Quote:
Originally Posted by maurus
Quote:
Originally Posted by incredible
width= 480 -(16*2) = 448! (totally correct would be also here subtracting 8*2px as the "active"! px area at mod16 of 480x576 on a Tv is -8+480-8=464px!
height= 576 -(16*2) = 544!
result is 448x544 (432x544)
|
My resolution is 448x544. I not understand 432 value...
|
As I told above ...
In 480x576 Mode ... On Tv the effective Active Pixel area is 468x576! And thats a fact the same as If you would encode to 720x576 where in that case the effective Active Pixel area is 702x576
But as we do encode mod16 based these values now will be rounded to mod16
468x576 becomes therefore 464x576! at 0 overscan!
Because ONLY the 468x576 of the 480x576 will be stretched to 768x576 1:1 PAL square pixels the rest is out of bounds and WILL NOT be displayed, even if your TV wouldnt have an overscan area! Thats also the reason why it makes no sense to reencode a 720x576 to 720x576 for TV purposes as the active pix..area of that one is as said 702x576! in mod16 = 704x576.
Now lets add the overscan to 464x576
-16+464-16 = 432 pix width
-16+576-16 = 544 pix height
Your NEW active pix area is now 432x576 which will be "expanded" by just adding the black borders to:
480x576
And ... as the vertical alignment should be macroblock based:
576-544= 32 and therefore Mod16
therefore 16px borders at top and 16px borders at height will be applied
480-432= 48! and thats NOT Mod 16
but we at leas keep the width centered aligned so ...
24px borders at right and 24px borders at left will be added, where each of the 24px borders do have a macroblock (16px) inside which gives us a gain in compression as our mpeg encodings are macroblock based.
You dont trust me?? ok, ... its a lot of theory.
But in case of doubts do watch this site:
http://www.uwasa.fi/~f76998/video/co...nversion_table
(look at the 480x576 row in the PAL table and you can see at "actual active picture size" the 468x576 size
)
This site is a reference used in FitCD, Moviestacker and my MenCalc.
BUT FitCD i.E. uses a "resized" overscan where mine uses "overlayed" overscan ... which results in diff. results
Also FitCD & Moviestacker do FIRST crop, Then Scale and then adding borders, mine does FIRST scale, then crop and then expanding (adding borders) ... so the approch is different (the same way as discribed in the link above) ... but the result on TV WILL be the same.