By Steve Shaw, Light Illusion
Last month we looked at edge violations and floating windows. I concluded that when viewing stereoscopic images on a small-sized screen, edge violations in negative parallax are something that must be considered, especially to the left and the right of the screen, and using floating windows can be a useful way to manage issues with objects exiting the screen unevenly and causing stereoscopic failure. This month, I will discuss interocular distance, miniaturisation, and convergence, and talk a little about focus.
Interocular distance (interaxial separation)
The amount of stereo effect is defined, in part, by the separation of the camera lenses, which defines the relative parallax differential between the left and right eye images.
If the interocular distance (the distance between the camera lenses – more accurately called the interaxial separation) is too large, objects in the close foreground, or in the far background, can be excessively separated by the exaggerated stereo effect, causing the human eye/brain combination to have to work harder than expected to bring the image into focus. This breaks the suspension of disbelief that theatres rely on to immerse the audience within the story, and can easily invoke headaches due to the unrealistic nature of the image.
Consider the following two images:
With the exaggerated stereo effect in the second image comes a lengthening of the apparent distances between objects, as well as the length of the objects themselves – obvious by comparing the above two images. This can cause some very unnatural looking images, especially where the human body is concerned, with overly long, and very thin looking arms, not to mention the effect of miniaturisation (see later).
It also becomes a lot harder for the human eyes to ‘fuse together’ the two images, specifically the close foreground in the above second image.
In addition, it should be remembered that the stereo effect is a relative one – depending on screen size and viewing distance (this will be discussed further in the next release of this series) as well as the camera interocular distance. A good stereo 3D calculator, such as the SGO Mistika one for Android phones, can be a help in controlling the overall scene depth budget.
From the camera perspective, stereoscopic 3D requires two cameras, as shown below.
The problem with rigs such as the one above, is that the interocular distance is too great for any relatively close-up shots, since the stereo effect will be too exaggerated, causing the issues outlined above.
The following rig uses two smaller Iconix Video cameras on a rig that can get to just about the optimal 2.5-inch interocular distance.
In the following image, a mirror rig is used which allows for the setting of small interocular values down to zero (and yes, that is me in the background!)
The amount of stereo effect is defined, in part, by the separation of the camera lenses, which defines the relative parallax differential between the left and right eye images. As already mentioned, extreme interocular distances can cause problems when attempting to bring the two images into focus. There is another effect of using a large interocular distance – that of miniaturisation.
The following image has been shot with a relatively normal interocular distance of approximately 2.5 inches, and looks fairly natural to look at.
With the exaggerated stereo effect comes both a lengthening of the apparent distances between objects, as well as the length of the objects themselves, but more importantly, the second image gives the impression that the objects are all miniature models, not life-sized.
This is due to the amount of ‘stereo’ being more equivalent to that generated by a small object very close to your eyes, rather than a larger object further away from you. Therefore your brain is telling you that with such a large amount of stereo, the objects must be small and close to your eyes.
Conversely, using small interocular distances, below that of the average human eye’s 2.5 inches, generates a feeling that the objects being viewed are larger than they are in reality, due to the reduced stereo effect.
Convergence and focus
Looking at the image below, the convergence and focus point is set to the first newel post, and the forward banister rail has excessive divergence where it exits the screen. Since the banister rail is exiting screen so close to the camera, it is always likely to suffer such issues, unless the interocular distance is set a lot smaller.
In the following image the convergence point has been moved close to where the banister rail exits the screen, while the point of focus is still the newel post:
This raises an interesting point as the point of focus is still on the newel post, while the point of convergence is now on the front portion of the banister rail.
The point here is that some people suggest that focus and convergence should be locked – but as can be seen here, there are valid reasons why the two should not be linked – apart from the fact that shallow depth of field can limit the 3D experience as it reduces the ability for the whole stereoscopic effect to be enjoyed by both eye and brain.
Yet another effect of moving the point of convergence further forward is that it lengthens the apparent distances between objects, as well as the length of the objects themselves – much as moving further away from the screen has the same effect.
It is important to understand the effect of interocular distance on the perceived image. Too large, and it can cause strain in trying to focus or fuse the two images, as well as causing objects to seem excessively separated, elongated and miniaturised. Too small, and objects are perceived as being larger than they are in reality.
Where the intercoluar distance cannot be easily made smaller, it may be possible that setting the point of convergence further forward can help, but this will have the effect of lengthening objects and distances. This needs to be understood during shooting, because if changes are going to be made in post, the result can be unexpected when seen edited within non-effected shots. The reality is that it is next to impossible to ‘fix’ incorrect interocular errors.
Next month I will look in more detail at scene focus and viewing distance.
Steve Shaw is a Partner in Light Illusion, a top consulting service for the digital film market, with offices in the UK and India.