Standard-issue human eyeballs are very adaptive and clever. Of course it’s the massively powerful image processing in the visual cortex of the brain that really allows us to resolve 3D images. Stereographers have been very practised over the years in achieving good camera set-up with only simple tools. The most common test uses a picture monitor showing left eye image subtracted from the right. Depth disparities show up as blackish or whitish edges on any vertical edge features creating parallaxes. This display also shows vertical disparities and any rotation (roll) problems. However, to measure actual values of anything, a plastic ruler might be needed. (This is frequently only too true.) Slightly better, some of the specialised monitors have a grid or an offset control allowing visual estimate of disparity values. With a complicated scene, there will be many different points in the image to consider. It is then very difficult when objects and people are continually moving around. That is one reason why stereoscopic analysers are really needed.
Where a production is destined for the cinema, the large screen sizes available predicate for large disparity ranges. So the depth budget will allow larger pixel or screen percentages than those smaller ranges essential for TV. Curiously, this means cinema is actually easier to eyeball set-up than TV. In any case, cinema folk are not going to be very happy to accept 3D with basic errors that could have been avoided by proper measurement and quality control. Using a 3D display monitor is not an option for measurement. The eyeballs of an experienced operator will tolerate a wider range of depth than real audiences.
A number of on-set and post-production tools are now available which directly measure depth budgets as actually being imaged by the cameras. Some of these only show numerical results or a confusion of tagging feature points overlaid on a monitor. You can use a depth map to ‘eyeball’ the depth accurately in a whole scene. The stereoscopic analyser translates disparity values found by comparing left with right eye images into depth value. The depth values across the analysed picture are then translated into colour values according to a look-up table. Rather like a contour map of the countryside, this reveals the topography of the 3D scene. The disparity extremes that exceed the preset depth budget limits can be portrayed in warning colours. Excess negative (in front of the screen) is flagged in mauve. Other discrete colours show the intermediate values in use. The Cel-scope3D analyser has this mode among its facilities. It offers a number of different LUT colour schemes which can be selected. The one illustrated has white corresponding tozero parallax or no depth. With a colour depth map, you can instantly see where and when you are off budget.
Robin Palmer is Managing Director of Cel-Soft and is currently involved with software solutions for 3D & TV quality control and measurement technology.