paranoid people removes metadata from files when upload them to internet, but what if some companies hiddes user information inside data, for example in pictures could be an unique id per device in certain pixels.
Paranoid people removes metadata from files when upload them to internet...
Jonathan Fisher
Logan Mitchell
>what if
That's been done since before the internet and continues today.
Brody Allen
Camera pixel noise identifies you anyway
Nathan Miller
It's already being done. Printers embed their identifiers on every page. Of course it's very well hidden, but government agencies probably have all the documentation.
instructables.com
It's not too far-fetched to imagine the same being done with digital cameras too...
Levi Rivera
The thing is, it's rather easy to spot. The identification codes in colour printers were discovered, as was the watermarking of WoW screenshots.
For digital cameras, it isn't even necessary to do it deliberately. CCDs have slight process variation which makes it relatively easy to confirm that a photograph was taken with a specific camera.
Significant reductions in image quality or extensive post-processing will defeat this, but the same is true of any intentional watermarking scheme.
Brandon Brooks
Hudson Scott
now, I'm affraid.
Logan Anderson
why?
do you fap or record cp?
Jacob Lewis
Did you have a point?
Robert Walker
>The thing is, it's rather easy to spot.
If you look for it. There are a lot of different places to suspect and investigate, meaning it is unlikely you will be investigating the right product unless it is pretty much universal.
>as was the watermarking of WoW screenshots.
Didn't that go on for a year or two before it caught attention?
>CCDs have slight process variation which makes it relatively easy to confirm that a photograph was taken with a specific camera.
I'd like a source on that, I believe the variations but doubt how effective it is for linking a photograph to a camera.
>Significant reductions in image quality or extensive post-processing will defeat this
Watermarking isn't the only method of embedding extra data in an image file.
Landon Johnson
>unless it is pretty much universal.
It is.
>CCD source
google "CCD fingerprint"
Lincoln Scott
>>unless it is pretty much universal.
>It is.
Every widespread camera brand uses the same method of encoding hidden additional metadata in images? Perhaps you've missed my point. It may be near-universal for such hidden data to exist, but a particular method seems like it would not be, as that would defeat the point of the information being "hidden".
As for your google suggestion, I found mostly results for video cameras except for an analysis of 2700 images from 9 cameras. I think I should point out that needing that much data does not seem "relatively easy" - more like something that would be done over a month with a warrant for a federal case.
Nathaniel Anderson
no but what if someday I wanna make a voyerist robot and upload some caps to internet?
Ethan Myers
If you have not thought of a single other way that your identity could be compromised, don't worry, hidden metadata will not be necessary.
Cooper Peterson
>en.wikipedia.org
one time way back when i downloaded a jpg off of /v/, changed the extension to a rar, and opened it and it had all 700mb of halo ce inside
i will never fucking understand that witchcraft but goddamn was it cool
Liam Wilson
...
Asher Edwards
> Watermarking isn't the only method of embedding extra data in an image file.
A watermark is anything that's supposed to preserved in modified copies of the image. It doesn't necessarily refer to a "logo"-style watermark that's visible to the eye. E.g. any form of steganography can theoretically be used as a watermark (the stored data is simply a unique identifier).
Any mechanism for embedding "extra" data in the image is vulnerable to deterioration in the presence of common image-processing steps (cropping, resizing, lossy compression, gamma adjustment, etc).
Different techniques will have different degrees of vulnerability to different processes (e.g. techniques which survive cropping are more vulnerable to resizing and vice versa). Various error-correction algorithms can be used to mitigate this, at the cost of reduced storage, and subject to the fact that a given amount of error correction can only correct a given number of errors.
Josiah Gomez
but how did they even compress it that much
Justin Rivera
Explain yourself now
Ayden Jones
>A watermark is anything that's supposed to preserved in modified copies of the image.
Thanks for clarification of proper usage of the word, but given the context of my response:
>>Significant reductions in image quality or extensive post-processing will defeat this
>Watermarking isn't the only method of embedding extra data in an image file.
It should be clear that the person I was responding to seemed to be using the word with the intent of describing encoding additional information in the pixel data.
>Any mechanism for embedding "extra" data in the image is vulnerable to deterioration in the presence of common image-processing steps (cropping, resizing, lossy compression, gamma adjustment, etc).
If your editor is shitty/good (depending on how you look at it), or if the hidden information is entirely in pixel data. I wouldn't consider the case of complete loss of non-pixel data and non-standard metadata "deterioration" and it would not be recoverable. I fail to see how any of your examples would touch non-pixel data in exception of, as I suggested, the editing program deciding to drop the unknown bytes.
Easton Brooks
> Every widespread camera brand uses the same method of encoding hidden additional metadata in images?
I'm fairly sure that he's referring to the fact that process variation means that a CCD has a "fingerprint" which allows you to confirm or refute whether two images were taken with the same camera, rather than something that's added intentionally.
> an analysis of 2700 images from 9 cameras. I think I should point out that needing that much data does not seem "relatively easy"
You don't need that much data. That was the total size of their data set. They observe that the reliability of the reference pattern starts to drop once the number of images used to construct it drops below ~50. That may be an issue if your training set is limited to what someone chooses to post online; it's not an issue if you have possession of a camera and want to determine whether a specific image was taken with it.
FWIW, the "2700 images from 9 cameras" paper is "Digital Camera Identification from Sensor Noise", Lukas, Fridrich, and Goljan, Binghamton University. That was from 2006, so there's probably been some advances since then.
Fridrich's publication list has rather a lot on this subject, as well as a fair amount on steganography.
Caleb Thomas
I'm talking about watermarks and fingerprints, not metadata (e.g. EXIF).
Metadata is trivial to spot by virtue of the fact that it's metadata, which is either documented as part of the file format or (arguably even more obvious) a chunk of undocumented data.
It's also trivially removed, either by exporting to a format which doesn't support it or by removing any chunks other than those you wish to retain (typically the basic header containing image dimensions and format, and the pixel data).
Jack White
You can get arrested by doing this.
en.rocketnews24.com
Asher Jones
how is attatching a rar file to a jpg in any way related to someone being arrested for using a 3-d printer to create an illegal firearm?
Joseph Sanchez
you can attach the files to print the gun in a .jpg which is illegal in some countries.
Eli Ortiz
Wow, that article sure had a lot to do with embedding a file in an image.
This is completely tangential bullshit without citing a case of someone being arrested for embedding said file in an image WITHOUT printing it. Further, if said file is considered illegal, no fucking shit it's still illegal to hide it.
TLDR: thx 4 contribootin
Bentley Wilson
>image file
We're not talking about the files, but the image itself.
That is distinct.
Brody Perry
Just look it up
Henry Baker
just save everything as a jpeg
Samuel Ramirez
recursive data compression m8
all files are bit strings, just round the bit string to a small formula plus or minus correction, then repeat with the resulting bit string and so on until reaching desired size, can achieve ridiculous compression
the hard part is finding suitable formulas and corrections