5972: Feature request :: Editor :: Blur text option for the redact tool. will help support get you a status report.

There is currently no ETA for delivery of the feature, it will be at the discretion of the development team leader.

I hope that helps.

[PHK]

...

I have raised a formal feature request around this, however please appreciate that this is a lot of work to do for a purely aesthetic/visual result and as such it is considered low priority. The ticket is for internal use only, however referring to RT#5972: Feature request :: Editor :: Blur text option for the redact tool. will help support get you a status report.

There is currently no ETA for delivery of the feature, it will be at the discretion of the development team leader.
...

I think this would be a useful addition to the application and I look forward to having such a tool available.

[Paul - PDF-XChange]

I was just informed that if we do this it requires a warning. The warning is that unlike true redaction where the content is removed from the document, blurring the content can actually be reverse engineered and the original text that was there can be discovered.

I will not go into the details of how but suffice to say that if the blur is a process applied to text, it can be, (with sufficient know how and effort) actually reversed, even with us removing the original text itself from the document. The blurred "version" is always something that can be reverse engineered.

Blurring text is not secure by it's nature. If it is important that the content absolutely not be available this option cannot provide that, you must use one of the existing options that removes the content and does not offer any "hints" as to what it was. A "blurred image" is a very strong "hint"...

[claude vidal]

I was just informed that if we do this it requires a warning. The warning is that unlike true redaction where the content is removed from the document, blurring the content can actually be reverse engineered and the original text that was there can be discovered....

I acknowledge the fair warning. I maintain that a blur option would be welcome, in adition to redaction.

[Stefan - PDF-XChange]

Hello claude vidal,

Thanks for the above!
The feature request is already in place, so our developers are aware of it and will consider it's actual future in due time!

Kind regards,
Stefan

[PHK]

... 'Redacted text' is simply a black rectangle. It's all about visual perception honestly. I realize now that redacting actually deletes the text and replaces it with a black rectangle image. What if the redaction process could pass the text to a routine that would apply a blur to the text and return an image that the caller could apply, instead of the black rectangle?
_____________________________________________________________________

Finding - I often want to 'make non visible' a column of dollar figures. I found that I can redact an entire column of whatever in one operation by first pressing Ctrl while dragging the cursor, basically defining an area instead of doing it line by line.

Of course, the selected rectangle does not have to show as black, it can be any color or even white.

[Stefan - PDF-XChange]

Hello PHK,

Thanks for the post!
Indeed - you can modify how the rectangles appears while you are marking areas to be redacted, as well as what the result should be once the redaction has been applied:

image.png

Kind regards,
Stefan

[DIV]

Just curious, for those who prefer the æsthetics of blurred text over the redaction 'rectangles' that are currently available, how would you feel about replacing the sensitive text with 'placeholder' text?
The placeholder text could be set to have a distinctive look (colour, bold, italic, etc. set by the user), while maintaining the same font and font-size as the original text.

For example:

"The Government accepted a bid of $2000 million for supply of submarine maintenance services, to be based in Timbuktu."
could become
"The Government accepted a bid of ********** for supply of ******************* services, to be based in *******."
or
"The Government accepted a bid of lorem ipsum for supply of lorem ipsum lorem ips services, to be based in lorem ip."

In such a model, it should be too difficult to offer the user the option of specifying their own placeholder text (repeated where necessary, as per the examples above).

Also, as a possible future enhancement, such placeholder text could be used as the template for a blurring operation. This would completely avoid the possibility of subsequent reconstruction of the original text: reverse engineering such blurred images could only recover the placeholder text. However, the æsthetics would be similar to that of the original proposal.

—DIV

[Stefan - PDF-XChange]

Hello DIV,

My post above yours shows exactly this - how you can specify your own placeholder text and change it's colour.
Maybe my choice of the word "redacted" makes it look as if this was automatic, but the option to update that to any other string is available through the properties of the redaction tool.

If I have misunderstood your post, please excuse me and would you mind explaining the main idea once again so that I can better grasp it?

Kind regards,
Stefan

[DIV]

I guess there was some miscommunication both ways, but no drama, Stefan.

In your example you have something that appears like a floating text-box in which the text filling the box bears absolutely no connection to the underlying text. Font size is different, line spacing is different, and (especially) your filled text appears in areas that (presumably) were originally blank.

That's quite different from my suggestion.

Remember that the original proposal was to blur the existing text. That text would not suddenly be created in previously empty regions, the line spacing would be respected, the font-size would be respected, even the influence of font style would be somewhat retained.
The same is broadly true for the existing (default) black redaction rectangles when used for a pure-text PDF (except for no influence of font style).
The same would be true for my suggestion when used for a pure-text PDF.

So I should qualify my suggestion by mentioning that my suggestion was (tacitly) only intended to be used for a pure-text PDF.
So it is very literally as per the example that I already gave:
"The Government accepted a bid of $2000 million for supply of submarine maintenance services, to be based in Timbuktu."
could become
"The Government accepted a bid of ********** for supply of ******************* services, to be based in *******."
or
"The Government accepted a bid of lorem ipsum for supply of lorem ipsum lorem ips services, to be based in lorem ip."

Think of censored curse-words in comic strips, "@?#!%".
https://www.merriam-webster.com/words-at-play/grawlix-symbols-swearing-comic-strips

So there should be (almost) a one-to-one correspondence between an original character and a replacement character (though I have to admit I didn't count in the above examples, and there are difficulties if replacing thin letters like "i" with wide letters like "m" or vice versa).


With your proposal (using existing functionality) it could look like

Code: Select all

                                 ⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿ
                                 ⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿ
The Government accepted a bid of ⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿ
                                 ⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿⁿ

(I've only formatted the above as 'code' to ensure alignment.)

I have checked the existing functionality, and for a pure-text PDF it would be possible in principle to create something like what I was proposing, although it requires the user to manually adjust the font and font-size (besides style and colour).
I have to admit I hadn't realised that this latter functionality was already possible. Nevertheless, what I was proposing would be/require a slight enhancement in terms of automatically choosing a matching/suitable font & font-size.

—DIV

[Stefan - PDF-XChange]

Hello DIV,

Sorry, I am not the sharpest tool in the shed today!
My screenshot was aiming to illustrate that you can redact out more than just the text, so I deliberately went outside of the text area.
You can of course use the redaction tool to "highlight" some text elements, and then the redaction would be applied to that area only.

I think developing the redaction tool to try and recognize what part of a redaction rectangle is text and blur that and add e.g. the suggested "******", and leave other areas 'plain' would be a bit too complicated. So it is up to the person doing the redaction to set his tool up with e.g. the "****" template, and set it to repeat, and then only select the actual text area.

Then again, applying a blur and text on top is what you are asking (even if the text on top is applied to the whole area), is again something that to me feels a bit too complicated, but I will leave it to our devs to decide. They will take into consideration this whole discussion when deciding on the feature request!

Kind regards,
Stefan

[DIV]

Hi, Stefan.

I think you are assuming my suggestion was more sophisticated than it really was

Basically, it was suggesting an automated method of what would be achieved manually by editing the text. Hence why it would only work for text documents (by which I mean not for scanned documents).

If I call it "overtyped text" would that make it clearer?
https://support.microsoft.com/en-us/office/type-over-text-in-word-for-windows-62c15c48-0936-4902-affe-4cadd71b7038
Think along the lines of subtitles on TV (perhaps for a 'reality' show where people are whispering, perhaps for a translation of a someone's speech on the news): "We will not put up with that kind of s***!"

As will hopefully be clear from the added description, and the attached sample,

PDF_overtype-versus-redaction_RevA.PDFA.pdf

blurring overtyped text would be no more difficult than blurring the original text. But also, at that point, any blurring added to the overtyped text would be purely for cosmetic reasons, because the content would have already been utterly obscured.

Anyway, it's not really something I need myself much as a priority. I was rather throwing the idea out there to see what those asking for the blur feature would think of it.

—DIV

[Stefan - PDF-XChange]

Hello DIV,

I understand it now.
So you effectively want to do it as a two step process -
1) Apply the redaction with a replacement text
2) Blur the replaced text (so even if someone reverse engineers this - they would still not get any viable information).

I do understand now what you mean - but then again - this will likely overcomplicate things!
I will include all of this discussion in the ticket and let our devs have the final word on the actual implementation, if such a blurring feature is at all approved.

Kind regards,
Stefan

[DIV]

So you effectively want to do it as a two step process -

Hi, Stefan.

Yes & no.

Actually I was mainly raising the option of applying "step 1" only. And asking those requesting a difficult & 'hackable' enhancement to Editor's functionality (i.e. text blurring) whether they would possibly be happy instead with a simpler & more robust enhancement to Editor's functionality, viz. "Apply the redaction with a replacement text". Why?

• redaction by replacing text (proposed feature) may be simpler to implement than text blurring (proposed feature);
• redaction by replacing text (proposed feature) cannot be reversed by recipients of the file, unlike text blurring (proposed feature); and
• redaction by replacing text (proposed feature) might, perhaps, be considered to have superior æsthetics to redaction with overlaid plain/patterned rectangles (existing feature).

The optional extension was to say that if this "step 1" (redaction by replacing text) were indeed implemented as a new feature in Editor, then it would provide a potential foundation for a more robust implementation of "step 2" (text blurring), if that were still considered useful by those who raised the original proposal in this thread.

—DIV

[Paul - PDF-XChange]

Hi all,

the current "Find and Redact" when choosing text to replace the redacted content doesn't always match nicely anyway, so using that to get a similar blur to the redacted text isn't going to be easy and I suspect that may be an influencing factor.

I just tried replacing "electric" with "chicken" and the results are not ones that would lend themselves to a nice blur that in any way simulates the original text:

image.png

I added a note to the ticket that there is an ongoing discussion here about this so hopefully these comments will be taken into consideration as the team develop whatever is to come from this.

We'll have to wait and see what Santa brings us I expect.