1. Sign-up to become a member, and most of the ads you see will disappear. It only takes 30 seconds to sign up, so join the discussion today!
    Dismiss Notice

S-video Color Resolution

Discussion in 'Archived Threads 2001-2004' started by Gary_Lar, Feb 11, 2003.

  1. Gary_Lar

    Gary_Lar Auditioning

    Joined:
    Feb 10, 2003
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    0
    I am a liitle confused. The bandwidth for S-video color resolution appears to be constrained to 1.5 MHz, but the Bandwidth for luminance appears to be unlimited. What is the reason for this?
     
  2. TimTurtino

    TimTurtino Stunt Coordinator

    Joined:
    Dec 17, 2002
    Messages:
    156
    Likes Received:
    0
    Trophy Points:
    0
    (I'm assuming the above to be partly correct, and giving a plausible reason.)

    Because the human eye is much less sensitive to color than brightness, usually brightness information is given much more bandwidth than either color (in S-Vid Y/C) or color difference (component video, Y, Y-R, Y-B) information. This is simply because we allocate more bandwidth for the places where we can perceive more information.

    Also, the bandwidth for luminance is not, I believe, unlimited. I believe that it's on the order of 3-6 MHz....

    Me
     
  3. Gary_Lar

    Gary_Lar Auditioning

    Joined:
    Feb 10, 2003
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    0
    A DVD player can output a color resolution of 360 MPA (Maximum Pixels Across or max pixels on a scan line) which is half that of the luminance resolution of 720 MPA. The bandwidth needed for this color resolution on an NTSC television is approximately 2.83 MHz. I believe that a component hookup can handle this bandwidth, but I am not positive about this. For an S-video hookup, all the info I have indicates that the color bandwidth is limited to about 1.5 MHz which translates into about 190 MPA. Does this 1.5 MHz limit really exist for an S-video connection, and if so what is behind the limit? Just curious on how things work.
     
  4. TimTurtino

    TimTurtino Stunt Coordinator

    Joined:
    Dec 17, 2002
    Messages:
    156
    Likes Received:
    0
    Trophy Points:
    0
    But isn't the bandwidth halved again because there are half as many color lines up and down? Or does that just happen in the encoding?

    Sorry for the simplistic answer above-- you're clearly asking a smarter question than I can answer.

    (Oh, but a component hookup can definitely handle it-- from some of the earlier threads, even a poorly designed component switcher's -3dB point is around 27 MHz)

    Me
     
  5. Gary_Lar

    Gary_Lar Auditioning

    Joined:
    Feb 10, 2003
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    0
    Tim, thanks for the reply. I'm just starting to learn about the technical aspects of video in general, so I am not sure if I am asking the right questions, or even if my math above is correct. I think the bandwidth needed for the luminance signal from a DVD player is about 5.66 MHz. This bandwidth provides 720 lines of luminance resolution: (720 x 29.97 x 525)/2 = 5.66 MHz. Please correct me if my terminology, math, or reasoning is incorrect. For DVDs, the color resolution is one-half the luminance resolution, so the bandwidth needed would be 5.66/2 = 2.83 MHz. I believe this number corresponds to 360 vertical color lines. The numbers above are based on the full width of the TV scan line.
     
  6. Allan Jayne

    Allan Jayne Cinematographer

    Joined:
    Nov 1, 1998
    Messages:
    2,404
    Likes Received:
    0
    Trophy Points:
    0
    The chrominance part of S-video is modulated onto (for NTSC) the approx. 3.58 MHz subcarrier. I believe although I am not absolutely sure that the maximum bandwidth of the color components so modulated is 1.79 MHz (half of a "nyquist frequency" of 3.58 MHz). Or maybe 1.79 MHz is the maximum bandwidth that can be demodulated with not too outrageously expensive circuits and 1.5 Mhz is all that can be demodulated using quite inexpensive circuits(?).

    The 720 MPA (540 TVL @ 4:3) luminance resolution and 360 MPA color resolution for DVD are the standards. For component video, the bandwidth of the Pb and Pr is theoretically unlimited, note that 1080i HDTV has 960 MPA color resolution.

    About the math: 720 MPA for 480i interlaced requires 6.75 MHz, for 360 MPA chrominance resolution it is half of that or 3.38 MHz. For 480i and 480p the visible portion of the scan line is about 83% of the total scan line (different for 1080i), the rest is the vertical retrace interval.

    Although for DVD every two scan lines share the same color, by the time the video is decompressed to 480i S-video or component video, or 480p component video, the color has to have been duplicated so each scan line has its own, then the full 3.38 MHz (480i) is needed for each of the Pb and Pr lines. Having every two scan lines share the same color and also with half the max. pixels across for color, and also many compression processes, drastically reduces the amount of digital data needed, sometimes expressed in 'kilobits per second', I don't know the figure for DVD.

    Video hints:
    http://members.aol.com/ajaynejr/video.htm
     
  7. Gary_Lar

    Gary_Lar Auditioning

    Joined:
    Feb 10, 2003
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    0
    Allan, thanks for the information. I knew that the 3.58 MHz color subcarrier was an NTSC broadcast requirement, but I was not certain that the requirement carried over to an S-video connection. Is there an industry standard that discusses the parameters for the S-video signals?

    The math: When the 0.83 factor is included in my math, the bandwidth becomes 6.82 MHz (720 x 29.97 x 525)/(0.83 x 2.0). Is the difference between 6.75 MHz and 6.82 MHz due to rounding of numbers, or something I have missed.
     
  8. Allan Jayne

    Allan Jayne Cinematographer

    Joined:
    Nov 1, 1998
    Messages:
    2,404
    Likes Received:
    0
    Trophy Points:
    0
    I am quite sure there is an industry standard for S-video which includes the 1.5 MHz for the color components that go into the C part of S-video. I don't know where it is published.

    About the active portion of the scan line being 83%, that is approximate and it is a rounding issue. The number 858 stands out in my mind as the total number of pixels in the scan line (360 dark/light pixel pairs in the active portion plus 69 duty cycle sync. pulses and separating spaces of the same size in the retrace interval).
     
  9. Gary_Lar

    Gary_Lar Auditioning

    Joined:
    Feb 10, 2003
    Messages:
    5
    Likes Received:
    0
    Trophy Points:
    0
    Allan, thanks for your AOL web pages on various video topics, and the replies above. They are all very helpful and informative.
     
  10. Cees Alons

    Cees Alons Moderator
    Moderator

    Joined:
    Jul 31, 1997
    Messages:
    19,474
    Likes Received:
    340
    Trophy Points:
    9,110
    Real Name:
    Cees Alons
    Gary,

    Your little math is almost correct - but you didn't take into account the 1/2 in both directions, so you have to divide by 4. DVDs (and mpeg-2) use a colour encoding known as 4.2.0, which essentially make good use of the limited resolution inside our eyes (the rods and cones) for colour.

    And because DVDs are digital, they don't need a carrier or subcarrier signal.

    Cees
     

Share This Page