What is a Video
A White Paper and
Introduction to Video Signal Processing
definition of a processor as it relates to this discussion is a device
that transforms a video signal or signals into something acceptable for
There are other
devices that share the term processor in their name, but this white
paper is specific to systems that drive single or multiple human viewed
video displays such as LCD screens, front projectors, DLP rear
projection displays, and the like.
There are many
other names that describe a Video Processor (referred to hereafter as
VP) such as Display Wall Processor, Display Wall Controller, Wall
Controller, Wall Processor, Video Server, Data Wall Processor, Data Wall
Server, Multiviewer, etc. We will refer to all of these as VP in this
There are 3 basic
categories of Video Processors:
Fig. 1 Single input,
Fig. 2 Multiple inputs,
Fig. 3 Multiple inputs,
This White paper
will focus on the multi input and multi output variety of these system
as single input or single output system will share some of the same
Processors are hardware devices that are built to order because the
number of inputs and outputs required are specific to a given
installation’s needs. In this white paper a number of input and output
complements will be shown and discussed, however the basic premise is
the same for most systems.
The most common use
of these Video Processors is to drive human viewed displays positioned in arrays
creating Video Walls. Other terms for these arrays are
Walls, Large Screen Displays, Visualization Environments, etc. Quite
often each display or screen in the array can be quite large, generally
between 40in – 120in each measured diagonally. When multiple displays this
large are tiled together it can create a very large overall display
hence the term 'Wall' showing video on it. (Width x Height is the
geometry order, 6 x 3 below)
Fig. 4 is a 6 x 3 Video wall array of 50in screens in a news room:
Video wall arrays
by simple multiplication have very large aggregate
As and example, if you have an array with 18 displays each with 2
million pixels, (referred to as 2MP) the overall display you have
created has 36 million pixels or 36MP. This is much larger than any
single video graphics chip can render, hence the need to gang multiple
graphics chips together in a single device that coordinates each pixel
relative to all others. This is the job of a Video Processor.
large physical and pixel count display arrays may come from the need to
display a single video signal across all available pixels in the array
or multiple video source signals each being shown on some portion of the
array, or a combination of both.
In our example
above of 18 displays with a total of 36MP, if we wanted to show a single
2MP image across the entire display we would have to stretch it 18 times
lager, or more properly scale it up to fit, also called
5 is an example of multi output processor Upscaling 1 image across a 4x1
Conversely if we
had 4 individual 2MP source signals we wanted to show in equal size on 1
of the 18 2MP displays in the above array we would have to shrink each
one 4 times, or Downscale it to fit. Upscaling uses multiple pixels in
the display array to represent 1 pixel in the source signal. Downscaling
must remove some of the source pixels through interpolation so that it
can fit into the desired display space. Full motion video signals are
less sensitive to scaling than text based images because of the human
brains ability to do its own interpolation.
It is quite common
for a video processor to perform up and down scaling functions at the
same time on different regions of the display array.
Fig. 6 is an
example of multi input and multi output processor doing Up and
Different from a
Processors differ from
matrix switches in 2 important ways,
A switch can accept source
signals at various resolutions and output those same signals to multiple
outputs at the same unchanged resolution. A VP can accept a source
signal then split that image onto multiple outputs each showing a
portion of the original signal or show multiple inputs one 1 output.
A VP will be driving the
display array at a constant pixel resolution, generally at the native
resolution of the displays used. (1920x1080, 1366x768, etc) The
resolution of the input signals can be totally different from each other
and the output resolutions since has the ability to do scaling.
Upon installation and configuration the output array geometry and
resolution of each output is entered into the operating system of the
Video Processor. The operating system internally creates a single large
logical pixel canvas that matches the display array in aggregate. When
an input signal is to be displayed on some portion of an output or
outputs, the Video Processor creates a window of some dimensions on the
pixel canvas. Then internally the input signal to be displayed is routed
and scaled to fit inside the window just created and sent to the
outputs. This process continues until all desired inputs being displayed
in the size and position desired by the humans viewing it.
Some simple single
input multi output processors that will only scale a single signal
across all outputs do not run software. These are dumb devices that
split the source signal into equal parts and send that portion to each
Most systems that
allow flexible window manipulation are generally carried out internally
with hardware for speed. This hardware in turn is controlled by some
sort of software generally consisting of an operating system and a user
interface that can be interacted with by human system operators.
Below in Figure 7
is and example of a user interface configured to drive a 4x3 array of
displays. On the left hand window pane the 16 input sources are arranged
vertically and indicate if there is a signal on that input, and of what
signal type is present by virtue of the green box around the icon that
indicated the type.
The input window is
represented by a wire frame and labeled by the input signal that is
being shown within it on the output array.
Fig. 7 is an example of a user interface showing input windows
represented by wire frames.
In addition to
hardware input signals being shown in windows, this system can open
other software applications such as Microsoft Internet Explorer in
this example, and make those applications part of the window complement
being sent to the outputs and being displayed.
8 is an example of the same user interface as above but showing the
input data in each window.
sizes and positions can be modified via mouse control by corner dragging
for size or left button hold down for whole window movement, similar to
any PC software application window. Window size and positions can also
be modified numerically by locating the X-Y pixel position of the top
left corner of a given window, and the size of same via a dedicated
configuration window as below. Many other properties of a given window
can be modified as shown in this screen shot as well.
Fig. 9 Shows the
display properties of a given input window including size and position.
There are a myriad
of other display details that are beyond the scope of this white paper
and will vary from system to system.
Saving of window
complements or Layouts is an important feature that allows system
operators to save time by not having to recreate the window
configurations and details about each window each time a system is
started. Multiple layouts can be saved and written to files so that a
single mouse click or command can be issued to recall those layouts
depending on the data that needs to be viewed at a given time.
control of Video Processor functions is an important feature that allows
external control systems such as AMX, Crestron or Layout Touch to
automate and simplify operation depending on the environment and
operators needs of same. Many of these systems utilize touch screens for
simplicity and intuitiveness of operation.
Below is a screen
shot of the Layout Touch software control system that allows operators
to recall layouts via any web page browser that has access to a video
processor. Each button has a layout name associated with it and by
selection of that button, the layout is recalled.
Fig. 10 is an
example of a touch screen program for recalling of layouts.
This white paper is
a work in progress and will be amended and modified from this form as
time becomes available to the author. Because of the vast range of
system configurations, signal types, user needs and environment
variables it is difficult to describe in detail all the possibilities
available in a single document. This document is intended as a primer to
the overall subject of video processing. Further in depth system details
quickly become germane to the maker of the devices in question. Specific
device details can be found by going to
This document is
authored by James Thornburg of Pixell and can be reached at