T436 - Fall 2007 - Week 7

Readings: Blain Brown, chapter 12 "video & HD "

Video & color sampling

Agenda:

review quiz
cover video
reality check

Notes:

• Production applications for Spring 2007 are now on-line!
• The production schedule is online. I'll be making some changes in weeks 3-7 to make sure everyone gets an even dose of field and studio. (right now it's unbalanced)
• We still need to decide on a few more projects later in the semester. In other words we need a few more workable or polished scripts.

Video signal

Experimental broadcasts began in the US in the late 1930s. The NTSC was established in 1940 and came up with the first set of standards in 1941.

Called for 30 frames per second with 2 fields. 4:3 (1.33) aspect ratio came close to matching existing 16mm and 35mm film formats, which used the Academy Aperture (11:8 or 1.375 aspect ratio).

On a cathode ray tube (CRT) display, the image is created by an electron beam, which excites phosphors on the face of the screen. The electron beam scans each row from left to right, and then jumps back to draw the next line. The excited phosphors on CRT displays decay quickly after the electron beam makes its sweep. Because of the decay, images displayed at about 30 frames per second, presented a noticeable flicker. In order to reduce the flicker, the display frequency had to be increased. To achieve this, the frame was broken down into two fields. The first field displayed only the odd lines while the second displayed only the even lines. So instead of drawing approximately 30 frames per frame, interlacing uses two fiel

People began to want color TV.

In order to broadcast in color, the original NTSC standard for B & W television had to be revised. NTSC updated it in 1953. Creating the new standard was no easy task as engineers had to make color broadcasts backward compatible with the large base of existing black and white televisions. (10 million sets had been sold by 1949.) To do so, engineers split the signal into two components, luminance, referred to as luma, which contained the brightness information, and chrominance, which contained the color. The color information was encoded onto a 3.58 MHz subcarrier added onto the video signal. Black and white sets could ignore the color subcarrier using only the luma portion, while color sets could take advantage of both. Unfortunately, the color subcarrier interacted with the sound carrier creating minor visible artifacts. In order to reduce interference, the field refresh rate of 60 Hz was slowed down by a factor of 1000/1001 to 59.94 Hz. So instead of running at 30 frames per second, broadcast television downshifted to 29.97 frames per second.

Composite video signal

• Horizontal sync pulse (start of every scan line)
• Horizontal blanking (beam turned off as it returns to draw another line)
• Vertical sync pulse & blanking (beam turned off to return to top. Can insert data here)
• Color reference (3.58 MHz)
• Reference black level (7.5 IRE for NTSC , 0 for digital)
• Luminance
• Saturation information
• Hue information

HDTV

A number of industry associations, corporations, and educational institutions formed the Advanced Television Systems Committee (ATSC) in 1982. The ATSC is a not-for-profit organization that develops voluntary standards for advanced television systems (www.atsc.org). Such advanced systems include enhanced analog TV, digital TV (DTV), standard definition TV, high-definition TV, and data services. The ATSC’s published broadcast standards are voluntary unless adopted and mandated by the FCC.

In December 1996, the FCC adopted most of the standards proposed by the ATSC, mandating that broadcasters begin broadcasting digitally. According to the ATSC, within one year of the November 1, 1998 rollout, more than 50 percent of the US population was in a position to receive digital broadcasts. During a transitional period, television would be broadcast both digitally under the FCC’s digital terrestrial television (DTT) guidelines and through traditional analog means. At the present time, Congress has voted to terminate analog broadcasting by February 2009, though the deadline could be extended.

Standard definition television (SDTV) can use either the 4:3 or 16:9 aspect ratios, HDTV always uses the 16:9 aspect ratio.

 

HDTV/SDTV	Horizontal lines	Vertical lines	Aspect Ratio	Frame Rate
SDTV	640	480	4:3	23.976p, 24p, 29.97p, 30p, 59.94p, 60p, 59.94i, 60i
SDTV	704	480	4:3 and 16:9	23.976p, 24p, 29.97p, 30p, 59.94p, 60p, 59.94i, 60i
HDTV	1280	720	16:9	23.976p, 24p, 29.97p, 30p, 59.94p, 60p
HDTV	1920	1080	16:9	23.976p, 24p, 29.97p, 30p, 59.94i, 60i

While HDTV content is designed to fill a 16:9 frame, the display of programming from other sources with varying aspect ratios is also possible. Programs shot in the 4:3 aspect ratio or in wider, cinematic formats can easily be displayed inside of a 16:9 frame without distortion by shrinking the image. Unfortunately it’s quite common to see broadcasters delivering images with the improper aspect ratio (Example A of figure 2.3). Traditional, 4:3 content is ideally viewed on widescreen displays by presenting the image as large as possible, centered within the frame. (Example B) This is sometimes referred to as pillar boxing. This allows the original image to be seen as it was intended. Some broadcasters magnify the 4:3 image so that it fills the entire 16:9 frame. (Example C) This can often be identified by the lack of headroom. Content from cinematic formats with wider aspect ratios can be accurately displayed within the 16:9 frame with letterboxing. (Example D) It’s also frequently necessary to present widescreen programming inside of traditional 4:3 displays with letterboxing.

Content with varying aspect ratios displayed within a 16:9 frame.

 

Color

Computer-based digital imaging systems typically operate in an RGB color space or a variant of it, while broadcast video transmission adopted a color difference model. This was not only because the signal had to be compatible with existing black and white televisions but it also had to take up as little bandwidth as possible. Video cameras capture images into an RGB color space via three CCDs or CMOS (complementary metal oxide semiconductor) sensors. Initially captured in uncompressed form, the RGB values are processed and converted into a color difference mode.

In the color difference system, the color signal can be numerically represented with three values: Y, B-Y and R-Y. Mathematically, Y represents the value of the luma portion with B-Y and R-Y representing the two color difference values. The formulas used to derive the color difference values vary depending upon the application. YPbPr uses a slightly different formula optimized for component analog video, while YCbCr uses a different scaling factor optimized for digital video.

Humans are more sensitive to spatial detail in brightness than in color information. Because of this, most of the important detail needed to comprehend an image is provided through the luma portion of the video signal. Engineers found they could throw out more than half of the color information and still get pleasing results. Compared to RGB, Y,B-Y,R-Y can store color data in a smaller amount of space and thus use less bandwidth when broadcast.

Color Sampling

Unless working in an uncompressed RGB mode, the color signal is converted into a color difference system. After converting the RGB, the signal is sampled, quantized, compressed (usually), and then recorded to tape, hard drive, optical disk, or in some cases a memory card.

Color sampling figures convey the manner in which the luma and color components are sampled for digitizing and are typically presented as a ratio with three figures (x:x:x). The first figure is usually four and refers to the number of luma samples. The second two figures correspond to the number of samples for the two color difference signals. For instance, DV’s 4:1:1 states that for every four luma samples, only one sample is taken for each of the color difference samples. A 4:2:2 format (such as DVC Pro50 or digital Betacam) means that for every four luma samples taken, two samples will be taken of each of the color difference signals. A 4:1:1 format would record half the color information that a 4:2:2 format would. When a codec is represented by a 4:4:4, it is typically referring to an RGB signal.

The 4:2:0 color sampling format comes in a few different variants. As usually employed in MPEG-2, the color difference signals are sampled at half the rate of the luma samples, but also reduced in half, vertically.

While formats using lower color sampling ratios require less bandwidth, those with higher sampling ratios are preferred for professional editing, keying and compositing.

Pulldown

A common frame conversion task is required by the frequent need to change 24p content into 60i. Such is the case when converting film (which runs at 24 fps) into 60i. Sometimes called the telecine process, it’s also required when changing 24p video into 60i. Some systems employ a 2-3 pulldown, which while reversing the order achieves the same end result.

The basic idea between the 3-2 pulldown is that 4 frames of 24p footage are converted into 5 interlaced video frames. It’s called 3-2 (or 2-3) because each consecutive 24p frame is transferred into 2 fields followed by 3 fields, then 2 fields, etc. One of the steps is to slow the film down by .1% to the rate of 23.976 frames per second. In the example below we have 4 frames of 24p material, labeled A, B, C, & D. The first video frame contains two fields of frame A. The second video frame contains one field of A and the second field of B. The third video frame contains one field of B and one of C. The fourth video frame contains two frames of C. The fifth video frame contains 2 fields of D.

Illustration 2.4 The 3-2 Pulldown. Illustration courtesy Tabletop Productions.

Setting Up a Monitor

(refer to book for various values)

Note that standard SMPTE bars will appear slightly different on an HD monitor compared to the downconverted image on an NTSC monitor. On an NTSC monitor the two small grayscale stripes will blend in as they are 3.5 and 7.5 IRE. Since HD monitors show us digital (0 IRE black level) we should be able to see the difference between the two.

 

 

Back up to the T436 index