Configuring Soft- and Hard-Edge Arrays, Part II: Real-World Example

Last issue, we discussed how to configure hard-edge and soft-edge arrays, detailing the differences between the two. This column, we'll detail a typical example of this technique, from conception to finished application.

Putting it All Together

In examining a real-world example, let's start by defining a typical but simple, array application. Our goal is to create an image 71'×16' as a backdrop for a televised event. This type of application is appropriate for an array, as the desired aspect ratio is very different than the native aspect ratio available from any electronic projection system. The content will be a mix of static scenery and motion video. Beyond creating this ultra-wide aspect ratio, the image must visually integrate with the elements of the stage set, so that it appears striking and realistic through the unforgiving camera eye. Thus, colorimetry, uniformity, and brightness of our 71'×16' screen are clearly critical points of emphasis.

Given the imagery will be motion video and static scenery, our goal is to create the most realistic, continuous image possible. Therefore, we elect to employ a soft-edge blend. When it comes to determining front vs. rear screen, our requirement to produce lifelike imagery that withstands the rigors of television set lighting demands, we employ the contrast-preserving benefits of rear screen. We will pick a screen with unity gain that enhances projector black levels a bit.

Now, we need to determine the most efficient number of projectors that will allow us to fill the array and meet the resolution and brightness goals. Given the screen width and its prominence on the broadcast set, the customer has asked for the highest resolution possible. We will employ three-chip DLP projectors with SXGA resolution. If we match the screen height of 16ft., each 4:5 aspect ratio SXGA projector will produce a native image that is 20ft. wide.

Next, we must account for the image width we will lose as we overlap the vertical edges of the projectors for a soft-edge blend. To make this calculation, we deduct the width of the overlap as a percentage of a single projector's image width for each overlap in the array.

A straightforward way to make this calculation is as follows: a 20% soft-edge blend results in projector No. 1 maintaining a 20ft. wide image, while each additional projector will only provide 16ft. of incremental image width due to the overlap (20ft.-(.20'×20')=16ft.). Thus, as we add projectors to the array, calculating our total screen width follows this simple math:

One projector: 20ft. image width

Two projectors: 36ft. image width

Three projectors: 52ft. image width

Four projectors: 68ft. image width

Five projectors: 84ft. image width

Unfortunately, employing a 20% overlap with four units results in an image width that is 3ft. less than than our 71ft. goal, and five units overshoots our goal by 13ft. We can either significantly increase the overlap area employed on five units, or we can go with four units and decrease the soft-edge blend area a bit. The benefits of a soft-edge blend greater than 20% are questionable given the additional resolution and light that is sacrificed in the increased blend area. We are satisfied that four units will provide adequate resolution and light output for the application. Thus, we are comfortable with reducing the soft-edge blend area a bit.

By taking that approach, we apply a 15% soft-edge overlap to the four-unit array. Projector No. 1 still provides 20ft. of image width, but each additional projector now provides 17ft. of incremental image width (20ft.-(.15'× 20')=17ft.). Therefore, we now have screen width as follows:

One projector: 20ft. image width

Two projectors: 37ft. image width

Three projectors: 54ft. image width

Four projectors: 71ft. image width

Bingo. Using four SXGA projectors with a 15% soft-edge blend results in the screen size that matches the application. We now turn to defining the projectors' required light output to support the screen brightness requirements.

Each unit is projecting light onto a 320-square-foot patch of screen (16'×20') that will be exposed to broadcast stage lighting. We therefore know we will have to bring significant lumens to the task. Based on our understanding of the stage lighting and our selection of a high-quality, rear-screen material, we feel 40ft. lamberts is the minimum screen brightness needed to achieve exceptional on-screen dynamic range. The simple calculation below defines the required projector brightness, in lumens, to achieve 40ft. lamberts:

Required projector lumens=320 (square footage of screen) ×40 (desired foot lamberts) ×1 (screen gain) =12,800 lumens per projector (minimum).

We can conclude that four Digital Projection Lightning 28sx units (at 16,000 lumens each) will provide more than enough headroom to excel in this application.

Although we want to avoid short-throw lenses, rear-screen projection space is limited on the broadcast set. Fortunately, in our example, the end-user can provide us with enough throw distance to employ a 1.5:1 throw ratio. The 1.5-2.0:1 zoom lens that accommodates this ratio benefits from good (but not quite perfect) geometry. Given we have a 15% soft-edge blend area between each projector, and the content is scenery and motion video, we feel this lens will more than fit the task.

Prep Time

Now that we have defined our screen type, array size, projector model, and lens, we are ready to do some advanced prep work to ensure an efficient setup on site that creates an optimized result.

One unit at a time, each of the four projectors is configured in our imaging lab to create a 16'×20' image. Using an incident (illuminance) colorimeter such as a Minolta CL 200, we measure each projector for brightness, black level, luminance uniformity, color uniformity, and native color temperature. We take copious notes of all measurements based on the ANSI standard.

Once we have characterized all the units, we compare their performance with an eye toward mapping the projectors into the array such that each of the three soft-edge blend areas benefits from projectors with complimentary performance characteristics. Units demonstrating the greatest performance deviation will be mapped to one of the two side positions of the array. Units with significant performance deviation are brought back into spec with appropriate maintenance.

We now turn to studying the geometry performance of each projector/lens combination. When executing this test, geometry is evaluated with the projector installed on vertical and horizontal axis with the 16'×20' image at a throw distance that exactly matches the throw distance we will employ on site. Our goal is to verify we have no unusual geometry performance due to a faulty component or optic.

Although all three-chip DLP units offer significant lens shift flexibility, by definition, utilizing lens shift actually moves the light patch out of the center of the lens. With nearly all optics, geometric distortion is minimized in the center of the lens and increases toward the edge of the lens. Thus, our objective for all array applications is to install the units “on axis” with respect to the screen, both horizontally and vertically. If lens shift must be employed to meet mechanical requirements of the application, we take every effort to use as little as possible.

On Site

With all the prep work we have done, our initial time on site is spent locating the units to the positions defined on the array map, verifying the mechanical installation of each projector is on axis (or as close to on axis as possible), and making small tweaks to mechanical positions to assure optical geometry is optimized. Great care is taken with respect to leveling the screen and individual projectors, as well as to the alignment of the overlap area that comprises the soft-edge blend transition. A crosshatch pattern created by the image generator driving the array is the only way to be certain this critical mechanical, optical, and electronic alignment process optimizes the geometry of the soft-edge transition of the actual source.

With the projectors in place and geometry dialed in, it's time to set our sights on the final alignment of projector brightness, black level, and dark and white field color coordinates that equate to our target color temperature (defined by the customer, converted by us to specific X and Y values). To make direct comparisons and adjustments between projectors, we use our array image generator to create a multi-step grayscale pattern laid out so that each level of gray appears on every projector in the array. Depending on the adjustment being made, tweaking of projector contrast, brightness, lamp power, and RGB drives and gains may all be employed. Some projectors, such as Digital Projection's Highlite and Thunder series, offer software-based automated color balance capability. This feature can obviously simplify the color-matching process.

Whether automated or manual, to execute this critical final tweak it is best to use a high-quality luminance spot meter such as a Minolta CS 100. We use a spot meter instead of an incident meter because it is vital to judge, quantify, and fine tune the performance of the array from the perspective of the typical viewer or primary camera position. That perspective allows us to align array imagery that appears visually correct in the context of the event environment. Depending on ambient light, viewing positions, and screen selection, proper use of a quality spot meter can guide us to a very different final alignment than an incident meter would incorrectly achieve.

With the projectors completely matched, we turn to the image generator to finely tweak the subtle visual components of the soft-edge transition. As referenced above, adjustments for tuning the shape, colorimetry, and gamma of the contrast ramp may all be used. In our experience, these final adjustments, which make the continuous nature of the array visually believable, are best made under the critical judgment of a well-trained naked eye. As this is a broadcast application, we will verify the final result and make the final tweaks by viewing the array, as captured through the most relevant set camera, on a studio monitor.

Now we get to sit back and enjoy the fruits of our labor — an engaging, ultra-wide and large-scale image that, to the majority of the world, is simply there to serve impressive entertainment. The application of the technology is completely transparent.

At that point, those of us in the projected-images business recognize we have done our job. Even so, it is nice to know, when it comes to the displays we create, there is always more to the story than meets the eye.

Mike Levi is president of North American Operations for Digital Projection, Inc., Kennesaw, Ga., and is a veteran of the premium, large-screen display market. Email him at