A Little Pixel Knowledge Goes a Long Way

Over the past five years of teaching "seamless switching 101," I've discovered that the one basic concept that separates the A/V expert from the novice is the understanding of how total and active pixel counts work together. When you master this relationship, you can maximize the image quality of both the switching and display systems used for your event.

In extreme applications such as multi-projection blended images, the pixel count relationship is critical to producing a quality image. For example, if you set up a two-screen blend and you don't fully understand total and active pixels, the blend will not be seamless, and critical video might be missing. On the other hand, if you do understand the concept, your images will sing.

Active and Total
Most of us are familiar with the screen resolution setting on our PCs, which can be accessed by right-clicking on the desktop, and selecting the "Settings" tab in the Display Properties Window. Common resolution settings are 800 x 600, 1024 x 768, 1920 x 1080, etc. The "active" pixels are those that we see on the display device, such as the monitor or projector, and these pixels are part of the total pixel count that makes up a complete raster (the entire pattern of lines and pixels that comprise a video image).

For example, in the common 1024 x 768 format, the horizontal active pixel count is 1024 and the horizontal total pixel count is 1344. The vertical active count is 768 and the vertical total count is 806. The differences between the active and total amounts are those pixels which we don't see (or which we shouldn't see if things are adjusted properly). These "missing" pixels occur in the horizontal and vertical blanking intervals, which are essentially zones above, below, to the left and to the right of the active image in which synchronization and timing pulses occur. (Ever wonder why you hardly ever see a TV image rolling vertically anymore? It's because modern CRTs and displays do a "very" good job of vertical synchronization within the blanking interval.)

But I digress - the key point is that when you set up a digital display device that includes a scaler, you need to account for the "total" pixel count.

Setting up the Sampling
The best way to ensure that your display system samples the correct amount of pixels is to utilize a "Burst" or "Pixel On/Pixel Off" test pattern. When you route one of these test patterns to your display device, and when you simultaneously enable a "raster box," you now have a visual means to match the input signal to the display's native sample rate - on a pixel-for-pixel basis. Note that a raster box is a signal generated by many contemporary seamless presentation switchers. When enabled, a dotted border appears on your display that is precisely defined by the edges of the switcher's selected output resolution - in other words, the proper number of "active" pixels. If the output is 1024 x 768, the raster box encapsulates that format, and it provides a perfect representation of the first and last pixels of visible information.

Typically, the display's "Auto Acquire" mode measures the full input signal and adjusts it for a one-for-one sample. We've all seen this occur when the picture on our desktop LCD monitor shifts horizontally as the PC boots up. The LCD is doing its little "auto acquire" dance. There are times, though, when the Auto mode fails, and the switcher (or display device) activates an "over-sample" mode instead. In this mode, the unit samples more pixels than are actually present in the input signal (such as the pixels in the blanking interval), and then scales this over-sampled region onto the display's active raster. On screen, this result in artifacts, extra pixels, and the sound of unhappy clients.

For optimum picture quality, the goal is to match the input signal's pixels (e.g., 1 through 1024 horizontally, 1 through 768 vertically) to the exact edges of the raster box - nothing more, nothing less. When an input to the display device is sampled correctly, the raster box's edge perfectly surrounds the four edges of the displayed image or screen.

Testing the Pixel Relationship
In addition to an "Auto Acquire" mode, most display devices, image processing devices and seamless switchers provide you with lots of "handles" for adjusting the input's horizontal position (H POS), vertical position (V POS), horizontal size (H SIZE), vertical size (V SIZE) and total horizontal pixels (H TOTAL).

To quickly test if the active area is in the correct position in relation to the total pixel count, use the presentation switcher's H POS adjustment to move the image one pixel to the left or right. As you adjust, the raster box should "fall off" the screen on either the left or right. You can also use V POS to determine if vertical position is correct. However, in each case, if you have to move by more than one pixel to shift the raster box off screen, you'll need to use the H SIZE and V SIZE adjustments to perfectly match the input to the raster box.

Next, you can visually optimize the image using a "Burst" or "Pixel On/Pixel Off" test pattern. When your pixels are aligned properly and the display is not over-sampling, this pattern appears bright and flat. However, if the display is over-sampling, or if H/V size or position is misadjusted, the test pattern will appear to have noticeable bands of dark and bright areas. In this condition, you might have to use the H TOTAL adjustment in the offending device to minimize (or eliminate) the banding, until the image is perfectly flat and bright. To complicate matters, if the over-sampling functionality can not be disabled (as is the case even with many mid to high-end projectors), you'll have to interactively adjust all the H and V handles to optimize the image.

The Optimum Image
Proper understanding of "total" and "active" pixels, the use of the raster box and the use of a "burst" test pattern will net you a presentation that includes superior images. This knowledge will ensure that your system is displaying each pixel as perfectly as possible.