Greg's Sandbox

Production Video Techniques for Film Production

PRODUCTION GUIDELINES

From 1982 to 1996, Video Image was the leading company specializing in the production of video material for use as photographed imagry in motion pictures. Many specialized issues were developed over that 14 years of experience.

Over that time, we formulated many special considerations that should be adhered to if the best video image is to be photographed. The following guidelines are just some of these considerations.

The most common production method employed when generating imagery for filming is to generate the video or computer animation at the standard 29.97 frames per second rate. This is converted by our video department through a specially modified scan converter to our 24 frame per second format and recorded on specially modified 3/4" Umatic or Betacam SP videocassettes. These tapes are then played back on the set at 24.00 or 24.02 FPS, in sync with the movie camera.

When producing a screen graphic for eventual video tape playback, certain guidelines must be followed. They include considerations for line width, text size, color information, and scene length.

PHOTOGRAPHING INTERLACED VIDEO

Both 30 and 24 FPS video use an interlaced scan method. When a video display is synchronized with the motion picture camera, the camera is synchronized so it photographs only one field of video from each video frame, either the odd or the even. To understand why this is true, we should familiarize ourselves with the operation of a motion picture camera.

The motion picture camera is adjusted for a video shoot by setting the shutter at 180 degrees. The shutter rotates once for each frame of film exposed. While the camera is running, the shutter is closed while the film is being advanced to an unexposed frame, and it is locked down on registration pins. This takes 1/48th of a second. Next, the shutter opens, and the frame is exposed for 1/48th of a second. This process is repeated 24 times a second, as long as the camera is running.

When photographing a video monitor on which a 24 frame video image is displayed, the camera's shutter is open when one field is being displayed, and closed when the next field is displayed. This means the camera only sees half of the scan lines, or picture information, that is actually on the screen. Each time it is turned on, the movie camera will synchronize with either the odd or the even field, completely by chance.

SCREEN GRAPHIC DESIGN

What this means to the screen graphics designer is that if a horizontal line of a single pixel in width is used, it has a fifty-fifty chance of not being photographed by the camera. This is particularly evident when using fonts which are made up of single lines. The diagram below shows the effect of dropping every other line on two forms of the letter "E", one which has single pixel strokes, and another which has double pixel strokes. To ensure all lines are displayed, we design all our graphics with horizontal line widths of at least 2 pixels. We also do the same for vertical lines, but for a different reason.

COMPUTER GRAPHICS AND BANDWIDTH

Vertical lines, and all small details in general, are subject to a degree of degradation when converted from a digital format to a composite video signal. What you see, for example, on your computer screen is by no means what you will get when you record imagery on video disk or video tape! Small, high contrast details, especially small text and graphics, require a large change in the video signal at their edges. Computers were designed to handle small graphics, NTSC and other composite video modes such as 24 frame video are not. Since a computer works with digital data and high resolution displays, those small details are not a problem to display, but when we convert them to the analog world of composite video, problems arise. Consider the display of a single vertical line. On the computer screen, the waveform generating the display would be nearly perfectly straight and square, as shown below. A video recording of the same image would show a waveform that would be rounded and stunted as shown below, since the frequency response , or resolution, of the video system is not wide enough to accommodate such fine details. Therefore, vertical, as well as horizontal details should be drawn with least 2 pixels wide for good looking displays in the video world and for video playback.

LIVE ACTION VIDEO CONSIDERATIONS

Interlace has little effect on the photography of live action video. There is, however an overall reduction of the photographed video resolution.

In 525 line, 30 frame (actually 29.97) video, the maximum number of lines that can be photographed with a film camera running at 30 FPS with a 180 degree shutter is 486/2 or 243 lines.

In 655 line, 24 frame (either 24.02 or 24.00) video, the maximum number of lines that can be photographed with a film camera running at 24 FPS is 615/2 or 307 lines. Since most 24 frame video is converted from 30 frame video, the actual amount of information that can be obtained from the 307 lines of photographable video is still 243 lines. However, a sophisticated Standards Converter such as the ADAC used at Video Image actually extrapolates all 486 lines of 30 frame video to create the new 655 line video signal. Therefore, more of the original 525 line 30 fps video signal is made available for photographing in the 24 fps playback process.

HIGH FREQUENCY INFORMATION AND NTSC

Like small, one pixel details, high frequency information in an image may take on an unexpected appearance when photographed as a screen graphic by a movie camera. Think of a pattern of alternating black and gray horizontal lines. If the lines that make up this pattern are a single pixel in size, the pattern will either appear as all gray or all black, depending on what field the camera syncs with. One pixel size vertical stripes will interfere with the NTSC color information, causing distorted, or unintended, colors.

COLOR LIMITATIONS OF NTSC VIDEO

Most computer graphics are designed as 24 bit RGB images. There are certain colors in these images that are "illegal" NTSC colors, colors that cannot be accommodated by an NTSC system. Usually they are the brighter shades of certain fully saturated colors. Many programs, such as PhotoShop, have filters that can filter the colors in an image so that they fall within the legal limits of NTSC. Make certain you make use of those tools as you design your imagery. If in doubt, consult with the video department for an evaluation of the colors you are designing with.

In addition to illegal colors, the composite system used by NTSC and 24 Frame Video has a difficult time dealing with saturated red. This color range tends to become particularly blurred and smeary during playback. Avoid using text or fine details colored in saturated red hues. If the chroma, or intensity of the color is reduced, or if it is taken toward orange, much of this problem will be eliminated.

ULTRASCAN and VIDEO PLAYBACK

1. When graphics are displayed using the exclusive Video Image UltraScan (c) process, the composite video is converted to a computer-like, non-interlaced signal, and the monitor used for display scans every line of picture information in 1/48th of a second. Therefore, the movie camera sees all the visible scan lines for each exposed frame, instead of only half of them. Two pixel wide details are not required in this situation.

2. We prefer to use a computer on the set to display screen graphics, bypassing the video recording step altogether. In this case, the display is also scanning the entire image in 1/48th of a second, so we can display single pixel details here as well. These set-ups, which consist of both special hardware and software, are known as RTVGA (c), for IBM compatible computers, and RTMAC (c), for MacIntosh computers.

3. When certain Liquid Crystal Display devices are being filmed, like a laptop computer, or LCD projector. Many of these devices do not require synchronization with the movie camera, since their display does not decay between refreshes as does a CRT does. Single pixel details will also display properly on these devices.

It is often impossible to make a determination as to the final display set-up when one is designing a screen graphic. Therefore, it is always recommended to err on the conservative side, and use double pixel widths as a rule.

VIDEO SAFE AREA

Although all computers display the entire working area of the screen, a video displays, TVs and monitors, particularly those manufactured in the United States, usually are set up with a portion of the display hidden behind the frame, or bezel, which surrounds the screen. Therefore one should design all video imagery which is intended for video playback with an amount of bleed area, and keep the important information with in the "TV SAFE ACTION AREA".