From the crude grayscale charts used in early film production to present-day sophisticated color alignment systems, test patterns have been fundamental to the production of consistent, high-quality images. High-definition and advanced television systems have further emphasized the need for precision test materials. The aim of this paper is to examine the different types currently in use and to offer suggestions in the selection of appropriate systems for different applications.
Owners of new "smart" cameras sometimes question the need for test patterns and camera alignment. Out- of-the-box image quality is usually good, but a camera's performance can invariably be further improved or modified to an owner's personal preference. This is particularly true if the images are to be intermixed with those from other cameras. Despite manufacturers' common goals of good image reproduction, "all cameras are not created equal." Each brand has its own "look"; there’s the Sony look, the Ikegami look, etc. In addition to variations in reproduction between makes, there are also differences between models from the same manufacturer. These differences cannot only be minimized, but a good technician with suitable alignment tools will in fact make cameras match. A change of lens or scene illuminant can also have a significant effect on color reproduction - for optimum picture quality and image matching, camera matrix adjustment is imperative.
Defining a Reproduction Goal Individual preferences dictate whether the goal is simple camera matching, improved dynamic range, consistent color, a sharper image, or some other parameter. Regardless of the emphasis placed on a particular facet of image quality, the fundamental goal in video production is the same as it has been in film for over 100 years, namely control and image consistency. To achieve control requires not only the use of meaningful, calibrated, and consistent, test targets, but also an understanding of their attributes and idiosyncrasies. Test targets fall into two categories, each with its benefits and limitations: Charts, typically printed on paper or board, are lit from the front Slides/transparencies, lit from behind with transmitted light.
Front-lit charts are fast and easy to use and have always been popular for studio and mobile applications. Charts use ambient or set lighting, do not require an illuminator, are portable and are less expensive (Fig. 1). Potential drawbacks are narrow dynamic range, fading, and color inconsistency. Being less robust than slide systems makes charts more susceptible to handling damage; however, the major problem with front-lit charts is invariably flare.
Rear-lit slides are generally preferred in labs and maintenance shops because of their greater precision, image stability, wider dynamic range, and absence of flare. Transparencies, however, require an illuminator, typically a light box, integrating sphere, or an ambient light illuminator (Ambi). Light boxes and spheres tend to be cumbersome, require power supplies, and are generally expensive. Ambi was designed to reduce these limitations.
Figure 1. Front-lit chart and ambient light illuminator with rear-lit Combi-1 pattern.
Slide Illuminator Options
Light boxes
The spectral distribution of a light box seldom matches that of the scene. Cameras aligned using such illuminators can exhibit significant color differences, even when the camera is subsequently white balanced to set lighting. This is particularly true when using discontinuous light sources on the set or in a light box. Fluorescent tubes would be the logical illuminant for light boxes, except that their CRI (color rendering index) in typically low and the illumination level is considerably higher in the center of the tubes. Care must also be taken in compensating for the color component of the diffuser material, which we have found to be typically 100 to 200o K. warm (orange/red).
Integrating spheres
Spherical Illuminators are excellent in a lab environment. To take advantage of an integrating sphere's evenness capability requires a minimum 1-m. diameter sphere to evenly illuminate a standard 10-in. (254mm) slide. Being less than half this diameter, spheres used in broadcasting are usually undersized. If accurate color temperature is a concern when using spheres, it should be noted that the white barium interior coating should be regularly refinished as it discolors with age and is easily marked. Spheres and light-boxes are somewhat cumbersome and seldom used in a production environment.
A Different Type of Illuminator
Broadcasters have expressed the need for a sturdy illuminator that had a sphere's evenness, was compact, and used any light source. The ambient light illuminator was designed to meet these criteria and has extended the use of rear-lit test patterns not only in engineering and operations, but also in production. Its benefits include:
| Inherently even illumination. Reflector panels, together with a light source remote from the color corrected diffuser, produce typical evenness levels of + 2%, or +1% with critical alignment. |
| Adjustable light level. By adjusting the light source, brightness can be set to virtually any level, from camera makers’ recommendations to many times higher or lower. |
| Versatile. It can be used on a stand, hung from the lighting grid, or placed on a news anchor's desk, to be illuminated by an overhead hair light |
| Matches scene lighting. The illuminator can use virtually any light source; quartz, fluorescent, HMI, daylight, sodium, etc. |
Test Pattern Anatomy
Various imaging technologies are used to produce test patterns, including ink printing, photographic processes, and high-definition jet and laser printing. Each has its benefits and disadvantages.
Photo screen-printing is often used for both charts and slides. This technology is efficient and inexpensive, but evenness is hard to achieve and the dot pattern can be troublesome, particularly with HD cameras. The problem is most noticeable using small format charts when the camera’s imaging pixel matrix "beats" with the test pattern's dot matrix, resulting in moiré patterns and other artifacts. To ensure CCD compatibility, DSC uses continuous tone photographic, dye or high-definition random dot images in both rear and front-lit patterns.
Chart Challenge From the Space Industry
Because light boxes were too bulky for use in spacecraft, there was a need for a precision front-lit grayscale/colorbar chart having an extended dynamic range. Such a chart would also be useful for earth-based broadcasters in ENG, mobile and production applications. A manufacturing feasibility study revealed a number of potential problems that would have to be overcome. Spectrophotometric plots were made of charts and construction materials. Of the dozens of printing papers, cards and plastics tested, few were flat across the visible spectrum; the same was true of black inks and dyes. It was common to see a black ink trace increase dramatically from 620 Nm up through infrared. Aligning cameras to charts printed with such inks could result in differences in color reproduction between cameras, depending on the cutoff wavelengths of their IR filters. Serious as these problems were, resolving surface flare issues was a greater challenge.
Matte Surface Questioned
In attempts to eliminate reflections and achieve a Lambertian surface, charts have traditionally been produced with a matte finish. While this eliminates or reduces visible reflections, in reality the reflections have only been converted to flare. Few video operators would align a camera with a chart having 2% reflection in the dark steps; it would be visible and unacceptable. However, cameras are frequently set up to charts having surface flare of 2 to 5%. Chart flare effectively lowers the contrast ratio, or dynamic range, of a matte surfaced chart. This happens differentially, not only with changes in the position of illuminants relative to the camera and chart, but also with the type of lighting and scene conditions. Chart flare is particularly detrimental to detail in darker areas. For example, 2% surface flare at the chart's white step is relatively insignificant, but at step 11 flare results in a 100% error by doubling the chip reflectance from 2 to 4% (Fig. 2). Some DPs attempt to address this problem by adding a black velvet patch or by cutting a hole in the chart, which does provide a truer black. However, it is important to be aware that 2% chart flare also increases the reflectance of the crossover (gamma) step by 7.5%. A camera set up under these conditions will typically produce images having crushed blacks and reduced shadow detail. The problem is compounded when multiple cameras are aligned to the same chart as the flare level will be different for each camera.
Figure 2. Surface flare on matte charts can increase dramatically from gamma crossover through darker steps creating problems in image quality and consistency.
Dynamic Range of Standard Charts Relative to that of Modern Cameras
The 9-step DSC test pattern has a 25:1 ratio grayscale, while the ratio of the 11-step is 50:1 (surface flare further reduces these values). Modern cameras have a dynamic range that exceeds 3000:1; setting up such a camera to a 50:1 ratio test chart leaves a large section of the camera's tonal range effectively uncalibrated. By comparison, rear-lit systems have no problem achieving such a dynamic range; for example, DSC's 72dB 13 step OSG® has a white to black ratio of 4000:1.
Unique Solution Proposed
While defying conventional practice, lab tests showed that charts produced with certain combinations of highly reflective materials could greatly increase dynamic range. Flare would be replaced with reflections and, unlike flare, reflections are relatively easy to control.
Having established that the desired dynamic range could be achieved, beta versions of the new charts, called CamAlign, were introduced in 1999. There was enthusiastic response from all but one individual who muttered that the concept was ridiculous - he would not wait for an explanation. Engineers and DPs were asked to provide a "wish list" of features for the new charts. Recurring requests were for charts that are stronger "less susceptible to damage." They should have a tougher, non-mar surface and be resistant to fading and discoloration. Realistic true blacks are a necessity. In response, the following specifications were drawn up: Rigid aircraft aluminum backing with resilient, protective edging Laminated, smooth, non-diffusing surface Stable, long life printing materials Wide dynamic range with spectrophotometrically neutral whites and optional folding cavity blacks Precision calibration sheets for consistency Continuous tone or random dot high-resolution images
The end product is a chart that is washable with mild detergent - greyscale/colorbars have even been used submerged for underwater photography.
Reports From the Field
While the patterns are highly reflective, this has not proven to be a problem (Fig. 3). Reflections are invariably easy to eliminate by simply tilting the pattern a few degrees. Because the effective contrast ratio is considerably higher than that of matte charts, any minor residual reflections have been deemed to be insignificant. For unusual locations where reflections could be a problem, such as a space station with white walls, a device called AntiReflect has been developed. It simply traps all reflections in black velvet.
Figure 3. Typical application of front-lit chart: chipping multiple cameras at network ball game.
Selecting Appropriate Test Patterns
DPs and engineers are individualists; some still want separate patterns for each test element (backfocus, flare, grayscale, colorbar, multibursts, window, etc). Others prefer combined patterns having multiple test elements on the same slide or chart. The most important criteria are that test patterns be accurate and consistent, and provide relevant, meaningful data.
Guidelines for Selecting Patterns
There are several criteria for selection of patterns. Combined patterns save set up time and cost, by reducing the number of slides or charts required. A standard DSC grayscale with extended dark tones is essential; black chips should be true and represent those of a scene. Note, accurate grayscale alignment and tracking will produce great monochrome pictures, but do not assure optimum or consistent color reproduction. An accurate, calibrated colorbar is also essential, particularly when shooting under discontinuous light sources such as HMI and fluorescent. Accurate vector colors are considered a superior engineering tool and speed color alignment. Uncalibrated color chips can be confusing.
Flesh tones are also useful. All patterns should be supplied with calibration sheets and should generate meaningful electronic signals. Multibursts, zone plates, back focus, etc. are invaluable for testing resolution, aliasing, focus, etc.
Precision Test Pattern Applications
Precision test patterns assure quality in camera manufacturing; pre-purchase camera evaluation; day-to-day maintenance; camera matching on the set. They are also used as a home base for robotic studio cameras and as film and video production reference standards As a production tool, test charts are under-utilized in all but the most efficient production facilities. A few frames of an accurate grayscale/colorbar, shot during production, are invaluable in post, both for video matching and film-to-video transfers. Colorbar information also provides a useful reference standard for SMPTE to PAL transfers and vice versa. Simply position the grayscale and color signals to the appropriate colorimetry settings. Similarly, if a camera’s color reproduction is less than perfect, a few frames of the grayscale/colorbar on tape or film will speed color correction in post.
Conclusion
This paper outlined the benefits that can be realized from the use of precision test patterns. It recommends that "a reproduction goal" be defined and that, while the goal could have many facets, the fundamental aim is image control. To this end the relative merits of front and rear-lit systems were examined. Three types of rear-lit pattern illuminators were reviewed; integrating spheres, light boxes and ambient light illuminators. Problems associated with flare in front-lit charts were identified and a practical solution proposed. Guidelines for the selection of useful test patterns were provided. Finally, it was suggested that engineering test patterns can be beneficial to the production community and are underused in that application.
Bibliography
Benedikt, F., Corley, D.F.E., Ross, R., "An Engineering Approach to Consistent Images," SMPTE J., 107: 961-966, Nov. 1998 Corley, D.F.E., "Color Telecine Alignment Slides," SMPTE J., 78:145-148, Mar. 1969. Corley, R.F.D., "Test Materials for the Alignment of Telecine Colorimetry" SMPTE J., 90:1064-1071, Nov. 1981. DeMarsh, L.E., "Color Rendition in Television," IEEE Trans. On Consumer Electronics, CE 23, May 1977. Holmes, L.H., "A Method of Producing Telecine Test Materials of Specific Density," SMPTE J., 70:699-701, Sept. 1961. Quinn, S.F., "The CBC Color Bar Chart and Color Bar Slide," CBC Engineering Review, 3:10-12, May 1968.
THE AUTHOR  David Corley
was educated at Kings, Canterbury, and Christ Church Cathedral School, Oxford. He emigrated to Canada in 1950 and worked in many aspects of motion picture and television production. His main interests are in the technical aspects of image quality. Corley and his wife Susan established DSC Laboratories in 1962 to provide precision products and services to the television and AV communities. He has patents for seamless (soft-edged) panorama technology, a dichroic additive printer, electronic shutter, ambient light illuminator, and optical signal generators.Corley has been a SMPTE member since 1966 and a Fellow since 1986. Active in SMPTE engineering committees, he was honored with the Fuji Gold Medal in 1994. He is dedicated to the Student Chapter Program and to providing ongoing education to Society members through the Web. (Reprinted with the permission of the SMPTE Journal) |
