HOWTO spot a photoshop job
Link (Thanks, Barry!)
Surrounding lights reflect in eyes to form small white dots called specular highlights. The shape, color and location of these highlights tell us quite a bit about the lighting.In 2006 a photo editor contacted me about a picture of American Idol stars that was scheduled for publication in his magazine (above). The specular highlights were quite different (insets).
The highlight position indicates where the light source is located (above left). As the direction to the light source (yellow arrow) moves from left to right, so do the specular highlights.
The highlights in the American Idol picture are so inconsistent that visual inspection is enough to infer the photograph has been doctored. Many cases, however, require a mathematical analysis. To determine light position precisely requires taking into account the shape of the eye and the relative orientation between the eye, camera and light. The orientation matters because eyes are not perfect spheres: the clear covering of the iris, or cornea, protrudes, which we model in software as a sphere whose center is offset from the center of the whites of the eye, or sclera (above right).
the latest
latest episodes
Hahah awesome picture, you don't really need to go to eye level detail to see it's a photoshop!
I've been visiting this site every day, you'll love it http://photoshopdisasters.blogspot.com/
Yikes, I agree with Mauric. You can spot this as a fake from the first glance, no specular highlights needed. Look at Ryan looking like he was copied and pasted behind the others.
Ben: digitally signing an image doesn't guarantee that it hasn't been tampered with before the signature was applied. It only detects modification after that point. Also, I don't know why you think a "voice signature" is special in any way.
So, Randy and Paula were probably really there - their light sources match, and the posing interacts a lot more with the environment.
I would be *really* surprised if the above photo was untouched, even after a quick glance at the size on this page.
It wouldn't surprise me if the guy in the middle-back wasn't even in the same city when this was comped together.
Yet it fails to bother me at all. This isn't journalism we're talking about here, it's not even art. It's commercial advertising. There's no integrity expected or required.
The other day I was shooting four people in my studio, one-by-one side-lit in front of a black backdrop to make an image that will be comped together in photoshop. Of course I make sure the catchlights are all the same, for a closeup like I'm doing. If they're not, I'll clone in the correct ones.
Ethically, the way to determine if using photoshop to change or create some aspect of an image is to ask yourself if a drawing or painting would be appropriate in the same spot. If so, (advertising, illustration,) then go ahead. If not, (news, journalism,) then no.
Hany Farid, the Dartmouth researcher who wrote this article, was recently on our public radio show. You can listen to him here, if you like. His work with digital forensics really is pretty incredible. We also talked to him a bit about the ethical aspects, like the blurring line between "photos" and "photo illustrations", and whether society's expectations are changing.
You could have the original photographer attach a digital signature with all the photographers public keys in an open database. Thus, this makes sure that only the photographer has an opportunity to tamper with the picture.
Some camera makers, (Canon for one,) have the technology to sign the image in the camera, preventing even the photographer from digitally-altering the image.
They do this so that the images can be introduced in court as evidence.
So, to sum up, you really can "tell by some of the pixels, and from having seen quite a few 'shops in [your] time."
Don't we all already assume that everything we ever see or hear made by anyone involved in the entertainment industry (or any other form of commerce for that manner) has been manipulated and that nothing may ever, ever be taken at face value?
The importance of getting the eye-reflections right may be even deeper and more interesting than Photoshoppers think. The philosopher and phenomenologist Maurice Merleau-Ponty wrote about the different ways eye-reflections are perceived, and he wrote about the impact of eye-reflections on our experiences of faces.
According to Merleau-Ponty, there are different ways of perceiving an eye-reflection, and when we perceive an eye-reflection "incidentally" (as a typical viewer of a photograph might) it fosters our experience of the expressiveness of the face as a whole. He has an interesting way to understand why it is that, if a Photoshopper gets the reflection wrong, the whole facial expression looks weird. Rather than take up too much space in this comment section, for those who are interested, I'll post some Merleau-Ponty quotes and explain them here.
I agree, it's very easy to tell this is photoshopped without looking at the eyes. But what's the problem? As Jim states in #6, this is clearly just a PR photo, not journalism. If they weren't able to get everyone there for the shoot at the same time and didn't have time to wait, photoshop is the next best thing -- that's one of the reasons we have it!
I use photoshop all the time at my work for much the same reason; I handle publications and we don't always have time to re-shoot photos before a deadline. If something or someone is missing (or isn't supposed to be in the photo), I have no problem retouching it to add or subtract. While that may be ethically wrong for a news journal or National Geographic or the like, it's no biggie at all for our purposes. We're not trying to pull the wool over anyone's eyes, just trying to have the photos and our publications look as good as they possibly can.
Speaking of eyes, if I have to re-do a photo for some reason -- and I'll say right now that I work primarily with photos of animals -- I am very careful with how I place the highlights in the eyes because I don't want them too look weird. That just seems a like common sense rule. In the ten years I've been doing this, I have rarely had anyone notice say something looked odd with any of the photos I've retouched.
But I appreciate the tips above for making them look even better! :o)
One of the most shocking of all photoshoped images
as true ,was that aircraft( we were lead to believe)
behind the tourists posing on top of the NYC World Trade center at the moment of attack.
Though false it still sends chills up my spine as
I was still smelling the fires of that attack in
my neighborhood close by.
We can of course copy an encoded digital high resolution print on my hi res scanner and avoid the Canon Company signature and then re-manipulate the image.
I don't think we can ever avoid these technological
fabrications when they are well done.
It worked decades ago for the music industry in making everyone from pianist to ballad
singer in perfect pitch and precision as we would "tape splice" a "live"classical performance (in the vinyl age) mixing studio touch ups onto a live location recording or mixing several days of performance as one complete continuous concert.
Who was to know?
For you AI buffs, you can use PCA to determine if an image was altered. Should be a cinch to throw something together in Matlab.
What the hell is going on with Simon Cowell's thumb? He looks way too dull in general too. I would be lying if I said that I noticed it was 'shopped after a cursory glance though.
@16 XENU
Thanks for that tip! Since OpenCV already has functions for PCA and what-not, it almost sounds like a 15-minute hacking exercise..... :)
This looks shopped, I can tell from some of the pixels and from seeing quite a few shops in my time... ;)
@#9: Thanks, Jim, I hadn't known about that.
@#15 - it's not quite that simple, since you still have to claim that the photo came from a particular camera, and the photo has to match that camera. That involves faking the camera's digital signature, as well as it's unique "fingerprint" (see this Bruce Schneier article on this. Nonetheless, this is quite doable, as is faking the Canon digital signature (see here - search within for the comment by "Dan Schwartz".
@#16: I can understand how PCA could be used for watermarking. (There are a few journal papers around describing this, and even a free 3D graphics program that provides the option.
But, how could you go about determining if a photo was manipulated just by PCA? (short of having the original copy).
@ SC Wolf: damn it, you beat me to it. I need to up my game, I assumed I'd be the only one who knew the sacred memes here (or who would bother to post them).
Was it not enough that the think just looks fake as hell as is? Rofl...
I really don't care if a picture is photoshopped
or not. I care about whether I like the way it
looks. I am impressed by creative use of photo-
chemical methods of producing pictures, but using
Gimp or Photoshop is a skill in it's own right.
As long at it's not represented as photojournalism,
who cares?
Why should we care, you ask? If you're not aware of how a picture can be manipulated, that makes it all the more likely that someone will try to pass off an altered picture as photojournalism. You'll recall that picture in 2006 of the Israeli attacks on a Hezbollah rocket site (and surrounding apartment blocks) that the photojournalist thought would look more impressive with added smoke and fire.
Then there's the grey area between reportage and publicity photo illustrated in the SciAm article linked above by the Bush rally with its cloned soldiers. I would have said that picture should have been regarded as reportage and not manipulated. Whoever did it obviously disagreed. With that particular picture there's not much harm done, but having done that one, how easy would it have been to manipulate something more important, if it hadn't been discovered?
The sad truth is, the skills and tools are out there, which means they will get used for less than pure ends. The more the general public are aware of these tricks, the less likely they are to succeed.
in this photo, there is one glaring reason why this is a photoshop: ryan seacrest is bigger than simon, whereas in real life, seacrest is only about 3'4
Check out the recent New Yorker article about the master of photochopping. Great stories about tweaking models for major magazines.
Hummm, Paula looks like she has 6 fingers too
I really couldn't care less if i know a picture is photoshopped or not, even though it's very easy to tell. Look at Randy's neck and Simon's arm...since when to people's arms have a big black spot on them?
But even if Paula IS photoshopped or not, I'd still do her...
the eyes in photos tell everything about lighting placement... the same goes for space-suit helmet visors and how they reveal the studio light sources on alledged moon walks.
Simon looks like a freaking pasteboard
One of the things I always took for granted as an artist was my ability to spot the fakes in videos with special effects. I held onto this until about the first Jurassic Park movie (first time a dinosaur was shown), then thought all was lost until the first Harry Potter (the bricks folding back in Diagon Alley) -- where I then realized that areas in altered images (even those carefully cloned through PhotoShop and PhotoShop like methods) do not have the same Red/Yellow/Blue pixal mix as their un-altered counterparts. To sum the run on: Just tweak the color blend on your home television/computer so that it is either too red/yellow/blue and those sneakily altered digital patches should show up like huge dead-zones. In closing however, I must say that I been a been nurturing classical artist tendencies with a good sense of color since I was age five, so it always does amaze me how people who do not have the same priviliages can be tricked by sloppy jobs. (Of course, this also makes me wonder what frighteningly professional jobs out there got passed off as 'facts' I now believe wholeheartedly to be true . . . :(
As a commercial advertising designer, I take offense to the notion that integrity is neither expected or required. Some of us are quite ethical, and just because you can make something better in photoshop doesn't mean you should...
I can tell you what exactly is photoshoped just by looking at it. The row of seats that all the judges are either sitting on or leaning on are one image. Ryan Seacrest is another image and the rows behind all of them are another image making up the background. 3 Layers at least from that size and quality. If it were larger and I could zoom in, I could probably identify more. I've done my share of high end retouching though.
Uhh kids, I don't need to check out the pixels to know it's been pshopped. Just one look at Simon's thumb tells me that!
I just have one question...WHO CARES!!! If the image has been doctored so what. If not, so what again. What boggles me is the fact that people care SO much about celebrities personal lifes that they would actually care whether or not, oh I don't know, Jen and ...whoever her current boy toy, were actually seen together or if the image had been faked to create a story. These people are here to entertain us, by acting in films and tv sets only. Get over it!
The only time I have a problem with doctored photos is when its actually IN a movie. Like the opening for Gone in 60 Seconds, for example. When its showing the family photos. Grace Zabriskie's head is so much bigger than the body it is placed on. Whoever was in charge of visual props on this film did a half ass job.
I clicked on the artical because I thought that it was going to be something interesting...not something that everyone in the free world already knows.
And if you want my opinion on the actual topic. Randy and Paula were both actually there when the picture was taken. Simon and Ryan are photoshopped in. Both from seperate pictures. Guaranteed.
Re: post #24....why in articles about the war do we need pictures? Why do we need to see the guns firing to understand what is going on? Because seeing is believing these days isn't it. You can't trust what you read or what someone says, only what we can see with our own eyes. And our world is so far gone that people have done everything possible to tell a lie. Even in a picture, that not too long ago wasn't important in order to believe or understand a story. When the radio was all we had. When we could trust the celebrities and the media and our political leaders. I guess, we're not so blessed anymore.
Jewellsomm,
Um. The picture is just an example. I don't think that anyone is suggesting that this tutorial exists solely to help people analyze celebrity photos.
BoingBoing made the front page of Yahoo with this one...
thnk y cld phtshp lttl nc vw nsd pl's skrt? tks t ll th gsswrk nd hrdwrk f mgntn!
shrn stn? bsc nstnct? ymm!
@#7 Jim:
So which is it, commercial advertising or journalism? People who think there is no integrity required in journalism have a problem.
Even commercial advertising requires some modicum of integrity to the extent that we should not endeavor to mislead our audience.
So I guess it's not just about journalism after all. Integrity should simply be a fact of life.
i don't really know that this information on photoshop is really that useful, i mean come on! the picture they used, lol, is completely fake. I think it's a pretty funny photoshop picture.
i guess there are a couple of useful pointers here, but when you are photoshop a funny celebrity or politician, like the ones seen here: http://www.phunnyphotoshop.com , no one is really looking for these kind of details. they just want it to be funny, photoshop edited or not.
Paula, Paula, Paula...cover up, PLEASE! You are making me ill. I can not believe that you are even a judge any more on American Idol. You try to grab all of the spotlight, I just want to say PLEASE SIT DOWN.
I think most of you are missing the point! I don't think point of the article was that this is a fake photo...it was a how-to-spot a photoshopped photo and this was the example they used.
Um. This really isn't a thread about American Idol. Really. It's about detecting photoshopped pics. Really.
So does anyone know what happens if you are using more than one light source? When you light a stage you use multiple lights to create an even blend across the stage. That means that some people are only lit by one light while others - on the edges of a light beam are lit by two. Could that account for the double specular highlights? Not to say this isn't photoshopped, but if the lighting is artificial it seems to me that there could be variation.