There simply is not enough data in the picture to resolve the matter of which leg is the right and which is the left, what’s moving inwards and what’s moving outwards, etc.
[QUOTE]
This is not so. When the shadow is taken into consideration, the figure moves counterclockwise and not clockwise, as proved by the shadow (makes no difference if you want to call it a reflection although it is not; she is a silhouette).
Imagine that she is jumping on a clockface, with Noon away from you and 6:00 toward you. She makes one jump each rotation. If she is rotating clockwise, she lands on her left axis foot with the right leg extended away from you at 12:00. The left foot and left foot shadow touch, but there is no right leg shadow at all.
This should not be the case; we should see the right leg shadow if it is extended away from us.
But, you might argue: the angle of light is so low and the leg is raised so the shadow is not see because it is too far in the foreground. This, however, is not possible. If she is rotating clockwise, the extended right leg is at 6:00 when she is at the height of her jump, and we can see the shadow of the extended leg at the height of her jump. If we can see it at the height of her jump when it is extended toward us, we must also be able to see it at the lowest point of her jump when the leg is extended away from us. However the shadow of the extended leg is absent.
Now consider the CCW direction. Here we see the extended leg shadow (now her left leg) when the leg is at 12:00 away from us. Despite the jump, the extension of the leg is still far enough away from the body to create the shadow in the foreground. In the CCW direction the shadow disappears when the extended left leg is pointing at 6:00 (directly at the viewer) because the shadow is offscreen into the foreground.
It may take a bit to see why this is true, but it is. Part of the SDMB fun is convincing others…
If she had a shadow, that would resolve the ambiguity. But she doesn’t; she has a reflection. A reflection of an object has all the same optical qualities as the object itself, so if she’s ambiguous (she is), then so is her reflection.
No, not at all. All that’s necessary is that the image be “reversible”, meaning that it equally corresponds to two different. I made a little experiment. Open up the following links in two browser tabs or windows:
Look at the still image first. It’s a Necker cube, the most basic reversible image there is. Because of the isometric perspective, there’s no way to tell which face is front and which is rear. Look at the still image and decide that the bottom right face is frontmost. Then look at the moving version. Go back to the still cube and now look at it as if the upper-right face is in front. Again to the moving image. What happens?
(You might need to repeat this several times…)
This is exactly the same illusion as the dancer. The shadow/reflection, the nipples, play no role whatsoever, if anything, they make the illusion harder to reverse.
I find it interesting that so many people almost aggressively refuse to believe that this is an illusion. If the direction changes, then it must be because the animation is changing.
The father of modern vision research, Hermann von Helmholtz, defined vision as a form of “unconscious inference”. In other words, vision is an educated guess as to the state of the external world. At the time, this definition was very controversial because it denies that our knowledge of the world is direct. Nowadays, Helmholtz’ views are widely accepted among researchers but most people are nevertheless convinced that “seeing is believing”. In many cases, however, you see what you believe.
Absolutely. The main difference between the left and right hemispheres is that language is handled mostly on the left and spatial ability is on the right. That’s all that has been found scientifically. Not logic vs. emotion, not “big picture” vs. details, not facts vs. imagination.
This is definitely untrue. I’ve examined the gif image frame by frame - it comprises 34 frames that make up one complete revolution. The switching is occurring in your brain, not on the screen.
That’s what I’ve thought through most of this thread, but after re-reading Chief Pedant’s explanation, I can’t account for how a reflection would work such that when she is spinning clockwise, we see the reflection of the leg when it swings forward (from 3 o’clock to 9 o’clock), but not when it’s on the backswing (from 9 o’clock to 3 o’clock).
It doesn’t matter if it’s a reflection (which it’s not, since she is a silhouette) or not. Call it a reflection if you must. The extended leg reflection disappears inappropriately if you make her spin clockwise. In a CW spin, when the axis foot is on the ground, the ‘reflection’ of that foot touches the figure foot but there is no reflection of the extended foot. This is not because of the angle of the reflection, since in a CW rotation we see the extended foot reflection at the highest point of the jump when the foot is extended toward us. If it is visible at the highest point of the jump extended toward us, it will be visible at the lowest point of the jump extended away from us, reflection or shadow notwithstanding.
I do not understand the insistence on calling this a reflection. The ambiguity of the figure arises in the first place because silhouetting it removes the A-P dimension.
The animation is playing BACKWARDS! Consider her ponytail. Her head is tilted to her left (if CCW, but to her right if CW). Normally, the ponytail will remain pointing downward like a plumb-line, and not remain aligned with her head, as it is in our animation.
Now, her head is also orbiting the axis of rotation. And the way we have it, the ponytail leads, scooping in the air ahead of her body in the direction of her head’s orbit, rather than trailing behind, as it physically should.
This problem exists in both the CW and CCW interpretations of the figure. It also persists in a mirror image of the animation (with the whole image itself flipped about a vertical axis).
The only way to rectify the situation is to reverse the direction of animation, and then her hair naturally and gracefully trails behind her head.
Furthermore, the shadow problem was created by the laziness of the animators. Instead of modeling a real light source and real shadow, they simply inverted the black figure, moved it down, and made it gray. This results in a perfect lateral symmetry in the plane of our 2D image.
I don’t think the image is supposed to be an accurate representation of the visual impression that would be created by a real-life event. Rather, the image is just an animation that has been completely blacked out. Almost all the information about lighting and so on has been erased, because all we get to see is the outline of the figure, and the outline of the shadow/reflection/whatever. Inside the ouline its black, and outside the outline its white.
You’re right - with a real-life human sized figure at normal viewing distances, we’d probably be able to determine when the leg is moving away from us by perspective clues. With a computer generated image, it’s possible to move the virtual camera a long way from the subject, then zoom back in, essentially destroying perspective and making the rotation ambiguous.
For an extra weird variation, and to demonstrate (as if analysis of the image file weren’t enough) that this is an illusion the perception of which can be controlled at will, try this: In your mind’s eye, keep her extended leg behind her at all times, as if she’s swinging back and forth through 180 degrees instead of turning in a full circle. It’s a little more difficult than switching the direction of rotation after several full turns, but it can be done, and it makes the illusion even more interesting.
Only if you are reflecting a vertical 2D cut-out that is perpendicular to the mirrored floor. For a 3D extended object, it is different.
Consider this: You are looking straight on to (but from slightly above) a long rod suspended parallel to a mirrored floor; the rod is extending away from you. You will not see the top surface of the rod flipped about a horizontal axis. You will see the the *bottom *of the rod, with the far end of the rod’s reflection away from you (and higher in your field of vision / the vertical image) and the near end close to you (lower in your field of vision).
That is correct. When her foot is extended away from us, we should see its shadow higher in our field of vision, higher in the 2D image.
If you trace the tip of her toe, you find that it traces an ellipse, not a horizontal line, in our field of vision.
Since we are looking at her foot’s orbit from the side and above, the circular orbit is foreshortened into an ellipse from our perspective. When the extended foot is at its lowest point, it must be closest to us, since we are viewing the orbit from above. (We are certainly viewing her feet from above; otherwise we would be unable to see the shadow on the ground below them.) When it’s at its highest point on the ellipse, it must be farthest from us.
According to this reasoning, she is rotating clockwise.
According to my rod analogy above, the “shadow” cannot be a reflection in a mirrored floor. **Chief Pedant **is correct that it is a shadow illuminated from behind. (The shadow is on our side of her. If she were illuminated from our side, her shadow would lie away from us, behind her.)
The only problem with Chief Pedant’s reasoning is that the gray image is not a modeled shadow – the animators skimped and merely inverted the image and made it gray.
Since the animators cheated, the shadow argument and the ellipse argument conflict.
The ellipse cannot be due to the jump, since it is in the *opposite *direction than the jump would move it anyway.
That doesn’t follow. When she is rotating CCW, she is tilted, and we do see the bottom of her feet. But that wouldn’t prevent us from seeing her reflection in the floor. After all, she is floating, so there is space to see both.
