We built "Lenny" as a 3D character within the computer and built a "set" around him. We added shaders to the objects so that the brick looks like brick and the spandex looks like spandex. We used virtual lights within the computer to light our scene, just like we would in the real world.
We then rotated the scene as if on a turntable (while keeping the virtual camera stationary) and rendered images at various angles . The number of images we rendered was based on the final lenticular resolution, and the amount of change in the angle of each image was adjusted to give the maximum illusion of depth without leaving too much space between each view. Otherwise the viewer would interpret this extra space as a jarring effect as they look across the image, instead of the desired smooth transition from image to image.
Note to the Interlacer: Our goal was to create a 3D scene where the viewer would believe to be seeing slightly around objects (smooth transition) before the parallax shift takes place. To accomplish this we needed to test various degrees of scene rotation. We then would capture each view frame (images of various angles). How many to capture depends on the target resolution and the lenticular lens LPI. We found that using a higher degree did give a bit deeper 3D illusion however there is a trade off being the smoothness of the action while looking around the objects (jarring effect) and the sharpness of the foreground and background elements. Going with a smaller degree did sharpen the scene but the price paid would be less depth. We needed to find that perfect balance and after a few variations we believe that we did.
Perspective and Field Of View
Perspective plays a vital role in establishing the perceived depth of a scene. The more the exaggerated the perspective is, the "deeper" the scene will appear to be. Exaggerated perspective is accomplished by the lens of the camera (either real or virtual).
 These two cubes are the same object, rendered through very different virtual lenticular lenses. One box looks very flat. It was rendered through a lens with a 10 degree Field of Vision (FOV), the physical equivalent of a 206mm lenticular lens in the real world. The other box is wildly exaggerated by a lens with a 100 degree Field of Vision.
 This would be the real-world equivalent of a 15mm lenticular lens. The more you push the perspective of your image, the closer you come to a "fisheye" lens effect that distorts your image.
  |
These two images show the cube and the cameras used to render it as viewed from above. You can easily see that the camera on the left has a much more narrow field of view and is much farther from its target than the camera in the right.
The camera on the right not only exaggerates its subject, it would show far more of the surrounding environment as well. Because of this, you must be careful to create enough art to fill the entire view of your camera, and be aware of your camera's position within the scene. |
  
For the Super-Lenny character, we experimented with a number of different virtual lenses to achieve the desired amount of perspective without pushing the character too far out of proportion. When we changed the Field of Vision of the lens, we also needed to change the position of the camera to keep the Lenstar logo where we wanted it. This has the effect of opening up the scene and showing more of Lenny's cape and environment.
You can see how small Lenny's feet (which are extending back into the image) are compared to the hand that holds the logo (which is projecting out towards the camera).
To read about more lenticular techniques, please continue to page 2. |
Although this is one of the easiest ways to create a 3D image, the results tend to appear as flat planes in 3D space because that is essentially what they are. You can lessen this through the use of "depth masks" which are Black and White images that tell the 
