Digital Photos Give Away a Camera's Make and Model

holy_calamity writes "Engineers at Polytechnic University Brooklyn have discovered that digital snaps shorn of any metadata still reveal the make and model of camera used to take them. It is possible to work backwards from the relationships of neighboring pixel values in a shot to identify the model-specific demosaicing algorithm that combines red, green, and blue pixels on the sensor into color image pixels. Forensics teams are already licking their chops."

Comments

stretch?

[Speare]

As even the cellphones are producing 3 megapixel images now, very few people need to be passing full-resolution originals around. If you scale the image down to a screen-usable 1 megapixel image, there's not going to be a lot of bayer mosaicking information still available.

Re:stretch?

[narcberry]

Maybe so, but you still have the full size images on the camera. If someone were to get a hold of that they would be able to tell what camera took the images after a few weeks of intensive forensic study.

Re:stretch?

[Anonymous Coward]

Joke------>
          You

Re:stretch?

[trum4n]

Feed your RAW images to photoshop, then hit NTSC color mode, then compress to jpeg. All their secret information is gone forever. Only idiots would let this work. and oh yea, those idiots didn't delete the Meta-Data anyway, cause they dont even know what it is.

Re:stretch?

[lysergic.acid]

or you could just use a camera with a Foveon X3 sensor. there's no demosaicing involved since it employs 3 vertically stacked photodiodes (red, green, blue) at each pixel sensor to capture color information.

here is a diagram showing how a multijunction photosensor works. unlike bayer filter sensors, Foveon X3 sensors produce no color artifacts.

Re:stretch?

[Tinik]

But then wouldn't the lack of demosaicing itself be the tell-tale sign that it was taken with a Foveon X3 sensor?

Re:stretch?

[cunniff]

Eric: Horatio, there does not look like there is any mosaicing information in this image.
Horatio: Eric, that means the image was taken with a camera with a Foveon X3 sensor.
Cally: Zeroing in on professional camera stores... I have an address
Horatio (menacingly): Eric, get on it.

Where's my CSI: Miami royalty check?

Re:stretch?

[the_womble]

It does, but rather than narrowing it down to a particular model, it narrows it down to any digital SLR (and maybe other types) camera made by Sigma.

Add to that the fact that reducing the image size will probably get rid of the evidence, using a raw image and demosaicing on a PC will tell you what software was used instead of what camera was used, there are a lot of limitations.

On the other hand, most people do not know all this - then again, most people are unlikely to think of deleting the meta-data either.

Re:stretch?

[XSpud]

Seriously if you carried a suspect image around on a different camera you could now call an expert witness to show that it was not taken on this camera - and since the only other people who had access to the camera were the police it must be a frame.

Commonly called a "picture frame" I believe.

Re:stretch?

[jeffmeden]

You forgot the obligatory

Horatio: It looks like this photographer captured... himself
[Horatio puts sunglasses on over existing sunglasses]
[music: Yeeeeeaaaaaaaaaaaahhhhhhh]

What about multiple cameras using the same sensor?

[Chirs]

Quite often there are different manufacturers using the same sensor. Since this locks in the physical aspects of the sensor layout, I would expect the demosaicing algorithm to be basically identical across all these bodies.

Even after image manipulation?

[MR.Mic]

I wonder if this method still holds up after noise removal, or even something as simple as an image size reduction. Anyone more knowledgeable on the subject care to speak up?

Re:Even after image manipulation?

[morgan_greywolf]

I doubt it. I do a LOT of image processing, and I have to say that after color correction, noise removal, etc., I very much doubt that this technique would hold up.

Re:Even after image manipulation?

[jools33]

There are photodesk editors with experience who claim to be able to identify the camera model, the lens used and the post processing software used to produce the jpeg / tif file. This is possible due to various characteristics that are introduced at each stage in the digital photographic process. Giveaways for the camera body used are the base resolution, any colour casts in the image, iso performance, dynamic range. For instance many Canon DSLRs are criticised for producing muddy greens in their images - especially at higher ISOs as the dynamic range is pushed to extremes. Then you can usually work out the lens - by the obvious field of view first, then the flaws in the shot - various lenses from various manufacturers have different flaws. For instance the 70-200 f/2.8 from Nikon has characteristic vignetting that can often be noticed even after post processing, then other cheaper lenses give various defects to the image such as chromatic aberation. The flaws in the image give away the body and lens. Also the sensor used gives certain image characteristics that are fairly easy to spot even to the keen amateur photogs eye - for instance telling the difference between a full frame sensor and a smaller APS sized sensor - the full frame image typically has a much smoother more film like attributes with less digital artifacts. Also the same can be said for post processing. This software usually leaves various characteristics - that remain with the image, and this varies for each different software vendor.

So this is all possible to a well trained human eye - don't see why it shouldn't be possible in software - but not sure of the real benefits of being able to identify this - as many photogs often leave the exif data - and that tells you everything.

Raw images?

[evanbd]

So, if I shoot in raw mode, and then postprocess in software to get a jpeg, the demosaicing signature should merely identify the software, right?

Of course, I wouldn't be surprised to learn that the distortion uniquely identifies the lens used...

Re:Raw images?

[Facegarden]

...Of course, I wouldn't be surprised to learn that the distortion uniquely identifies the lens used...

Nah, not likely. Unless you knew exactly what the scene was supposed to look like, from that exact angle and everything (and even then it would likely be impossible), you just can't know what is a distortion from the lens and what is part of the scene. Unless, like, the scene happened to be a highly accurate checkerboard pattern. Then you can look and see what lines aren't quite straight and get some distortion information, but that would be tough.

I know software can correct for lens distortion if it has a distortion profile for a certain lens (which is probably made by shooting a checkerboard type pattern...), but knowing to move every pixel to the left one is a lot easier than knowing if every pixel was moved to the left one by the lens, if that makes any sense.

Put another way, it's easy to put soda in your mouth and have yellow stuff come out of your underbits, but very difficult to do the reverse.
-Taylor

Re:Raw images?

[Speare]

You really don't need anything so clean and nice as a full-scene checkerboard, to calculate a lot of lens details. Two or three moderate-length manmade straight lines that are at different angles should be enough. Like two edges of a table, a tall building, etc. That should be enough to give you the general curvature coefficients, which in turn would be pretty close to giving the right field of view. I don't think you'd be able to tell Sigma from Canon from Nikkor from Leica from Tokina from Zeiss glass.

Re:Raw images?

[AliasMarlowe]

So, if I shoot in raw mode, and then postprocess in software to get a jpeg, the demosaicing signature should merely identify the software, right?

Yes, they're just fingering the in-camera raw conversion to jpeg. Using external conversion changes the game.

There are comparisons of demosaicing algorithms used on the same raw image at several places on the net, such as http://www.rawtherapee.com/RAW_Compare/. The software can make a huge difference, especially regarding moire and related artefacts. Most of the raw converters default to a much too aggressive approach for small scale features, in my opinion. As a result they often create chromatic moire in the JPEG, and accentuate the problem further by sharpening (to hide the softness of typical cheap lenses). This is clearly seen in the examples at the linked site.

Identical detectors on different cameras usually differ in the optical antialias filter used, which can affect their susceptibility to moire on sharpening. This may leave some residual information to allow the camera to be identified even with external conversion of the raw image. It would first be necessary to identify the demosaicing algorithm/software, so identifying the camera just from residual artefacts from the antialias filter would not be easy.

Of course, I wouldn't be surprised to learn that the distortion uniquely identifies the lens used...

Raw processing packages such as Bibble Pro also include a database of distortion characteristics for many lenses, including zoom lenses across their zoom range. Optionally, the image processing can compensate for the distortion for any recognized lens. Of course, the removal of distortion may leave a signature, which will perhaps allow both the lens and the software to be identified. This would not be a trivial task, of course, since "identical" lenses differ in their optical characteristics, and probably none exactly matches its nominal profile.

Re:Raw images?

[Facegarden]

Put another way, it's easy to put soda in your mouth and have yellow stuff come out of your underbits, but very difficult to do the reverse.

I'm never drinking Mountain Dew again.

And you're probably better off. I think Mountain Dew is the byproduct of people who never quite realized that you can't make soda from pee... Well, unless you're NASA I suppose...
-Taylor

So What?

[Lord Apathy]

So what if they can identify the make and model of camera. I own a D70. There are 300 billion d70 out there. Good luck on tracking a picture to my camera.

Re:So What?

[canajin56]

They don't have to. All they have to do is say they used forensics just like on CSI, and it shows it was taken by the same kind of camera you own. Sure, you can say there are a lot of them out there so it proves nothing. But you know who else says that? The bad guys on CSI, and their smug, latte drinking lawyers. Always demanding warrants and to be released if they aren't being charged with anything! EVIL

Re:So What?

[John Hasler]

> Good luck on tracking a picture to my camera.

That's not the purpose. Knowing the photo was taken with a D70 eliminates all the zillions of cameras out there that aren't D70s. It's like knowing that a bank robber is a 6' tall blue-eyed blond male.

Re:So What?

[xIcemanx]

So what if they can identify the make and model of camera. I own a D70. There are 300 billion d70 out there. Good luck on tracking a picture to my camera.

RTFA:

While many people own the same camera models, Pollitt believes that this technique can still be used forensically. He says that because digital cameras have a shelf life of only 18 months, this can help to narrow down when and where it was sold. Just because it won't immediately narrow it down to a single suspect with perfect accuracy doesn't mean it won't be helpful in investigations.

Re:So What?

[taustin]

Forensically, it's as useful as saying "this bullet came out of this model pistol." Not conclusive by itself, but one piece of a larger puzzle.

99% of criminal investigation is eliminating who didn't do it, and this can be useful for that.

Re:So What?

[cayenne8]

"It's like knowing that a bank robber is a 6' tall blue-eyed blond male."

A bank robber with THAT description really would stand out.

Hardly any of them look like that from what I see on TV.

Re:So What?

[fuzzyfuzzyfungus]

I hope the "there are 300 billion of them out there" defence works better for you than it did for certain owners of cheap, common watches.

Re:So What?

[Anonymous Coward]

The bad guys on CSI, and their smug, latte drinking lawyers. Always demanding warrants and to be released if they aren't being charged with anything! EVIL

The only EVIL on CSI shows is the way the motherfucking cops use extortion.

"You don't want to give up privileged information on this guy? Fine, we'll be back with a warrant. Of course, we may have to dismantle your office for a couple of weeks to do a thorough search. What does a couple of weeks mean to your business? You do understand, don't you, that when we seize (God, how those bastards love the word "seize") your computer, our clumsy techs might return it with some important files no longer readable? So sorry. ... Oh, yes, ma'am, that's the perp we were inquiring about. Thank you for your cooperation."

Re:So What?

[vux984]

So what if they can identify the make and model of camera. I own a D70. There are 300 billion d70 out there. Good luck on tracking a picture to my camera.

Its useful the same way knowing the car that raced away from the scene of the crime used a particular tire, with a particular wheelbase. Or that a bullet was fired from a particular make of gun.

Neither will positively identify anyone, but if you were already a 'person of interest' in a long list of people peripherally related to a case that detail might put you on a MUCH shorter list if it comes up that you have that model. Plus its useful when they are asking a judge for a warrant. Judges really like specificity with warrants... A "We want to search his home and car for a Canon Powershot X"; he blogged here about buying a Canon Powershot X, he was caught on this surveillance tape leaving the scene carrying an indistinct object, the dimensions and shape of which are consistent with a Powershot X, and we know the photos in question were taken with a Canon Powershot X"... that's got a lot more weight than... "We want to search this guy for a digital camera, because a witness said he owns a camera, and he was caught on tape holding an indistinct smallish object which could be a camera, oh... and the photos we're interested were taken with a camera."

A reasonable person would view the second as a complete fishing expedition, based on no evidence, practically everyone has a camera and he could have been holding anything on that tape. The first request is specific - the photos of interest were taken with that model, and there is reason to beleive the person HAS that particular model, and that he had it with him on the that tape. Sure it could be a coincidence, but a warrant for that particular camera if he has one to check it out, might not be unreasonable.

Its also not unlikely that they can pair photos to a particular camera if they have both on hand due to micro-scrathes and other unique lens defects... the same way they can pair laser pritners to printed output.

Re:So What?

[Anonymous Coward]

And you know that 1x1 pixel is all the evidence we need.

"Chief! We only got a 1x1 pixel area of the jpeg of the criminal!" "Okay take it downstairs to Abby and McGeek to check it out".

[Wiggly Lines]

"Yeah Chiefy it was pretty easy. We zoomed in on the 1x1 pixel image and we got this picture of the criminal which was inside there. And we got the model of camera he used too, an Motorola RAZR Smartphone." "Yeah Chief, there's only been ninety jillion RAZR phones made, so we hashed the 1x1 pixel image and got the phone's SIM number out of it. It's him all right. Science never fails. We get the criminal every time, just like in real life." "Good work Abbs, have a cookie. And McGeek, get a haircut, you look like a hippy." "Thanks Chief". [Exeunt all, Abby munching cookie]

Re:So What?

[Anonymous Coward]

Looks like somebody tried contesting a speeding ticket and failed. ;)

Re:So What?

[mysidia]

Yes... CSI is very realistic in that regard.

Standard police tactic in real life, i'm sure.

Use fear/insinuation to get whatever info needed.

Except IRL it's not merely extortion, insinuated threats like that are all real.

Killjoy

[ciaohound]

Well, that kind of takes the fun out of this kind of story in which images from a Canon point-n-shoot are indistinguishable from those taken by a $40,000 Hasselblad.

Re:Killjoy

[Colonel Korn]

That isn't the conclusion that the author came to. If you compare http://www.luminous-landscape.com/images-85/h2.jpg and http://www.luminous-landscape.com/images-85/g10-comp.jpg there is definitely a different between the yellows and the depth of focus on the expensive camera is far better (compare the red leaves on the upper left).

Depth of field isn't a question of better or worse, you know, it's just different. If you want to poke at that story, you might just point out that a the limited image size makes the comparison pointless. These days, cheap digital cameras make incredibly expensive pro cameras more useful for either flexibility or niche markets (like >13" prints). That doesn't mean professional cameras aren't worth it, just that they're not worth it for everything.

Re:Killjoy

[johnlcallaway]

My wife and I have two point and shoot cameras, a Nikon and an Olympus. We also have a Pentax dSLR.

Looking at the images at 100% scale and you can see a tremendous difference in the amount of noise in the backgrounds. That is mostly caused by the smaller size of the CCDs and the quality of the sensor itself. Plus, the higher end cameras have far better noise reduction software built in.

Depth of field is EVERYTHING to taking pictures. By using a long lens and a large aperture, bars around zoo cages disappear, the annoying crowd behind the bride also disappears, or that person just standing behind your subject gets just the faintest blur so your eye is drawn to the subject. Or use a small aperture and everything is brought into crisp focus.

Then there is being able to use higher quality optics. I recently used the Pentax camera to take some campfire scenes using a 50mm(film) lens set at 1.4f. I was able to take clear, handheld images around the campfire. Try that with a point and shoot.

I'm not knock the PS cameras. I use them when I'm riding my motorcycle to get action shots of those I ride with. That would be impossible with a dSLR or SLR camera, they are just too big and bulky.

But if someone wants to take high quality snapshots to share, nothing beats a dSLR. Pricey, yes. But well worth it for the serious photographer, be they professional or hobbyist.

Re:Killjoy

[swillden]

Looking at the images at 100% scale and you can see a tremendous difference in the amount of noise in the backgrounds. That is mostly caused by the smaller size of the CCDs and the quality of the sensor itself. Plus, the higher end cameras have far better noise reduction software built in.

In addition to noise, the small digitals show lots of other image degradation as compared to a dSLR. They lack sharpness and have issues with distortion, color accuracy and chromatic aberration. The ultimate source of all of those issues is the glass. You simply can't get the same level of quality out of a half-inch lens as out of series of two-inch lenses (assuming similar technology applied to both).

I just shake my head at the ever-increasing megapixel numbers on compact digital cameras. I know they're great for marketing, but for the camera owners they do nothing but produce bigger files, with no better image quality than if they'd had a smaller pixel count. Once you get beyond the resolution of the glass, there's just no point in adding more pixels.

I just read the Haselblad/Canon comparison, though, and I have to point out that the Canon G10 is not what most people think of when you say "cheap digital". It's not an SLR, but it's close, with larger, better glass than most P&S cameras and a larger sensor. 15MP is a bit much for those lenses and that sensor size, but it's not that crazy.

Re:What about multiple cameras using the same sens

[fugu]

After the sensor takes the RAW data, the camera processes the image (some noise reduction, curves, and compression) to get a jpg. Since this conversion would vary between manufacturers (or even RAW software) I'd imagine that the process would leave behind similar "fingerprints."

Meta data?

[Ian Lamont]

Is the make and model data already included in jpeg's file header? See here

Re:Meta data?

[John Hasler]

> Of course someone who is stripping the exif data will never resize the image and run
> some sharpening over the image just to cover their traces, right?

Some will, some won't. Criminals are notoriously careless and stupid.

> Yep, this one was taken by a Canon Powershot A510 of which only 5.7 million were sold.
> We also know that this particular model was either sold in North America, Japan, Europe,
> Africa, Australia, South East Asia and South America. That should narrow it down.

Yes. Of the 18 initial suspects only two own that camera. Concentrate your investigation on them.

Re:Really...

[afidel]

Film can be identified down to the batch, MUCH more unique than a highly quality controlled part like a CMOS sensor.

Re:What about multiple cameras using the same sens

[qwertphobia]

You insensitive clod! I shoot my digitals in the raw!

Oh my!

[winkydink]

I better sell my Nikon D300. They'll be able to trace it back to me. Or one of the other gazillion people who also bought one. Hmm... on second thought...

Re:Oh my!

[winkydink]

No. Talking about my glass is showing off. :)

Re:Oh my!

[AgBullet]

I better sell my Nikon D300. They'll be able to trace it back to me. Or one of the other gazillion people who also bought one. Hmm... on second thought...

you're missing the point. there usually are other factors in play in any investigation. having only the camera make to go by is useless, but not so if you've managed to narrow - through other criteria - your suspect list down to 10.

Re:Oh my!

[ceoyoyo]

Or you could shoot RAW and then the camera doesn't do any demosaicing at all.

Re:What about multiple cameras using the same sens

[sjs132]

Ahh... That must of been you I saw at the park!

Good Gawd man, the poor lady didn't know what to think, haveing a bald, fat, nude man run up and take her holiday snap like that!

You left so sudden, she didn't get your number. For the picture, I believe.

Cherios....

This just in...

[fireman sam]

Cops didn't realize that most pictures posted on the interweb thing are usually post processed.

What about standard designs?

[EmbeddedJanitor]

People like Ericsson Mobile Platforms (http://en.wikipedia.org/wiki/Ericsson_Mobile_Platforms) provide the same design to multiple handset vendors. As the industry progresses we can expect to see growing commonality.

Rethink

[techdojo]

At first blush this struck me as similar to the printers that revealed a specific device by a faint set of dots printed on each piece of paper. On further thought, it occurs to me that the difference would be that the dot-tracking was shady where-as this is a triumph of statistical observation. The former being slimy and the latter sheer brilliance.

_________________________
http://techdojo.org/