By Steve Shaw, Light Illusion
In Part 5 of this series, we looked at the cut-out effect, scene focus and viewing distance, and I concluded that while it is tempting to increase the interocular distance in order to compensate for the cut-out effect, this can make the problem worse. I also showed how focus, particularly with objects in the foreground, must be considered, but can also be used to guide the viewer to other locations within the image, and that viewing distance and screen size must be considered when determining the amount of stereoscopic effect to use.
In this last part of my series, I will discuss viewing issues.
Viewing 3D Images
There are numerous different systems available for viewing stereoscopic 3D images; from monitors to projectors, from systems with glasses, to systems with none; from full-colour images to colour-restrictive anaglyph. Let’s look at some examples.
The image below shows a dual-projector setup, using polarizing filters working in conjunction with polarized glasses to provide the separate left and right eye images. This is one of the most simple techniques, and provides a very good image if set up properly and managed well.
The following image shows an LCD-based monitor system that uses two polarized screens and a ‘pepper’s ghost’ reflection screen to provide a 50/50 mix between the two polarized images.
The following image shows a single-screen LCD monitor, where each alternate image line is polarized differently, allowing for viewing of the stereo image without the need for two screens, but at the cost of reduced resolution.
All of the above systems require glasses to see the stereoscopic image, but it is of course possible to employ a system that does not requires glasses. The lenticular approach is glasses-free, but is reduced in resolution, and reminds us of the old 3D image cards you used to get for free in your Wheatybangs breakfast cereal!
Headache-inducing Problems: Accommodation and Convergence
Throughout this series on stereoscopic 3D, there have been a number of comments on various effects that cause problems when viewing the stereoscopic image. Most have come with suggestions of techniques for avoiding the problems, and hence reduce the probability of causing headaches within the viewing audience. There is one issue however, that cannot be overcome by different shooting techniques, or through post-production.
Accommodation is where the muscles in the eye distort the eye’s lens to focus on a given object at a given distance, while convergence is where additional muscles in the eye rotate the eyeball so that the left and right eye images align to produce a single image (with parallax differences due to the interocular distance between the two eyes).
In the real world, this muscular adjustment for accommodation and convergence are linked, as the point of focus for the eye’s lens and the point of convergence for the eyeballs occur in the same place. This is known as fusion; where the point of focus and convergence are the same. The eye’s muscles are used to working this way.
When looking at a stereoscopic 3D image however, accommodation and convergence do not work like this. The focus (or accommodation point) for the image is always at the screen plane; while the convergence point varies with the perceived depth of the image. The only point at which accommodation and convergence are natural, and in agreement, is when the object being viewed is on the screen plane.
This is far more of a problem when viewing images on a small screen, such as you are probably doing now, than viewing the same image on a large theatrical screen, since the eyes’ focus point (accommodation and convergence) is much nearer to infinity. This is one reason that stereoscopic images tend to look (and feel) better when viewed further away; from the back row. It is also a reason why positive parallax images (images placed behind the screen plane) can be easier to look at than negative parallax images, where the image is all in front of the screen plane, since the positive parallax images are less stressful on accommodation and convergence.
Could this be the problem that stops stereoscopic 3D going mainstream – especially on DVDs and TV transmission, where the TV screen is small?
Divergence (the need for the eyes to go wall-eyed, i.e. rotate outward to fuse the two separate left and right eye images into one) is a problem that can be easily overcome when shooting, but is impossible to fix in post (at least with a realistic budget!)
Divergence is a problem most associated with shooting wide interocular distances and using camera convergence to set the zero parallax point, with background objects some way away from the convergence point. In such a set-up, objects behind the screen plane (the zero parallax point) can become too far separated, requiring the viewer’s eyes to diverge to mesh, or fuse, the image into a single occurrence. For many people this is impossible to achieve if divergence is required at all, and for those who can accept some divergence, a value of no more than 1 degree, or 1.5 degrees from the eye’s convergence angle at zero parallax (the screen plane) is acceptable. However, even a 1-degree divergence is a huge value!
This is yet another reason why shooting parallel can be preferable, since it is all but impossible to cause a divergent image, except when shooting very wide interocular distances and horizontally re-positioning the images in post to push most of the image behind the screen plane.
Producing stereoscopic 3D images that are comfortable to watch depends not only on the production and post-production processes, but on the final display media. The future of the 3D market depends on all of us, as a fledgling industry, using our creative and technical skills to make engaging content that is a pleasure to view. It is incumbent on us to be informed about what works and why, what doesn’t, and what the latest best practices are.
Digital stereoscopic 3D production is still in its infancy. It isn’t just about making 2D with an extra dimension, it is a new art form and there is still much to learn and new techniques to develop. In this series, I hope I have given you a good grounding in the basics, but I expect you to move on and discover more, and I look forward to your comments as you do so!
Steve Shaw is a Partner in Light Illusion, a top consulting service for the digital film market, with offices in the UK and India.