- Slides: 32
Visual Illusions Playing with Perspective CS 99 D Final Project By: Jason Anderson Professor Marc Levoy
William Hogarth 1754 - "Whoever makes a DESIGN without the knowledge of PERSPECTIVE will be liable to such Absurdities as are shown in this Frontispiece. " Source: W. Hogarth, 1697 -1764 Trustees of the British Museum
Theories of Geometrical Illusions • Eye-movement perceived length • Perspective cues • Transactionalist approach • Adaptation-level theory
Eye-Movement Theory • Line length eye movement • Testable, but usually fails – initial perception, eyes are stable • Finding an index of eye movements a problem
Müller-Lyer Lines • Eye-movement theory: Arrowheads influence extent of eye movements
Perspective Cues • Pictures converted in our brain from 2 -dimensional drawings to represent 3 -dimensional scenes • Different level of explanation – does not propose a mechanism for perception • Well established, although some ‘loopholes’ have been found
Müller-Lyer Lines Revisited • The same illusion through perspective cues • Oculomotor Macropsia/Micropsia
Transactionalist Theory • The world is a product of perception, not a cause of it – Hamlet: Do you see yonder cloud that’s almost in shape of a camel? – Polonius: By the Mass, and ‘tis like a camel indeed. – Hamlet: Methinks it is like a weasel – Polonius: It is backed like a weasel – Hamlet: Or like a whale? – Polonius: Very like a whale Old Man Hamlet (Act III, Scene II) • Change our way of looking Perception will change
Adaptation-Level Theory • Helson, 1964 – “spatial pooling” • Green & Stacey, 1966 applied to illusions • Past stimulation current stimulation • “stored norms” • “Top-down processing” • Some flaws – Ames room
Depth Cues on a Flat Surface • 1967, R. L. Gregory – all pictures are “impossible objects” • Conflicting depth cues in the content of the picture with the flat surface on which it is presented
Retinal Disparity • No retinal disparity on a flat surface • As a viewer of an image, we choose to suppress the cue of retinal disparity Source: http: //frank. mtsu. edu/~pyskip/splec 6. htm
The Acceptance of Perspective • We have come to accept that although we are seeing a flat surface, that the objects on it represent 3 dimensional concepts • Pictorial cues: interposition (occlusion), relative size, linear perspective & texture gradients • Ambiguous dimensional cues can lend themselves to be great visual illusions
Depth Ambiguity • Because of the way everything we see is projected onto the retina, there is a great deal of ambiguity
Wundt’s crosses • Hering (1879) & Wundt (1898) • Most ambiguous of all figures • Infinite number of interpretations, but perceptual system tries to settle with a ‘best’ one
Sanford’s figure • Sanford, 1903 • Although there may be an obvious ‘best’ interpretation, once can easily be persuaded to accept an alternate one!
Of Ambiguous Figures and Depth Reversals • Necker cube • Mach Book
Of Ambiguous Figures & Depth Reversals 2 • Not enough information in the image to make a decision as to the “best” interpretation • Taken advantage of to create “impossible” figures
The ‘freemish’ crate • Cochran’s photo of his ‘freemish’ crate (1966).
How did he do that? ? ? Any guesses?
Viewing from a single, special perspective • Viewing the image from a misleading perspective • Viewing from another angle wrecks the effect • Monocular viewing required • Occlusion
Misleading depth cues • Stage scenery – gives impression of greater depth • The Ames Room
The Ames Room
Of Giants and Dwarves? • Of course not! • But how?
What’s going on here? • Adelbert Ames, Jr. (1946) – concept by Helmoltz • Special viewpoint – monocular • Floor, ceiling, some walls, & windows are trapezoidal • Inclined floor • Appears as a normal cubic room
So how does it work? • Peephole removes stereopsis • Forms an identical image of a cubic room on your retina • Both corners of the room subtend the same visual angle to your eye – appear equidistant • Seckel & Klarke: Past experiences not relevant
But what about the people? • A split between perception & expectation • Apparent cubic perspective overrides sense of size constancy • Stanford psychologist Robert Shepherd – use background & relationship to the horizon to judge size
Retinal Size != Apparent Size • Distance cues: relative size of elements, separation, density, clarity, background
But is the Ames Room necessary? • Seckel and Klarke: only charm • An apparent horizontal path is all that’s necessary • Richard Gregory: same effect, ambiguous background
The Moon Illusion • Perceived distance, visual angle, & linear size != physical values • Illusion from comparison of perceived values at the horizon & at the zenith • Subtends. 5º in the eye no matter what • Not atmospheric • Illusion disappears in a “mooning position”
Theories • Apparent distance theory – appears farther away larger – Size-distance paradox • Distance, visual angle, & linear size illusions work together – Oculomotor micropsia / macropsia visual angle – Distance cues macropsia for horizon moon
The Mystery Spot • Tilted house • No visible horizon – assumed horizon with internal reference frame of house • Your body is on a tilt as well – enhances effects as much as 3 x • Application to pilots
A new perspective on seeing • Many theories, none are allencompassing yet • New ways to see things – become more aware of space through witnessing these illusions • Perspective is a powerful tool – in ‘imitating’ reality, it can also deceive • Seeing is believing Perceiving is believing