VideoHelp Forum
+ Reply to Thread
Results 1 to 16 of 16
Thread
  1. 720x576, the usual deal, but it's only 4:3 content and the black bars are useless. What's the math on this? I don't want to just eyeball it and risk distortion if the picture is only meant to be over 704 pixels.
    Quote Quote  
  2. Member DB83's Avatar
    Join Date
    Jul 2007
    Location
    United Kingdom
    Search Comp PM
    720 * 576 pal 4:3 displays as 768*576.

    But surely black bars are normal if you display that on any screen that is wider than 4:3
    Quote Quote  
  3. Resize the frame to a 4:3 ratio, then crop the black bars. That will leave you with square pixels and the correct aspect ratio.

    If it's an ITU cap crop 16 pixels off the width, resize to a 4:3 ratio, then crop any remaining black bars.
    Quote Quote  
  4. Originally Posted by DB83 View Post
    720 * 576 pal 4:3 displays as 768*576.

    But surely black bars are normal if you display that on any screen that is wider than 4:3
    The bars are baked in. I want to get rid of them so it displays full-height on 4:3 and 16:10 displays, and co-operates with my 4:3 menu a bit better.

    Resize the frame to a 4:3 ratio, then crop the black bars. That will leave you with square pixels and the correct aspect ratio.

    If it's an ITU cap crop 16 pixels off the width, resize to a 4:3 ratio, then crop any remaining black bars.
    How does one tell if it's ITU? And is there not a more mathematical way of doing it, rather than just guessing?
    Quote Quote  
  5. Member DB83's Avatar
    Join Date
    Jul 2007
    Location
    United Kingdom
    Search Comp PM
    Full height ?

    That kinda tells me you have wide-screen video that was then capped to 4:3. Quite common back in the day when wide displays were rarer.

    But it might be more useful if you posted a sample.
    Quote Quote  
  6. ITU caps usually have about 8 pixels of black at the left and right edges.
    Quote Quote  
  7. Yes, full height. As in, filling the full height of the screen.

    It's a VHS tape from 1992. When it was digitised, it was for some reason digitised as a 16:9 image, so there are baked-in pillarboxes. Don't ask why, it wasn't me.
    Quote Quote  
  8. Originally Posted by jagabo View Post
    ITU caps usually have about 8 pixels of black at the left and right edges.
    But this has lots of pixels of black at the edges, because it's pillarboxed.

    Is that a no on there being a mathy way to work it out? Are there preferred resizers out there? It's in AviSynth at the moment, which has a lot of choices.
    Quote Quote  
  9. Originally Posted by koberulz View Post
    Originally Posted by jagabo View Post
    ITU caps usually have about 8 pixels of black at the left and right edges.
    But this has lots of pixels of black at the edges, because it's pillarboxed.
    Sometimes you can see a difference between the black pillarboxing and the ITU black at the edges (ie, they're not exactly the same intensity). In the absence of that you have to guess. Typically, anything made by capturing from analog tape is ITU. You'll never notice the 2 percent difference anyway.

    Originally Posted by koberulz View Post
    Is that a no on there being a mathy way to work it out? Are there preferred resizers out there? It's in AviSynth at the moment, which has a lot of choices.
    The resize/crop method given is mathematically precise. Use whatever resizing algorithm is appropriate for your material. For very sharp material I use BilinearResize(). For less sharp material I use Spline36Resize().
    Quote Quote  
  10. I've not really used resizing at all, so I have no idea about any of them.

    How is the resize/crop mathematically precise if I'm guessing at where the picture area starts and thus how many pixels to crop? Even zoomed in with VDub the edges look kinda fuzzy.

    EDIT: Hang on, your process for ITU is crop, resize, crop, resize, right? That seems inefficient, and isn't there extra quality loss from the double-resize?
    Last edited by koberulz; 10th Nov 2018 at 11:00.
    Quote Quote  
  11. Member DB83's Avatar
    Join Date
    Jul 2007
    Location
    United Kingdom
    Search Comp PM
    As I already said, that was not un-common back in the day.

    But if you want full height from 4:3 you must crop the sides and lose detail. So what you really need to do is crop the top and bottom (4:3 back to 16:9 is often 72 pixels top and 72 pixels bottom for pal and you might also have to de-interlace)

    And is not pillar-boxing but letter-boxing.
    Quote Quote  
  12. Why are you so determined to believe I'm wrong?
    Quote Quote  
  13. Member DB83's Avatar
    Join Date
    Jul 2007
    Location
    United Kingdom
    Search Comp PM
    Talking to me ?

    Pillar-boxing is bars on left and right. Letter-boxing is bars top and bottom.

    You already stated that you have a 16:9 video with bars from a vhs. Neither 720*576 or even 704*576 (both pal) are 4:3 and you simply can not get full-height from such picture without losing detail from left and right.

    But do not believe me since you know better.
    Quote Quote  
  14. Originally Posted by DB83 View Post
    Pillar-boxing is bars on left and right. Letter-boxing is bars top and bottom.
    I know.

    You already stated that you have a 16:9 video with bars from a vhs.
    Yes, pillarboxing.

    Neither 720*576 or even 704*576 (both pal) are 4:3
    ...what?

    and you simply can not get full-height from such picture without losing detail from left and right.
    Except that there isn't anything on the left or right to lose, except the pillarboxing bars.
    Quote Quote  
  15. Originally Posted by koberulz View Post
    How is the resize/crop mathematically precise if I'm guessing at where the picture area starts and thus how many pixels to crop?
    If the source is 4:3 DAR then resizing to a 4:3 frame size gives you exactly the right aspect ratio. Then, no matter how much you crop the content of what remains will have the correct aspect ratio. For example, draw a picture on a piece of paper. Then cut away some amount form the edges of the paper. No matter how much you cut away, what remains of the picture has the right aspect ratio, circles remain circles, squares remain squares, etc.

    If you want a specific frame aspect ratio just use that aspect ratio when you crop. It sounds to me like what you have is a 16:9 DAR source with a 4:3 picture pillarboxed in it. Resize to 1024x576 (1024 = 576 * 16 / 9) then crop down to a 4:3 frame size 768x576 (768 = 576 * 4 / 3).

    The general equation for aspect ratios:

    Code:
    DAR = FAR * SAR
    
    DAR = display aspect ratio, the final shape of the picture
    FAR = frame aspect ratio, frame width:height
    SAR = sample aspect ratio, the relative spacing of pixels horizontally:vertically
    When you have rendered the picture to square pixels SAR becomes 1:1 and the equation becomes:
    Code:
    DAR = FAR
    Last edited by jagabo; 10th Nov 2018 at 11:38.
    Quote Quote  
  16. Member DB83's Avatar
    Join Date
    Jul 2007
    Location
    United Kingdom
    Search Comp PM
    Ok. I probably mis-read what you wrote. Did not help the situation when you wrote 'full-height'

    So if I now read correctly you have a 16:9 image with hard-coded pillar-boxed black bars.

    Now here's the 'raw prawn'. By all means crop these away so you get a true 4:3 video. But display that correctly on a 16:9 display and the player adds soft-coded pillar-box bars due to the 4:3 DAR.

    Bottom line. Why bother ?
    Quote Quote  



Similar Threads

Visit our sponsor! Try DVDFab and backup Blu-rays!