The Appearance of Motion
How We Perceive Motion
In both non-moving objects and in physically moving objects
Race Horse 1878 by Eadweard Muybridge
Eadweard Muybridge: text says: "'Sallie Gardner,' owned by Leland Stanford running at a 1:40 gait over the Palo Alto track, 19th June 1878" -
Photographs made at intervals of 27-inches and 1/25th of a second - exposure time less than one two-thousandth of a second
Getting good answers requires asking the right questions. For centuries we were aware that we perceived motion but gave it little real thought (Note exceptions for philosophers debating time and existence such as Parmenides). Existence and perception were givens but there are biological mechanisms behind our perceptions. Research shows that seeing still objects and seeing moving objects come from different parts of the brain. Damaged brains give us insights. There are some people who can see objects only when they are moving and there are some people who can see objects only when they are still. In a sense, still-pictures in sequence are an opportunity similar to brain damage, in that it presents a circumstance we would not otherwise run into, which allows us a glimpse of how our brains work.
When the first sequences of still pictures displayed in rapid succession gave us a sense of watching motion, the question was asked about how we could see those still pictures as moving pictures. Later, with the first cinema, the question was still there. How do we see the sequence of pictures on the roll of film as if we are watching real motion. But it was the wrong question.
Zoopraxiscope of Eadweard Muybridge: Picture disc and the resulting "moving picture." The text reads "The Zoopraxiscope 35 A Couple Waltzing Copyright 1893 by Eadweard Muybridge - Patent applied for"
Instead of asking how we "see" motion in a sequence of still pictures we should have asked :: how do we perceive motion visually. Then we would have been looking for a mechanism or mechanisms which allow motion perception in both for real motion and for sequences of stills. By focusing on appearance of motion in looking at a still picture sequence conclusions are limited to conceptualizing the appearance of motion as dependent on a set of overlapping still images.
Before we go any further the correct name for the way we get the sense of motion from a set of still pictures (frames) in a movie or video is called Near Apparent Motion (or Short-Range Apparent Motion).
Not persistence of vision - disproved in a 1911 paper on apparent motion
Not phi motion or phi phenomenon (a common wrong "correction")
Not Beta movement
The wrong, and still-taught concept of "persistence of vision" does not exist. This concept assumes that the eyes (in particular, the retina) does the work of sensing movement. The idea, wrong by a century, is that
the last image remains on the retina as the next image arrive, overlapping images - WRONG
and that the overlap is sorted out in the retina and recognized as movement - WRONG
If the old persistence-of-vision theory were true, with one image being superimposed on the last one on the retina, we would be seeing something like this multiple-strobe exposure of a ballerina (1947, by Gjon Mili), although, unlike this picture, the most recent "frame" would be brightest and the rest would fade away in time, still overlapped. Even were it true we would then need yet another theory to explain why we do note see multiple overlapping images.
Even a quick look at this example shows a number of problems with the old persistence-of-vision theory.
I've put together four frames from video of Rose Taylor-Spann in a jeté (leap). The last frame is brightest with each previous frame more transparent. Not only would the retina have to decide which of the overlapping and fading still images is most important but it would have to continually parse out the new from the old-but-still-visible information. Which parts of the scene are brightest for example may seem simple except that you will note the places with little or no movement get brighter just from the overlaps, such as the dancer standing in the background and the floor and so forth.
The four video frames included at the top are the source of the lower combined image. Even though there is no sense of movement (apparent motion) from seeing these four frames laid out this way, you can tell, by comparing frames, what has moved and what is taking place. By contrast, the overlapped images have to be carefully sorted through to determine which image sections correspond to which frame assuming you can really parse such an overlapped image that well, indicating that getting a sense of motion from overlapped images would be very difficult. It is tough enought to see clearly anything in the combined image. Instead of combining images, the brain has to "sample" "frames" from the retinas and compare them, one to another. The results of that comparison are presented in what we experience visually as motion, that motion is "painted on" or created within the brain based on external samplings of visual data.
From wikipedia - http://en.wikipedia.org/wiki/Akinetopsia, following are excerpts from the linked page.
"Akinetopsia, also known as cerebral akinetopsia or motion blindness, is an extremely rare neuropsychological disorder in which a patient cannot perceive motion in their visual field, despite being able to see stationary objects without issue. For patients with akinetopsia, the world becomes devoid of motion. Most of what is known about akinetopsia was learned through the case study of one patient, LM. There is currently no effective treatment or cure for akinetopsia."
1911 - "Potzl and Redlich reported a 58-year-old female patient with bilateral damage to her posterior brain. She described motion as if the object remained stationary but appeared at different successive positions"
1918 - " Goldstein and Gelb reported a 24-year-old male who suffered a gunshot wound in the posterior brain. The patient reported no impression of movement. He could state the new position of the object (left, right, up, down), but saw "nothing in between"
1978 - "LM, a 43-year-old female admitted into the hospital" She perception of movement depended on speed and direction (up/down, left-right) - she had some small sense of horizontal movement at a speed of 14 degrees per second otherwise "sometimes at successive positions in between", but never as motion."
2000 - "70-year-old man presented with akinetopsia. He had stopped driving two years prior because he could no longer "see movement while driving". ... When objects (on television) began to move they would disappear. He could, however, watch the news, because no significant action occurred."
2003 - " 60-year-old man complained of the inability to perceive visual motion following a traumatic brain injury or TBI ... He gave examples as a hunter. He was unable to notice game, to track other hunters, or to see his dog coming towards him. Instead, these objects would appear in one location and then another, without any movement being seen between the two locations"
Predictive MotionsTo illustrate the multi-faceted nature of the the brain's predictive motion processing, another type of motion sense is used in catching fly balls in baseball. In 1995 researchers Michael K. McBeath, Dennis M. Shaffer and Mary K. Kaiser realized that the brain does not accurately calculate accelerations, which is all that the ballistic curve of a ball in the air is about. So how does an outfielder manage to be in the right spot? It turns out that outfielders run in manner which makes the ball's trajectory look like a straight line by running in a curve under the ball. So there is no automatic calculus being performed. This lead to other observations in which it is also how predators track their prey and fighter pilots track other planes. Neither position, nor velocity nor acceleration need to be calculated with this method.
- Our brains operate on their own, internally, "doing things" and using external "data" to modulate what it does
- Our brains operate on predictive models, changing operations when the prediction doesn't match external data
- Image detection happens at the retina and is fed to the brain which chooses what to do with it
- Motion detection happens in the brain and the appearance of motion is a construct
- specific visual images, as still excerpts, segments of external data, are picked up by one part of the brain
- the sense of motion with visuals from one moment to the next is produced by another section of the brain, as part of predicting the where and the when and the what for the outside world.
- The selection of what to be conciously aware of is decided in the brain
- Our "awareness" (all senses) is an assemblage which is "edited" or created by the brain running internal programs modulated with external data
- the "display" of experience, in this case the visual display, is "published" by the brain, in real time with a tiny delay
- The delay between input data and "displayed" data can be measured
- The brain makes decisions about what we will be aware of and decisions about what it will react to automatically without telling us. This is where training (think of so called "muscle memory") comes in. New learning and checklists require the fully conscious awareness with long delays to be useful. When you need to react quickly to changing conditions, you need to rely on training for "second nature" reactions. It is training which eventually pushed concious routines into the unconcious routines in the brain.
A Few Terms:
Gestalt Psycology, Near Apparent Motion, Far Apparent Motion, Induced Motion, Relative motion parallax, Beta Movement, Phi Motion, Flicker fusion
Often, movement is detected or created within a frame as a reference object.
An example of Phi Motion. There is no actual motion. This is constructed of two vertical bars as separate images, on separate layers. The only thing uniting them is that they are in the same frame and that one is displayed after the other. Our brain creates the appearance of "motion"
The blue vertical bar on the left is on for 0.2 seconds, blank for 0.05 seconds, then the right side bar is on for 0.2 seconds and off for 0.05 seconds.
Above is the setup in Photoshop CS4 to create this animation. Not shown is the "Layers" dialog to show that this is done with four layers. But here you can see all four frames used in the animation and their timing.
Example of Beta Motion. This works like the lights around a cinema marquee. Just a couple of lights at a time turning on and then off in sequence so that they seem to move around the area on a track or course.
In this frame Eric jumps up and folds his legs under him. In this next frame Eric jumps up and extends his legs to the sides. The animation below was created with only two photos, the two frames above. These are two successive frames shot with a DSLR during an in-studio rehearsal. The dancer is Eric Sobbe. The shooter, of course, is me.
Eric seeming remains airborne and seemingly moves his legs from under himself to the sides and back and out and so forth. In reality each of the two photos was taken at the top of two separate jumps, one right after the other.
But because of the timing between each photo and the amount of time each photo is presented, our brain gives us a sense of actual movement of Eric's legs by "painting" the motion, much like "tweening" the frames. You get the experience of watching actual movement.
And while Eric was certainly moving, the movement that we see in just these two frames didn't happen as we experience it.