Anamorphic question?

Discussion in 'Archived Threads 2001-2004' started by Chris Pendergraft, Jun 24, 2001.

  1. Chris Pendergraft

    Chris Pendergraft Auditioning

    Joined:
    Jun 23, 2001
    Messages:
    7
    Likes Received:
    0
    I understand the basics of anamorphic mastering, having read the primer at digitalbits.com. I understand that the image is squished and then unsqueezed by the 16*9 television and that the benefit is more vertical resolution and a cleaner picture. My question is this- how is this achieved without stretching the image, much like the expand feature of some 16*9 televisions. My friend, who I tried to explain anamorphic mastering to, insisted that in order for the vertical resolution to be increased the picture has to be stretched and distorted in some kind of way. All of my anamorphic dvds look fantastic and any comparison screen shots I've seen present a flawless picture. Thank you, in advance, to any and all posts [​IMG]
    Chris
     
  2. Seth Paxton

    Seth Paxton Lead Actor

    Joined:
    Nov 5, 1998
    Messages:
    7,585
    Likes Received:
    0
    Well, the best way to explain it is to clarify something.
    The 16x9 TV is NOT unsqueezing, at least not in an active capacity. The TV does NOTHING to the image. All the TV does is scan the lines of video info from left to right as is normal.
    The key to anamorphic is that the information per scan line is sampled with the intention of it being scanned in a 16x9 area. So when you scan the lines in a 4x3 area they looked too tall.
    Also, to help understand (and ignoring any DSP/digital work along the way) a scan LINE is analog, meaning that if you use a scope to look at the electric signal you will see a continuous wave pattern within the scan line area. Now each line starts and ends with sync pulses, color bursts and other markers that say to the TV "hey, new line here".
    But the image info within those markers is one continuous wave, it's analog. That's how it can be stretched back and forth without resolution loss. Now there is a natural limit to that, being the limits on any part of the system to respond quick enough. Any electrical or chemical "analog" systems do have limitations to the amount of detail they can capture, so if you make the scan lines big enough (stretched very far) then you would be able to tell. Think of film, where info is actually captured by particles collecting together. That is a real limit, just as pixels are.
    However, pixel-limits are hit MUCH sooner so the analog limits just don't come into play. [​IMG]
    Now the scan lines themselves, those ARE samples, in the vertical direction we DO HAVE SAMPLING which is in effect a digital, or at least discrete, system. And it's very limited sampling compared to the analog waveform of the lines themselves.
    480 total lines for NTSC, 1080 lines for ATSC (HDTV).
    So we want to try to use all of those 480 "samples" to capture picture info. Think of a picture placed on a table. Then think of drawing that picture with only 10 lines - meaning the vertical info can only have 10 different values if you followed a straight line down the picture. You may have a picture of the proper size, and as you go from left to right things would be "ok". But as you went downward you would basically have single color blocks.
    Make it 100 lines and you capture more detail, 200, 300 and so on.
    So anamorphic was someone's idea to use MORE of the sampling lines to capture MORE of the vertical info on pictures that were wider than 4:3. Knowing of the coming of 16:9 sets, an anamorphic video standard was created in which the info was captured with the intent of scanning it back out in a 16:9 area.
    Since that area would be WIDER than 4:3, that meant that for pictures wider than 4:3 the area would also be TALLER. And a TALLER picture means one that crosses more scan lines. So if you capture with the intent of showing it in a 16x9 space then you can capture with these extra sampling lines that you will be later scanning on.
    For NON-ANAMORPHIC pictures to be zoomed to fit a 16:9 set, an internal DSP chip will sample the existing LESSER lines of video and create FAKE new lines to get to the total number of lines the picture should have to fit the screen. That means less REAL info that an ANAMORPHIC picture would have.
    Say a non-ana capture the picture with 350 lines, and the picture will be scanned on a 16:9 set over 470 lines (which an anamorpic picture would use). Those 120 "missing" lines must come from someplace and that place is the DSP which samples the original 350 and creates a totally new set of 470 lines, many of which contain no real details from the original picture but simply DSP "guesses" at what the info might have been.
    Obviously the ZOOM function to make a non-anamorphic picture cover the same area as an anamorphic one IS an active effort by the TV. And THAT'S what we are trying to avoid.
    But the normal process of scanning a line is "inactive".
    Just let your 16:9 TV scan a normal 4:3 picture. There is no "new or fake" info being created. It's the exact same info simply being shown on a wider space. Like stretching a comic that you captured with Silly Putty. [​IMG]
    And yes, while your CPU or other video source may start with pixels in the horizontal direction, those pixels are turned into an analog line for scanning by a CRT.
    With DLP and LCD projectors we run into new problems in matching both scan lines and horizontal samples to the discrete pixels that can be shown. But even then anamorphic helps as it still represents capturing the data with MORE sampling lines than non-anamorphic does. And more sampling lines always means more vertical info to start with. You can always throw out info, but you can never add "real" info.
    Of course with 4:3 pictures, both methods use all 480 lines.
    Also, "sqeeze modes" on 4:3 sets involve making the 4:3 into a 16:9 by simply scanning all 480 lines in a 16:9 space on the tube. That means the rest of the tube goes literally unused. But that's the scan lines in that case are doing the exact same thing as they do on a 16:9 set, that is they are scanning a 16:9 area. The only difference is that a 16:9 TV has a tube that matches that area, whereas a 4:3 set has some "extra" glass above and below that area. [​IMG]
    Potentially, you could have a variable anamorphic system in which ALL SCAN SAMPLES were used no matter what the aspect ratio of the original picture. Then it would be up to the video display system (like your TV) to scan the lines in the right space. You certainly could "squeeze" a 16:9 TV just like a 4:3 TV if you had a DVD sampled for a 2.35 squeeze, or 480 lines intended to be scanned over a 2.35:1 area. Then you would ALWAYS be maximizing your vertical resolution.
    Good idea that will not be happening anytime soon. [​IMG] And not just because of stubborn business, but also due to potential technical problems that would come with some new spec.
     
  3. Seth Paxton

    Seth Paxton Lead Actor

    Joined:
    Nov 5, 1998
    Messages:
    7,585
    Likes Received:
    0
    I should have mentioned that with MPEG 1& 2 systems such as DVD, Digital Satellite, and HDTV the video is also sampled in the horizontal direction, and then those samples are turned into analog waves for transporting (via S-vid, Component, or Composite) and then scanning on a CRT.
    While the particular systems follow specs and limitation to how much to sample, they can always sample as much as the system can handle since they will become analog waves eventually.
    And also I made it sound confusing, but for most systems I know of, even LCD and DLPs take IN these analog waves (S-Vid, etc) and then convert them back to pixels. So for them it's analog -> sampled for DVD to pixels -> coverted back to analog by DVD player -> converted back to pixels for LCD projector.
    People like to scale the samples used to make these analog waves so that the pixel count being used matches the number of pixels per line on the LCD. Or better yet try to find methods for never converting out of pixel mode (digital) in the first place and using a DSP to scale the pixels to exactly match the pixels in the display.
    All of that is more advanced concerns that just the basic anamorphic app that you were discussing.
     

Share This Page