|
[ Pipeline ] |
|
PIPELINE ARTICLE 19960504
Title: Current Video Options on SGI Systems
Revision: 1.1
This article describes the current video options available
on Silicon Graphics machines. It assumes that the reader
has some familiarity with video concepts; it is designed to
help users decide which SGI video option best suits their
needs. The article provides an overview of the following
topics:
o Basic video terms and concepts
o Video encoding standards used in video transmission and
recording
o Video compression standards
o Video data transfer
o SGI video products and features
The research and testing for this article were performed
using IRIX 6.2. The concepts discussed here may also apply
to earlier versions of IRIX; however, not all of the
described products are supported on older systems. For
additional information on video options for older systems
or older versions of IRIX, readers should either contact
their sales representative or refer to the article entitled
"Video Options on SGI" in Pipeline, January/February 1994
(volume 5, number 1). This issue is available on-line using
InSight(1), if SupportFolio is loaded.
Basic Video Terms and Concepts
Alpha keying Keying of two frames (foreground
and background) based on the alpha values of the two
frames. The output pixel value is based on a formula using
the alpha values of the inputs. The formula may be varied
to produce different effects.
Alpha value Value of a pixel that specifies the pixel's
opacity, translucency, or transparency.
Blanking level The signal level at the beginning and end of the horizontal
and vertical blanking areas, typically representing zero output.
Chroma Keying Keying of two frames (foreground and
background) based on the chrominance (color) of the
foreground frame. Pixels whose chrominance value falls
within a specified range are replaced by pixels of the
background frame. Also known as chrominance keying.
Chrominance The color information in a video signal.
Specifically, it refers to the modulated subcarrier
sidebands in a composite video signal. It is also used to
describe the color difference signals, which carry the hue
and saturation values, in a component system.
Color burst A small portion of subcarrier inserted during blanking at a
fixed zero phase. It is used to decode the chrominance from
the subcarrier signal portion of the active video signal.
CCIR 601 The standard for digital component video
encoding. Specifies the valid color-space and sampling
frequencies. SMPTE 125M and SMPTE 259M define parts of the
digital video standard.
Component video An analog color encoding method that
separates color into three individual color signals, in RGB
or YUV format. The timing standards for the component
formats are closely related to the timing standards for the
composite formats.
Composite video An analog color encoding method or video
signal that contains the color, brightness, and
synchronizing information in one (composite) or two
(S-video) signals.
D1 Tape format for component digital video tape recorders.
Erroneously used to refer to CCIR 601 component digital
video. See CCIR 601.
D2 Tape format for composite digital video tape recorders.
D2 is not compatible with D1.
Field One scan of the video display, comprising one half
of a frame. In an interlaced video timing system (for
example, NTSC or PAL), one field consists of either the odd
or even scan lines.
Frame Conceptually the same as one frame of film, except
that a video frame is electronic in nature. In both the
NTSC and PAL standards, two fields comprise a frame.
Frame Rate The rate at which the frames come in to or go
out from the graphics/video subsystem. The frame rate
varies according to the standard or medium used:
o In NTSC, 29.97 frames comprise one second of video
o In PAL, 25 frames comprise one second of video
o In film, as in a movie theater, 24 frames comprise one second
Genlock Synchronization to another video signal or timing
source.
GPI General Purpose Interface. Used to communicate between
various video devices.
House sync A shared stable, synchronous source distributed
to many video devices at a site.
Interlacing Video transmission method that sends only an
image's odd or even scan lines at one time. Used by NTSC
and PAL. In an interlaced video format, only half of the
scan lines of each picture are transmitted at any one time.
See also Sequential scanning.
Luma keying Keying of two frames (foreground and
background) based on the luminance (brightness) of the
foreground frame. Pixels whose luminance value falls within
a specified range are replaced by pixels of the background
frame. Also referred to as luminance keying.
Luminance The video signal that describes the amount of
light in each pixel. Also referred to as luma.
NTSC National Television Standards Committee. Refers to a
television standard of 525 scan lines, interlaced at 60 Hz.
This form of video, used in the United States, Canada, and
Japan, is named after the organization that created the
NTSC television transmission system.
PAL Phase Alternate Line. Refers to a composite color
standard of 625 scan lines, interlaced at 50 Hz. This
standard is typically used in Asia, Europe, and Australia.
Pedestal See Setup.
RGB Red, green, and blue component colors used by graphics
systems and some video devices. A separate signal is used
for each primary color.
Sequential scanning Video transmission method that sends
each line of an image one after the other. Also referred to
as non-interlaced scanning. See also Interlacing.
Setup The difference between the blackest level displayed and the
blanking level. Used in NTSC, but not in PAL. Also referred
to as pedestal.
Subcarrier The constant-frequency component of a video
signal, added to convey chrominance information. The phase
and amplitude of the subcarrier encode the two color
difference signals, hue and saturation.
S-VHS Tape format for S-video analog video tape recorders.
S-video Video format where the Y (luminance) and C
(chrominance) portions of the video signal are kept
separate.
YCrCb The color difference format used in the CCIR 601
standard. Y is the luminance component, and Cr and Cb are
the color difference components. Cr and Cb are the same as
U and V except that they are scaled and offset to fit the
CCIR 601 color-space.
YUV Color difference format that uses a "subtractive" color
spectrum. Also referred to as Y (luminance), R-Y (red minus
luminance), and B-Y (blue minus luminance). Common YUV
formats include Betacam and M-II.
VLAN Video Local Area Network. Videomedia's proprietary
deck control protocol using 75 ohm video cable. Used
primarily to control video tape recorders.
VCR Video Cassette Recorder. Generic term used for the common
household video tape recorder.
VTR Video Tape Recorder. Generic term used for anything
from the common household VHS tape deck to advanced
broadcast-quality D1 decks.
Video Encoding Standards
Video encoding standards have diverged into two sets,
analog and digital. Analog and digital video both use two
major types of standards: composite and component video.
Analog video also uses an additional standard, S-video.
Analog
In analog composite video, luminance (Y) and chrominance
(C) are combined into one signal to facilitate
transmission. Because this format is commonly used by
television stations, all VCRs have adopted it. The
chrominance signal is filtered, then added on top of the
luminance signal. Although composite video is easier to
transmit, it has several drawbacks. The main drawback is
poor image quality, which is one result of limiting the
video bandwidth for luminance and chrominance to fit into
the video carrier.
In S-video, on the other hand, luminance and chrominance
are maintained as separate signals. Because the chrominance
signal does not have to be filtered or added to the
luminance signal, chrominance retains more of its bandwidth
than with analog composite video. Similarly, the resulting
video image is free of any "cross chroma" artifacts, which
occur when high frequency luminance transitions are
interpreted as chroma information.
In analog component video, each signal represents a
different color in one of two color-spaces: RGB or YUV.
This format is far superior to composite video mainly
because the signals have much higher bandwidth, thereby
conveying more color and image information. The most common
forms of component video for 525- and 625-line formats are
RGB and YUV. Two popular YUV formats are Betacam and M-II.
The RGB cables that feed the high-resolution monitor of a
Silicon Graphics workstation are also a form of analog
component video.
Digital
In digital component video, the analog component video
signals are sampled to produce a digital signal for
transmission. The components that make up the digital
signal are:
o Luminance (Y)
o Color difference signals (Cr, Cb), that are scaled, offset versions of the
Y, U, and V components of analog video
o The alpha channel (A), which is optional
Digital component video uses two major sampling formats:
4:2:2 and 4:4:4. When the alpha channel is used, these
become 4:2:2:4 and 4:4:4:4. These numbers represent the
sampling frequencies of the video components Y:Cr:Cb:A.
Therefore, the 4:2:2:4 format specifies that the luminance
and alpha channel are sampled twice as often as the Cr and
Cb data (see Figure 1).
- Y Y Y Y Y Y
|
scan | -- Cr -- -- Cr -- -- Cr --
lines | -- Cb -- -- Cb -- -- Cb --
|
- A A A A A A
Figure 1. 4:2:2:4 sampling frequency
The 4:4:4:4 format specifies that the luminance, alpha
channel, Cr, and Cb are all sampled at the same frequency
(see Figure 2).
- Y Y Y Y Y Y
|
scan | Cr Cr Cr Cr Cr Cr
lines | Cb Cb Cb Cb Cb Cb
|
- A A A A A A
Figure 2. 4:4:4:4 sampling frequency
Once the data has been sampled, it can be transmitted over
a parallel or serial interface. The parallel transmission
is carried on a 25-pin cable. The serial transmission is
carried on a standard 75 ohm video coaxial cable if it is
4:2:2 video, or two coaxial cables if it is 4:2:2:4, 4:4:4,
or 4:4:4:4 video.
With digital composite video, the analog composite video is
sampled; the resulting data is transmitted digitally rather
than as analog waveforms.
Video Compression Standards
SGI systems support many movie file formats, most of which
support some form of compression. Although it is outside
the scope of this article to discuss movie file formats, it
is worth noting that the only movie file format available
from SGI that currently has hardware support for
compression is Motion JPEG, or M-JPEG.
M-JPEG is a series of JPEG images used as a movie. JPEG is
a lossy compression algorithm, that is, the output image is
not identical to the input image. For most people, the
changes introduced by JPEG compression are not visible
unless they are using high compression rates. Cosmo
Compress for Indigo2 and Indy, as well as the IMPACT
Compression board for Indigo2 IMPACT, support M-JPEG in
hardware to allow real-time 30 fps (frames per second)
capture and playback of compressed video.
Video Data Transfer
This section assumes a working knowledge of the video
control panel (vcp(1)) and basic video concepts. Most
video boards offer two types of video transfer: to or from
memory and to or from the graphics subsystem. Video
transfers to or from disk must first go through memory.
Video transfers to or from the graphics subsystem require a
physical connection from the video board to the graphics
hardware.
Video Transfers to or from Memory or Disk
Video is composed of a continuous stream of individual
video frames. These video frames can be transferred to or
from memory using one of two basic methods: frame-by-frame
(single frame transfers), and burst transfers (continuous
frame transfers). If sufficient video bandwidth is
available, continuous frame transfers can become real-time
video transfers where no video frames are dropped during
the transfer.
Single Frame Transfers
IRIX 6.2 includes two programs for single frame transfers:
vidtomem(1) (Video To Memory/Disk) and memtovid(1)
(Memory/Disk to Video). Both tools store the video frame in
the RGB image format, by default. The capture(1) program
can also transfer a single video frame to disk. Many
video-editing software packages require precise,
frame-accurate control of single frame transfers to and
from a VTR. They generally use a VTR control protocol, such
as V-LAN, and an editing VTR that understands the V-LAN
protocol. Another method uses the GPI interface to control
single frame transfers.
Continuous Frame Transfers
The capture program in IRIX 6.2 is an example of continuous
frame transfers of video frames to disk. To perform a
continuous frame transfer, the video board is told to
continuously transfer as many video frames as possible, up
to the limit imposed by the hardware. Real-time transfers
occur when sufficient video bandwidth is available and no
video frames are dropped during the transfer. Some video
boards can achieve real-time transfers with optional
hardware. For example, the Cosmo Compress option for the
Galileo Video family uses JPEG image compression to reduce
the video bandwidth required for real-time transfers. With
the Cosmo Compression option, the Galileo family of video
boards can achieve real-time transfers of full-size video
frames using JPEG image compression.
Video Transfers to or from the Graphics Subsystem
When the video board is connected directly to the graphics
board, it can perform real-time transfers of video to
graphics (Live Video Input) or graphics to video (Live
Video Output). With this direct connection to the graphics
board, some video boards can do real-time blending of
graphics and live video input for live video output.
Live Video Input
The videoin(1) program in IRIX 6.2 provides live
video-in-a-window for those video boards that have a direct
connection to the graphics board. The capture program also
has the ability to display live video-in-a-window.
Live Video Output
The videoout(1) program in IRIX 6.2 provides real-time
transfers of graphics to video in two modes: full frame
(default) and full screen.
Full Frame mode uses a rectangular area of the graphics
screen that is the size of the video encoding standard
specified in the video control panel (vcp). The user can
move this area to the desired location on the graphics
screen.
Full Screen mode uses as much of the graphics screen as
possible, up to the limits imposed by the video hardware.
The screen is then filtered (scaled) down to the size of
the video encoding standard specified in the video control
panel. For example, for the Galileo Video family, Full
Screen mode is 1280 x 960 pixels for the NTSC encoding
standard.
Video Product Descriptions
This section describes the video products currently
shipping from SGI. Tables 1 and 2 summarize the
products described here. For information on older products,
see "References" at the end of this article.
VINO (Video Input No Output)
VINO is built into the Indy motherboard and is shipped with
an IndyCam on all Indy systems that have graphics. The
default video source for the capture program is VINO and
the IndyCam. This video board can achieve 30 frames per
second with half-size NTSC video frames using capture. The
image quality is considered to be very good for the video
professional; however, it is not broadcast quality. The
vino subsystem on the IRIX 6.2 CD-ROM (Part 1 of 2)
provides support for VINO.
Galileo Video Family
The Galileo Video (Express Video) family provides three
video boards: Galileo Video, Indigo2 Video, and Indy
Video. Although they are functionally similar, each one
has a different set of input and output video connectors.
The analog video image quality on these video boards is
considered to be of professional quality. Besides video
data transfers, they also support blending and keying of
graphics and video.
An IndyCam comes standard with Indigo2 Video for use with
InPerson. To use capture with Indigo2 Video and the
IndyCam, execute the command capture -ev1. For real-time
full frame transfers, the Cosmo Compression option is
available for the Galileo Video family. For digital 8-bit
broadcast image quality, the Digital Video (CCIR 601)
option is available.
The galileo subsystem on the Galileo CD-ROM provides
support for the Galileo Video boards. The cosmo subsystem
on the Cosmo Compress CD-ROM provides support for the Cosmo
Compression option.
Indigo2 Video for Indigo2 IMPACT
The Indigo2 Video for Indigo2 IMPACT is a redesigned
Indigo2 Video board that has new connectors to accommodate
Indigo2 IMPACT graphics. The Indigo2 Video for Indigo2
IMPACT has the same features and functionality as the
Indigo2 Video board, except that it does not provide
support for the Cosmo Compression option. An IndyCam comes
standard with Indigo2 Video for Indigo2 IMPACT for use with
InPerson. The impactI2V software subsystem on the IMPACT
Digital Media CD-ROM provides support for Indigo2 Video for
Indigo2 IMPACT.
Indigo2 IMPACT Video
Indigo2 IMPACT Video is a digital video board for the
IMPACT graphics family. Designed for post-production and
broadcast markets, the Indigo2 IMPACT Video board is a
10-bit component serial digital video board that interfaces
with CCIR-601/SMPTE-259M-compliant devices. With its direct
connection to graphics, it also supports blending and
keying of graphics and video. To support analog video,
optional third-party analog-to-digital and
digital-to-analog converters/encoders are available. For
real-time color-space conversion and video texture mapping,
the optional Indigo2 IMPACT Color Space/Video Texture
Mapping (CSCTEX) daughter card is available for Indigo2
IMPACT Video. The impactvideo software subsystem on the
IMPACT Digital Media 2.1 CD-ROM provides support for
Indigo2 IMPACT Video and Indigo2 IMPACT CSCTEX.
Indigo2 IMPACT Compression
Indigo2 IMPACT Compression is a stand-alone analog video
board with the ability to handle two JPEG streams (with
compression ratios as low as 2:1) for very high image
quality. The Indigo2 IMPACT Compression card can be
connected to Indigo2 IMPACT Video for added functionality.
The impactcomp software subsystem on the IMPACT Digital
Media CD-ROM provides support for Indigo2 IMPACT
Compression.
Sirius Video
Sirius Video is an option available for Onyx, CHALLENGE
L/XL, Power Series (pre-IRIX 6.2 only), and Crimson
systems. In a system with RealityEngine or InfiniteReality
graphics, Sirius can use the full power of the graphics
system to manipulate live video images. Sirius Video
supports real-time input and output in the full range of
broadcast video formats. Sirius Video supports blending and
keying of graphics and video in a number of ways,
including: alpha blending, chroma keying, and luma keying.
With the optional SD1 interface, Sirius is also capable of
serial digital component video.
References
"Video Options on SGI" in Pipeline, January/February 1994
(volume 5, number 1). This issue is available on-line using
InSight, if SupportFolio (formerly Support Advantage) is
loaded.
The following definitions apply to Tables 1 and 2:
EX Extreme Graphics
Express XS, XS24, XS24Z, Elan, and XZ graphics
High High IMPACT graphics
Maximum Maximum IMPACT graphics
Newport Indy 8-bit and 24-bit graphics
IN Capable of input only
IR InfiniteReality graphics
I/O Capable of input and output
OUT Capable of output only
RE RealityEngine graphics
RE2 RealityEngine 2 graphics
Solid Solid IMPACT graphics
STD Standard, part of the system hardware
VINO Video In No Out
XL Newport Graphics (24-bit Indigo2 graphics)
System | VINO Indy Indigo2 Galileo Indigo2 Video Indigo2 IMPACT Sirius
| Video Video Video for Indigo2 IMPACT Compression Video
| IMPACT Video
-------------|-----------------------------------------------------------------------------------------------
-------------|-----------------------------------------------------------------------------------------------
|
Indigo R4000 | Express
-------------|-----------------------------------------------------------------------------------------------
|
Indigo2 | XL, XZ, EX Solid, High, Solid, Solid, High
| Maximum High, Maximum
| Maximum
-------------|-----------------------------------------------------------------------------------------------
|
Indy | STD Newport
-------------|-----------------------------------------------------------------------------------------------
|
Crimson | RE*
-------------|-----------------------------------------------------------------------------------------------
Onyx and |
CHALLENGE |
L/XL | RE2/IR*
-------------------------------------------------------------------------------------------------------------
* Sirius can be used in a system without graphics, with a slight loss of functionality.
Table 1. System, Graphics, and Video Options
Video | VINO Indy Indigo2 Galileo Indigo2 Video Indigo2 IMPACT Sirius
Option | Video Video Video for Indigo2 IMPACT Compression Video
| IMPACT Video
-------------|-----------------------------------------------------------------------------------------------
-------------|-----------------------------------------------------------------------------------------------
|
Composite | IN I/O I/O I/O I/O I/O I/O
NTSC or PAL |
-------------|-----------------------------------------------------------------------------------------------
|
RGB | O* O* OUT I/O
-------------|-----------------------------------------------------------------------------------------------
|
S-video | IN I/O I/O I/O I/O I/O I/O
-------------|-----------------------------------------------------------------------------------------------
|
YUV | O* O* I/O I/O
-------------|-----------------------------------------------------------------------------------------------
|
Parallel | I/O* I/O* I/O
Digital |
Component |
-------------|-----------------------------------------------------------------------------------------------
|
Serial | I/O* I/O* I/O I/O
Digital | (with optional
Component | SD1 interface)
-------------------------------------------------------------------------------------------------------------
* Requires the CCIR 601 option.
Table 2. Video Formats
Maintained by sgips@beijing.sgi.com
