This seminar targets those who have occasional need to transfer computer generated images to videotape, for use at conference presentations, client communications, etc., where NTSC encoding and VHS tape often are employed. Differences between RGB computer video and NTSC video standards are discussed. General guidelines are offered for designing images in a way that anticipates and avoids common problems with color and resoultion that can occur with the transfer of computer graphics images to videotape.

These notes are designed for a presentation situation in which simultaneous computer projection and NTSC video displays are compared to illustrate various artifacts introduced by NTSC encoding of computer graphics images. So, the effectiveness of many of the following illustrations is lost on the web.

Computer Video versus NTSC Video

• Raster displays
  
  

  
  
  

  • Rows of RGB phosphor triads
    
    
    
    
  • Proportionally excited by voltages
  • Colors are ``added'' at the emissive display

  
  
  

• Computer Video
  
  
  
  
  • Typically non-interlaced
  • Typically 60 refresh cycles per second or higher
  • RGB component signal throughout the system
  • Various spatial resolutions, increasingly in ``workstation'' range (ie. 1280 x 1024 pixels)
  • Various aspect ratios
  • Various color depths - the fewer the colors, the less subtle tones that are available

  
  
  

• National Television System Committee (NTSC) Video - 1953
  
  
  
  
  • Interlaced
  • 30 Frames / 60 Fields per second
  • Encoding formats
    • Component (Y, R-Y, B-Y; Y, Cr, Cb)
    • S-VHS (Y/C)
    • Composite (like standard VHS, among others)
  • 4w x 3h (1.33) Aspect Ratio
  • 640 x 486 - NTSC(525)
  • 720 x 486 - CCIR(525)
  • In practice:
    • The horizontal resolution of composite VHS NTSC television is in the realm of 300 lines
    • In the encoded signal, the luminance channel carries about twice as much information as the chrominance channel
    • Displayed colors will vary somewhat from monitor to monitor
    • The area of the image displayed will vary slightly from monitor to monitor

  
  
  

• Computer images to NTSC video
  
  
  • Other than technical compatibility, there are three main areas of concern when transferring computer generated images to video:
    
    
    • Spatial resolution/sampling - how many pixels are in your source image as compared to how much resolution is available in the encoded video signal
      
      
    • Color resolution - the gamut or range of colors that can satisfactorily displayed on your RGB computer monitor versus the more limited range that is available in NTSC displays
      
      
    • Timing and sequence - videotape is a linear medium that progresses at a constant frame rate. The constant frame rate needs to be considered in any timing decisions for animated sequences.
      

    
    
    

• Scan convertor/encoder
  
  
  • converts/encodes between analog signals
    
    
  • often an external box

  
  

• Framebuffer/encoder/capture
  
  • encodes video output signal from digital image memory
    
  • usually an internal card
    
  • often bundled with video capture capabilities
    
    • uses digital disk for storage
    • uses compression to enable real-time capture and display
      
      
      

• NTSC encoding/connection methods
  
  
  • Composite - the entire video signal is carried on a single cable (usually RCA or BNC connectors)
    
    
  • S-Video - psuedo component - luminace bandwidth is increased and shifted on the tape from the chrominance signal, providing more luminance detail and somewhat of a separation of the luminance and chroma portions of the image. Uses a special S-Video (4-pin) connector.
    
    
  • Color-difference Component (e.g. Betacam) - luminance and two color-difference channels are separated on three cables (usually BNC).
    
    
  • "Digital" options are emerging (e.g. Firewire, digital Betacam, etc.)
    
    
  • Ultimately, for VHS distribution, we still are dealing with composite NTSC when we get to tape.
    
    

  
  

• Recording to video
  
  
  • real-time versus single-frame recording
    
    
  • editing at the computer versus at the recording device (increasingly an option)
    
    
    

Spatial resolution design issues

• Keep critical image within video safe area
  
  
  • ``Action Safe'' = 5% margin all around
    
    
  • ``Title Safe'' = 10% margin all around

  
  
  

• Tiny text is hard to read (even in scales and legends)
  
  

• Don't use single pixel lines (don't forget subsampling from higher resolutions)
  
  

• Avoid fine patterns (again after any adjustments to resolution)

Color definition issues

• RGB primaries
  
  

• NTSC ``legal'' colors have a compressed chroma and luma range
  
  

• Avoid highly saturated colors (particularly reds) and sharp chroma transitions
  
  

• Adjacent areas of RGB complements show pronounced color bleeding
  
  

• Chroma problems are more apparent with colors of similar luminance value
  
  

• Chroma and luma work together

Animating time series data

• Timing decisions
  
  
  • Analysis versus presentation
  • Comprehension - to understand something visual, people need to see it
  • Boredom - to maintain attention, people need to see something happening
  • Leave text panels up long enough to be read out loud, at a moderate pace

  
  

• In NTSC, 30 frames (or 60 fields) per second is the constant around which all timing decisions revolve.
  
  
• Persistence of vision requires a minimum of about 15 frames per second for the perception of continuous motion (depends somewhat on context and nature of motion)
  
  
  
• When planning for animation, the smoothest motion will be obtained if sufficient images are rendered to fit desired viewing time.
  
  

Field Rendering

• 30 frame-rendered images / second
• 60 field-rendered images / second

• Field rendering provides additional motion increments, each composed of a 1/2 vertical resolution image displayed on either odd or even scanlines. Field rendering provides the smoothest possible motion in NTSC video, but problems can occur with field order or field dominance within your production piepline. These will appear as fluttery motion ("two steps forward one step back") or strong aliasing on edges that are moving vertically through the video raster.
  
  

• Often, a 1:1 image/playback ratio is not practical. For example, consider a desired sequence that has 240 time series images of a complex physical phenomenon
  
  
  • Shoot on 1s (i.e. display each image for 1 video frame)
    
    
    • 240 images / 30 frames per second = 8 seconds
    • Very smooth motion
    • Not much time to evaluate process depicted

    
    

  • Shoot on 3s (i.e. display each image for 3 video frames)
    
    
    • 240 images x 3 / 30 = 24 seconds
    • Not as smooth motion, but still fairly continuous
    • More time to evaluate process depicted

    
    

  • For multimedia movie formats, frame rate can serve the same function as shooting on multiple frames (e.g. 10 fps playback is the equivalent of shooting on 3s)
    
    
  • Dissolve through (proportional transition from one image to the next)
    
    
    • 1 frame each image
    • 5 frames each dissolve
    • (240 x 1) + (239 x 5) / 30 = 47.83 seconds
    • Smooth, but fuzzy, "motion"
    • More time to perceive transitions
    • Can the process fairly be assumed to be continuous?

Assembling your images

• Image conversion/moviemaking programs & utilities
  
  • e.g. Debabelizer
    
    
• Multimedia compositing / non-linear editing programs
  
  • e.g. Premiere, AVID Cinema, Composer
    
    

Some simple guidlines

• Work at video resolution (640 wide x 480 high)
• Simplify image content and layout whenever possible
  • Avoid redundant titles and labeling
  • Avoid unnecessary increments on axes or scales
  • Label items directly when possible (versus a legend)
• Keep saturation levels under 80%
• Keep luminance levels under 80% (not synonomous with HSB brightness)
• NTSC legal filters
• Avoid fine lines or textures
• Avoid small text (even in keys and legends)

Don't use lossy compression schemes in the middle of your workflow!

Planning for video

• Don't treat video presentation as an afterthought
• Do color and format tests during production
• Review both computer monitor and NTSC monitor during production if possible
  
  
  
  
• Try to maintain consistent lighting between your creation and presentation environments
• If using a service, allow plenty of time for transfer

Adjusting your video monitor

• ``Brightness'' or ``Black Level'' should be adjusted to appropriate black level
• ``Contrast'' or ``Picture'' can then be adjusted for overal display intensity
• ``Tint'' to obtain neutral gray
  

• Tint to match color bars

Test images (TIFF format) for download via http

• NTSC525 (640 x 486) Colorbars
  
  
• NTSC525 (640 x 486) Grayscale
  
  
• CCIR525 (720 x 486) Colorbars
  
  
• CCIR525 (720 x 486) Grayscale

CAC publically available resources

• CAC services
  
  
  • CAC Distributed System Group labs
    
    (http://dsg.cac.psu.edu)
    
    
  • Faculty Media Center
    
    (http://cac.psu.edu/fmc/)
    
    
  • Visualization Group
    
    (http://cac.psu.edu/viz/ or e-mail vizgroup@psu.edu)
    
    

• WWW resources
  
  
  • Network Computing Services, Inc. Video Guide
    
    http://www.msc.edu/msc/docs/video/

  
  
  

• Publications
  
  
  • Jackson, Richard and Lindsay MacDonald and Ken Freeman, Computer Generated Color: A Practical Guide to Presentation and Display, John Wiley & Sons, 1994.
    
    Available from Engineering Library.
    Call #: T385.J33 1994
    
    
  • Poynton, Charles A., A Technical Introduction to Digital Video, John Wiley & Sons, 1996.
    
    ISBN 0-471-12253-X
    
    Available from Engineering Library.
    Call #: T385.J33 1994

  
  
  

This file was last modified on October 7, 1999.