Stereoscopic Systems, Part 1

Stereoscopic Systems-Part 1 Well, let's start with a bit of terminology. Uh, 3D animation refers to uh computer animation that's generated with uh programs that manipulate objects in a 3D space. Uh, this type of computer animation, uh normally renders images in 2D; uh but the general public when they think of 3D, they're thinking of 3D stereoscopic films. So, to avoid confusion with this terminology, uh we'll use the term stereo 3D or stereoscopic 3D, when we're referring to uh movies with uh stereoscopic view. Now, stereoscopic vision or stereopsis is the type of uh bi-binocular vision that we have with uh our two eyes seeing a different scene due to parallax. So, when we have say a red object closer to us, a blue object farther away. Uh, the two eyes see a slightly different view, and from those two different views as we see in these images here. Uh, from those, from the differences of those two views; the brain can figure out that the red object, in this case is closer and the blue object is farther away. Another effect that comes from stereoscopic vision is that one eye may see a different part of an object than the other eye. So, in this example the left eye only sees the front and the top of the cube, the right eye sees those two sides, and around to another side of the cube. So, with those two different views, the brain can judge distance and depth. Now, for the first stereoscope that allowed us to see 3D images from 2D drawings; uh was built in 1838 by Charles Wheatstone, and it had kind of a complicated form using a pair of mirrors to project different images seen by uh the left and right eye. Uh, much simpler system was invented in 1861, and it became quite popular because photography was developed about that time, and so it became very easy to use a pair of cameras to create stereoscopic images. Uh, later the View-Master stereoscope basically uses the same principle just in a different design. Now, for looking at stereoscopic images say on a screen or in a book. Uh, simpler system, is the system of anaglyph color pairs. So, here you see a couple of examples of the typical anaglyph glasses, which are used for stereo 3D. We have the red/cyan, green/magenta, and blue/yellow. And you see that these are color additive color complement pairs. Here�s a little video illustrating, what these glasses, uh what the view looks like through these glasses. So, you see the mannequin is wearing the glasses, and we put the red filter in front of the camera, and then the blue filter. So, watch her eyes, and watch how you can see through in one case and then the other case. So, here we have that one, there's um no filter; we see she has cyan on one side, red on the other. When the cameras looking through the red filter, uh the view through the cyan filter is blocked, uh and then vice versa when the cyan filter is in front of the camera, uh the view through the red filter is blocked and we see uh through the cyan filter. Here are the transmission curves for those two filters. So, you see that the uh cyan colored filter transmits light through most of the spectrum except it does not transmit in the red part of the spectrum, and then the red colored filter is the opposite, it transmits very well in the red part of the spectrum, and not very much at all anywhere else. Now, another combination, which is used uh is green magenta and uh it turns out that for a lot of displays like uh television displays just due to the spectrum of the individual LEDs, this tends to give a better stereoscopic result. So, here are the transmission curves for magenta; magenta colored filter transmits everything except for the green, and the green transmits primarily in the green part of the spectrum. Finally, there's another possible option, which is blue/yellow, those are also additive complements, it�s not as commonly used since our vision in the blue part of the spectrum is not as strong as in the other parts of the spectrum. So, when we're creating a stereoscopic film, uh you need to have a pair of cameras; one for each view for the left eye and the right eye. And so these cameras, record the scene, and this is of course very easy to do in computer animation, you can put as many cameras as you want and render different images from different views. In live-action, it's a bit more complicated, you do have to have a more complex camera rig. Then with a projection, the simple option is to take the film recorded by each camera and use dual projectors and display that, then the audience wears the appropriate color anaglyph glasses, and then they see uh with each eye what each camera was filming. The complication here is, the projectors have to stay perfectly synchronized. So, a better system is to combine the film from each camera into a single strip, which can be protected by a single projector. Now, these images have a stereoscopic effect, which means some of the elements in the scene appear to be inside of the screen and some of them appear to be outside of the screen. So, the more dramatic situations when an object appears to come out of the screen and into the theater space, and that's called negative parallax. So, in this image, uh you find that the paddle-ball, actually comes out of the screen towards the audience. And in this case the two images of the paddle-ball, which are seen by the left eye and the right eye are on opposite sides of the colors seen by those two eyes; in other words, they-they cross. Further in the background, in this image we have say the person standing in the back, they have positive parallax; so they appear to be inside the screen compared with the negative parallax of the paddle-ball, which was outside of the screen. Now, the location of uh zero parallax; that distance from the cameras is called the convergence uh distance. There's another important distance in stereoscopic filming, which is the inter-axial distance, which is the separation between the cameras kind of like the separation between your distance between your eyes. Now, the convergence distance is either adjustable by changing the angle of toe-in of the cameras, but a simpler adjustment is to do it uh post processing. And here's an example that in the first image, we have a rather extreme uh parallax. You see that the green-magenta images, everything is rather far apart. In the next one, what we've done is taken the-the green image and shifted it relative to the magenta image, so that uh the telephone is at screen depth, or at zero parallax. Now, we still have 3D effect, if you see the objects on the table, but it's not so excessive as uh in the first one, which if you try to have too much parallax, the-the images end up looking rather poor. Now another adjustment, but this one has to be done during filming, is fixing the inter-axial distance, so that�s the horizontal separation between the cameras. Now, the when this inter-axial distance is small, the objects in the scene appear to be close to screen depth. In fact, when the inter-axial distance is zero, everything is at screen depth. The larger you make the inter-axial distance, the greater the stereoscopic effect. So, things with negative parallax, look farther from the screen, things with positive parallax, look deeper into the screen as you increase the inter-axial distance. Now, you don't want to do too much of this because there's another affect, which is called Retinal Rivalry. So, each eye has a cone of vision, and uh when you have objects within that cone of vision within that angle, uh then the eye can comfortably uh resolve that image. Something which is too far outside of the cone of vision into the peripheral vision, uh we don't-don't see it clearly. In creating stereoscopic images, you want to have everything in the scene in this comfort zone which is the overlapping uh cones vision of the two eyes. In this example, this is a very poorly constructed 3-D um uh film because much of the screen is outside of the cone of vision of one eye or the other. So, we have objects which are in retinal rivalry like um this closest sphere, and this farthest square are in the uncomfortable zone, and will not be seen clearly by the viewer. Another affect that has to be considered is so-called breaking the frame, and this occurs when we have an object with negative parallax. In other words, it�s coming out of the screen. If that object touches the side of the frame, then we have conflicting's depth cues because, uh when it touches the frame, it gets cut off and so by occlusion, we consider to be inside of the screen. However, the parallax stereopsis, since it has negative parallax, we see it being outside of the screen, and this is a jarring a disruption of the stereoscopic effect. And so objects with negative parallax, filmmakers tend to try to keep them uh near the center of the screen. So, in summary; stereoscopic systems present a different image to each eye using parallax and occlusion revelation for depth perception. Anaglyph uh systems use colored filters with additive complement color such as red/cyan. An object with negative parallax appears as if it's between the viewer and the screen; in other words coming at the audience, uh positive parallax of the opposite is something that appears to be inside the screen. Uh, convergence distance is the distance from the camera to the screen, or in other words, the distance to zero parallax. And the inter-axial distance is the separation between the cameras and the greater the inter-axial distance, the greater the stereoscopic effect.