Experiments Reveal Previously Undescribed Effects
The following section is derived from an unpublished article submitted for a proposed special edition of New Scientist*. The Dark Side of Photo Reality explained how technically sound photographs can reproduce the same person with dramatically different body weights (even though they were were taken minutes apart and reproduced with their eyes at almost identical magnifications) like these two pictures of Melissa McCarthy above.
The article is expanded to include data and illustrations from the original PhD thesis, examples quoted earlier and illustrations from a presentation to the UK Experimental Psychology Society in 2013. It concludes with a summary of the main findings as submitted to its sponsors, the UK Independent Television Commission. The research found that stereoscopic television can increase naturalness while significantly reducing the size, shape and distance distortions often found in conventional 2D television imagery (see Times article on this ITC study below) .
The Dark Side of Photo Reality
How would you feel if no matter how carefully you used a word processor, your work was altered every time you clicked “Save”? While still resembling what you remember, its font size, style or spacing may all be slightly different and even portions of the text have been changed to conform to some unexplained literary convention. Such interferences would be intolerable and send most of us back to our trusty typewriters. For photographers however, something similar happens every time we take a photo. Many carry on regardless because they have not noticed the subtle differences. But even when they should be in plain view, highly significant distortions can lay hidden in the most finely crafted of photographs. And as research by Bernard Harper discovered, the secrets that lie in every photograph can manifest themselves in the most surprising and disturbing of ways.
“Not only can the camera lie, the camera always lies“
Malcolm Muggeridge, Christ and the Media 1977
Anecdotal evidence on the unreliability of photography is commonplace. Yet when there is no actual evidence of tampering or retouching, many believe photographs truthfully capture the view seen at the time they were taken. So in the UK press and popular magazines, we often see photos of celebrities accompanied by captions like: ” Who has got the best body: Kate Winslet or Kylie Minogue?”
The consensus at that time was Ms Winslet carried too much weight to be a leading actress,* whereas the child-sized Kylie more closely approximated the ideal female shape. It was taken for granted that the body shape and attractiveness of someone they had never met could be accurately judged from their photographs.
* Winslet has never been significantly overweight and had the ideal Body Mass index of 21 or less for all of her adult career.
When 3D Reality Becomes A 2D Image
One might think it is obvious photography is an optically reductive and distorting process: It compresses three dimensions into two while discarding size-constancy (true size and distance information), colour accuracy, resolution and tonal range. However, it is only when people are made aware of airbrushing, manipulation or technical failures that they question photography’s inability to be an accurate and trustworthy method of record. So if we look fat, unattractive or barely recognisable in our holiday snaps, the photographic process is never blamed. We often assume we are not photogenic because “The camera never lies.”
The need for naturalness and size constancy goes far beyond vernacular photography. It was reported that James Cameron wanted to shoot his epic film Titanic in 3D. While it is obvious this option would have helped him deliver the scale and spectacle required, it may have helped solve another problem too. The sylph-like Gwyneth Paltrow was the first choice to play the role of Rose, but she declined. Kate Winslet landed the part, only to be called “Kate Weighs-a-lot” by the director James Cameron throughout filming. Her on-screen size became a concern, perhaps because audiences might not believe that the boyish Jack Dawson (played by teenage idol Leonardo DiCaprio) would fall in love at first sight with an older-looking and seemingly oversized Rose. It was rumoured that Cameron even considered an optical narrowing (anamorphic squeeze) of her face in post-production to make her look slimmer. However if Titanic had been filmed in 3D, it is likely the size constancy and naturalness found with stereoscopic imaging would have been all that was required to show audiences how slim and youthful Winslet actually looked in reality.
James Cameron Is Not Alone In Believing
The Fattening Effect Is Real
Lorraine Kelly, TV presenter, Sunday Mirror Magazine 1999
A “Stone” weight equals 14 pounds (6.35 Kilograms), which on average is over 10% of bodyweight. This research was commissioned in-part by the ITC as a response to claims like these from TV presenters, newsreaders and actors etc.
2D “Flattening and Fattening” Effect
In professional photography, film and television it is widely believed the camera somehow “adds 10lbs” to perceived bodyweight. Surprisingly, the experiments reported here were the first scientific study into the fattening effect of photography * They proved that body size estimation in photography is not only increased, but also strongly gender biased against women and can be wildly overestimated.
So why are people largely unaware of this effect? Perhaps our facial recognition ability is responsible. Humans usually out-perform facial recognition computers. We can recall faces with extraordinary precision, even when altered by fashion, disfigurement or age. Perhaps we recall people as the shape and size we last saw them in reality, rather than the often distorted size conveyed in the same photographs when they are viewed by strangers?
* Harper, B., & Latto, R. (2001). Cyclopean vision, Size Estimation, and Presence in Orthostereoscopic Images. Presence, 10(3), 312-334. MIT Press
Latto, R. & Harper B. (2007) The non-realistic nature of photography: Some more reasons why Turner was wrong. Leonardo, Volume 40, 243-247
“Virtually There” co-author E.U. Presence Research discussion document New Scientist 13.04.02 p8.
Previous Explanations of Size Distortions In Photography:
“Wide-angle lenses will cause curvilinear distortions that appear to distort objects and people.” Langford Advanced Photography (1989)
“Short (wide angle) lenses distort by “fattening” facial features making skinny noses wider, with fat noses, faces and necks especially wider” Photo.Net (2008)
“With the camera farther away, the difference between camera-to-nose and camera-to-ear distances becomes negligible. So the person’s features seem to be closer to their actual sizes” James E. Cutting, Moving Image Theory: Ecological Considerations (2004)
“Distortions” in Reality: Seeing is Believing
When the public meet celebrities or people known only from photographs, they can sometimes look startlingly different from their preconceptions. Few people outside film or TV know that Tim Robbins, Liam Neeson and David Hasselhoff for example are over 6’ 4” tall. Or that President Clinton, Tony Blair and Gordon Brown are approximately 6’ 3” and that Nichole Kidman, Gwyneth Paltrow, Geena Davis, Joely Richardson, Mini Driver and Taylor Swift are aproximately 6ft tall.
However, Tom Cruise, Robert Redford, Bruce Lee, Kiefer Sutherland, Patrick Swayzee, Willem Dafoe, Ian McShane and Sylvester Stallone are all reported to be 5’ 7” or less. Martin Sheen, Al Pacino, Dustin Hoffman, Elijah Wood and Henry Winkler are said average less than 5’ 5,” while Madonna, Lady Gaga, Christina Aguilera, Shakira and all five Spice Girls and are all known to be exceptionally small and slim and average at less than 5’2”.
Remarkably, Dolly Parton, Kylie Minogue, Paula Abdul, Sheena Easton, Elaine Paige and Eva Longoria are all reported to be less than 5ft tall.
Side-By-Side Scaling Can Sometimes Reveal The Real Sizes Of Celebrities, Such As This Example Of Madonna And Taylor Swift
The Height Difference Compression Effect
The image above is a rare example of an extreme height difference being conveyed uncompressed (5’1″ Madonna, 5′ 9″ Swift). These extreme differences are usually compressed, particularly with on-screen couples: The size difference between child-sized Jada Pinkett (4’11”) and her husband Will Smith (6’2”) or David Duchovney (6’1”) and Gillian Anderson (5’2”) of the X-Files would be incongruous if we were to meet them in person. But on screen, they seem to be perfectly matched.
The reason may be that if we were to make real eye contact with them, the large vertical saccadic eye motion required to shift attention from one face to another would be a powerful cue to their true size difference. When viewing TV or cinema image however, such vertical saccades are likely to be much smaller. The reduced visual angles between gaze points in photographs may misinform our visual system that large size differences appear much smaller than they really are.
The Low Viewpoint Effect
Another form of misdirection can be derived from camera positions. Practicalities dictate that tripod heights are often set at about 4.5-5ft. So we are often looking upwards at faces in photographs or on the screen, regardless of their actual height.
In reality, if we need to look upwards to make eye contact, it usually means that the person is much taller than us. The Madonna-Swift performance photo was taken from waist height and when viewed individually, makes both look tall.
This may also be a significant cause of the oft-reported observation that actors tend to look much shorter in reality than they appear on the screen.
Low Viewpoints And Avoiding Side-By-Side Scaling Allowed Eva Longoria (4 ft 11) To Convincingly Play A New York Catwalk Model
And Martin Sheen (5ft 4) To Play President John F Kennedy (6.0 ft)
3D To 2D Curvature Effect
Mapmakers have known for centuries that it is impossible to accurately represent the curvature of the earth in 2D without distortions. Mercator (1512-1594) knew that his most famous projection grossly distorted Greenland as being the same size as Africa, when in fact it is almost 14 times smaller. His equal-area sinusoidal projections reduced these distortions, but even the best 2D maps can never be truly free of errors.
Perhaps photographers should think of 2D photography as a “mapping effect” and that flattening 3 dimensional reality into a 2D image will always cause distortions?
Da Vinci Effects
Photography of the linear perspective we experience in reality however appears be so accurate that few people have thought it could be affected by map-like distortions of size or curvature. Leonardo da Vinci (1452-1519) was well aware of the power of natural 2D perspectives to convey startlingly realistic scenes. Yet he was also aware that the curvature of a human face could not be fully represented in 2D, or be reproduced with its natural relationship to a distant background.
Leonardo’s Constraint 1
“A painting, though conducted with the greatest art and finished to the last perfection, both with regard to its contours, its lights, its shadows and its colours, can never show a relievo equal to that of natural objects, unless these be viewed at a distance and with a single eye.”
Leonardo’s Constraint proposed that the shape information available to a painter from their stereoscopic vision cannot be reproduced on a 2D medium such as canvas. This is because each eye sees a different perspective that can have large areas of important foreground and background information not viewable from the other eye’s position.
With Stereoscopic Vision, Objects Occlude Less Background Area And Appear Closer, Smaller And Slimmer Than In 2D
Background area AB is significantly larger than CD, suggesting the camera’s view is of a larger object (in relation to its background) than when viewed stereoscopically from the same position. If a near target perceived without depth subtends a large visual angle, the Human Visual System may be compelled to perceive it as more distant and therefore significantly larger that it would appear with direct stereoscopic vision. These explanations and the experiments detailed in the Experiments section were first published in the MIT journal Presence and New Scientist in the UK (see below).
* The New Scientist article was not published because the editor said that the existing publications (The Times, New Scientist, Presence and Leonardo) had adequately covered the research and therefore it was not new enough to publish.
Please feel free to comment on these findings using the email address below or this link. PDFs of individual PhD chapters exploring these effects in detail can be forwarded on request to: email@example.com