It began the way pranks sometimes start these days, as a post to an e-mail discussion group, Q-Skeptics in this case.  The message was along the lines of: I’ve got this photo (attached), and it looks very much like there is a ghost in the photo.  The photo was taken by the mother of the girl in the photo, using a friend’s mobile phone. There were independent witnesses present who saw nothing at the time, but can verify that the photo was taken innocently.
Being good skeptics, we shouldn’t immediately jump to conclusions, but investigate whilst casting doubt.  I’m often reminded that skepticism is a process, not a position.  In practice it is, but in reality it’s difficult.  Anyone like myself, with an interest in photography and a somewhat skeptical attitude, is frequently sent photos containing all kinds of accidental anomalies and challenged with “What do you make of that?” Very tiresome.

The posted photo was worth a second look: the ghostly effect was quite compelling, although the image was small and unclear.  Almost all other purported ghost photos I’ve had sent to me were of the usual orbs.
Orbs
If you’re not aware of the orb effect, it’s where floating dust reflects light from the on-camera flash, back into the lens.  These dust particles create ghostly out-of-focus blobs in the final image, and are the same shape as the camera aperture.

This effect is explained in great detail on many skeptically oriented sites, so there’s not much point in my going over it again.
In the case of the photo posted to Q-Skeptics, it clearly wasn’t dust orbs, but a noticeable human-like form.
Film Photography vs Digital
Not so long ago, film photography produced its fair share of anomalistic photographs, arising from dust orbs, fingers, camera straps close to the lens, accidental double exposures etc.  Deliberate fraud was also a problem, but a little more difficult to achieve.

With digital photography, deliberately putting fake ghosts in photographs is easier than sticking a bed sheet on your head.  Photo manipulation with software is pretty much the first thing we think of when faced with a purported ghost photo.
Was it possible that the background story in this case was accurate?  An innocent person taking a photo with a mobile phone? That’s the way the story was presented.
It’s the background story that’s important in this case.  You’d look at a ghost photo differently if you knew it came from someone proficient at fiddling digital images.
Almost straight away the Photoshop solution was suggested by another poster to the Q-Skeptics message board.  (Not by me, but more on this subject later.)  Then another, even more intriguing suggestion from someone else: “Was it the ghost app on an iPhone?”
Ghost App?
Wow! That was news to me, a “ghost app”.  How interesting.
I can imagine the market for such a thing.   A tech-savvy prankster with an expensive new iPhone wants to show it off.  What better way? Freak out your unsuspecting credulous relatives.  “Hey Gran, can I take your photo on my new iPhone?” Click… “Oh look, what’s that ghost-like thing sitting next to you on the couch?” (cue: shriek from Gran)
The nice thing from the prankster’s point of view is that it’s all so easy.  The photo manipulation software that puts the fake ghost in the photo is on the phone – fake spirit of lost soul in an electronic trice, who’d have thought?
With the iPhone/ghost app suggestion posted to the message board, the question now became one about the original poster: was the photo taken on an iPhone?
Several days passed, and I decided to do some investigation myself.
There is a clever method of confirming a story associated with a photograph.  Let me explain how it’s possible to let a digital image tell its own story.  Forget the image itself; use the back door method, the EXIF data.  I’ll now briefly digress from ghost stories to give the slightly tedious explanation of EXIF data.  Hopefully you’ll stay with me: it’ll be worth it.
EXIF data
Photo image files such as JPEG files are jam-packed with information.  They are deliberately compressed data, so much so, that the method of compression actually subtly modifies the original image in the effort to make the image file smaller. This is known as “lossy compression”. However the changes are so slight that people generally don’t notice.
The data within an image file can be very repetitive and redundant, i.e. blue sky pixel, blue sky pixel, blue sky pixel, blue sky pixel… etc a million times. The slight changes between adjacent pixels are stored rather than the actual pixel data: this saves a huge amount of file space whilst still producing normal-looking images.  [For all those serious data compression wizards out there, I apologise. I hope I haven’t oversimplified the complex subject of JPEG data compression, which I know includes such very clever things as discrete cosine transformations and matrix multiplication.]
In addition to the actual data used to recreate an image, there is the need to put additional non-image data in a JPEG file.
Image cataloging programs make use of this non-image data.  The kind of useful data that’s stored is: the type of camera used, the camera settings, colour space, date and time…etc…etc.  This non-image or metadata is kind of shoe-horned into the JPEG file. It’s known as “EXIF” data.
It’s not difficult to read the EXIF data of an image file; most image manipulation software will do it and that’s what I would be using to get the full details from the Q-Skeptics photo.
Even operating systems can read the EXIF data to some extent.  Select a thumbnail in Windows and hover the thumbnail with the mouse pointer: up will pop a small window showing some of the EXIF data.  In Mac, select a thumbnail then go to file -> get info. That’s also the EXIF data.

So what does this have to do with our reported ghost photo?
EXIF data can be used as a kind of backdoor method for checking someone’s story.  If someone has flash orbs in their photo but they insist that the flash didn’t fire, although the EXIF data says it did, it does somewhat dent the credibility of the person’s story.
Closer to the example at hand, if someone says they didn’t take their ghost photo using an iPhone, but the EXIF data says they did, that also leads to a loss of credibility.
There are some complicating issues, however:  EXIF data is not a fixed standard, and not all cameras will add complete EXIF data to an image file.  Camera phones are amongst the worst offenders, inserting very little EXIF data into the image files they produce.  It’s also possible to manipulate (hack) the EXIF data with the appropriate software, and thus it’s not “hoax-proof”. However, most weekend pranksters won’t bother to go that far.
Another important point: photo manipulation software, particularly the cheap/free stuff, tends to destroy the EXIF data.  This is because, as I’ve mentioned before, the EXIF data is kind of shoe-horned in amongst the compressed image data.  For some software programmers, keeping the EXIF data AND manipulating the image, it’s too difficult, so EXIF is discarded.
Even the act of e-mailing a photo can remove the EXIF data.  Some e-mail programs automatically shrink an image file to an acceptable file size, i.e. they manipulate an image before sending.
To get to the bottom of the ghost photo story, I’d need the help of friends with iPhones, as I don’t own one myself.
Native Resolution
Another concept to consider is that of “native resolution”.  Every camera, including a camera built into a phone, has an electronic image sensor.  That’s the part that converts light to digital bits, commonly known as the CCD (even if it’s not actually CCD technology).  CCDs have an inherent number of pixels, horizontal and vertical.  The number of hardware pixels is equal to the “native resolution” of a camera.
The specification for an iPhone says that it’s got a 2-mega pixel camera, made up of a 1600 x 1200 pixel array.
The first thing to do is to ask friends for a control image.  Any image from an iPhone at the full resolution of the iPhone (1600 x 1200) would do. That would ensure, as much as possible, that the image is the native resolution of the iPhone, and hasn’t been manipulated.  Then I’d check the EXIF data.
First snag: all the images my iPhone friends sent to me were much smaller than the native resolution. Ha! A little Google searching found the answer: the e-mail program on the iPhone itself automatically cuts down the resolution of images taken on the phone.  The only way to get the iPhone’s full native resolution was to hook up a cable and download the images directly from the phone to a computer.
In the meantime, the news that I was waiting for came through.  The answer was yes, the reported ghost photo to Q-Skeptics was indeed taken on an iPhone…..hmmm.  It seemed very likely that it was all down to the “ghost app” and an annoying prankster.
Yet more news:  a subscriber to the same message board, Lynne Kelly, an iPhone user, had loaded the ghost app onto her phone and had posted an image very similar, in its ghostly representation of a little girl, to the little girl in the image that started this episode.

Another friend, Mark, with an iPhone, had also tried to get the “ghost app”. His report to me was even more bewildering.  “There are 25 ghost apps to choose from. There is even a ghost video app that puts ghosts in the videos you produce on your iPhone”. Aaaargh!
I still wanted to investigate, although this news made investigation very difficult.  The news that there are so many ghost apps destroys the credibility of an iPhone as a source of images for ghost believers.
I still wanted to know the full photographic details.  What does the EXIF data show where an image is downloaded from an iPhone and not manipulated?  Is that data different from the EXIF data from an image that has been manipulated through the use of the ghost app?
Control Image
The EXIF data from a directly downloaded iPhone image showed:

Created: 11/3/2010, 11:58,47 am
Modified: 11/3/2010, 11:58,47 am
Make: Apple
Model: iPhone
F-stop: F2.8
Pixel Dimension X:1600   Y:1200
Orientation: Normal
Resolution X: 72     Y:72
Resolution Unit: Inch
Colour Space: sRGB
[All other EXIF fields are blank.]

Interestingly, it does say that the photo was taken with an iPhone, and the camera aperture is shown, although this is much less EXIF data than I normally see from an ordinary (non-phone) camera.  There was no indication of shutter speed, ISO rating, use of flash or metering mode – many of the details I’d like to see, simply blank.
Reference Image
My friend Mark had also managed to install on his iPhone the app that produces the little girl fake ghost, seen in the original image submitted to the message board, and in Lynne’s fake ghost app photo.
The EXIF data shows (a vertical orientation image in this case):

Pixel Dimension X: 679    Y:800
Resolution X:72      Y:72
Resolution Unit: Inch
Colour Space: sRGB

ie, very little EXIF information, just as if the photo had been manipulated, which it had, by the ghost app.  What was surprising was that it wasn’t possible to get an image at the full native resolution of the phone via the ghost app.  Images were always cut down in resolution.

It looks to me as though EXIF data is still a useful tool in relation to fake iPhone ghost photos. My advice, if you’re ever asked about ghost photos, is to try to suppress the urge to offer possible explanations of the image itself, and instead, first ask for the original native resolution image from the camera (1600 x 1200 for an iPhone).
At the very least, the EXIF data should say what kind of camera took the photo, (including if it was an iPhone).
If the EXIF data doesn’t mention the camera type, it’s very likely that the image has been manipulated before you’ve seen it.
In conclusion, the ghost photo that was sent to Q-Skeptics:
•  Has exactly the same ghost image added to it, the little girl in old style clothing, per other photographs known to have gone through the ghost app.
•  Has exactly the same ghost image added to it, the little girl, as shown on the web page advertising the ghost app (see images at the end of this post).
•  The image size is not the native resolution of the iPhone, which indicates is has probably been manipulated by the ghost app.
•  The EXIF data, imbedded within the image, is identical to other photographs know to have gone through the ghost app.
Ghost app in the news
To finish off, what do you think might happen when pranksters go one step further and try the ghost app on a serious ghost whisperer?
On the 10th of February 2010 the Melbourne Age newspaper reported on a story which developed at Western Australian’s radio station Nova 93.7 FM.
The upshot of the story was: a photo showing a ghostly looking little girl in old- fashioned clothing (that should ring some alarm bells) was passed on to ghost expert Anthony Grzelka.  Anthony had the photo checked by his own photographic expert, who confirmed it wasn’t a fake.  Anthony proceeded to do a reading of the photo.  Anthony said the ghostly little girl depicted was named Bethany, who died in a fire in the early 1900s.  Anthony then went on to read other interesting and personal details from the purported ghost photo.
If you haven’t twigged, the camera that took the photo was an iPhone. Ghost whisperer Anthony Grzelka had failed to read the most important thing from little Bethany: she leads a happy little digital life, in an iPhone 🙂
The “ghost app” might be very annoying to skeptics, but it’s not half as annoying to serious ghost whisperers.
Mal
Follow up: several weeks ago I sent an e-mail back to the original poster of the “ghost” photo.  I wanted to know if the owner of the phone was now willing to say if the ghost app was installed on the iPhone from which the image came. The original poster forwarded my e-mail to the owner of the iPhone. As yet, I’ve had no reply.

Links:

The Age article
http://www.theage.com.au/technology/technology-news/iphone-caught-up-in-ghost-reading-20100210-nqxd.html
Wikipedia – EXIF
http://en.wikipedia.org/wiki/Exif
Ghost app
http://www.appstorehq.com/ghostcapture-iphone-79452/app
Orbs
http://www.paullee.com/ghosts/orbs.html
http://www.csicop.org/si/show/nonmysterious_orbs/
http://www.skepdic.com/orbs.html