[mythtv] automatic 14:9 or 4:3 zoom based on black bars detection?

Petr Stehlik pstehlik at sophics.cz
Sun Jul 23 19:36:56 UTC 2006


Peter Schachte píše v Ne 23. 07. 2006 v 22:19 +1000:
> Now to the OP's idea of a 14:9 zoom.  This could be put in as a new
> adjust fill option called 14:9, and would also work equally well in
> letterbox and pillarbox situations.
> 
> But there's one more option that would be worth considering; I'll call
> in "compromise."  This would zoom (and clip), stretch (distorting the
> aspect ratio) and leave black bars in equal proportions.  This is what
> my analog widescreen TV does by default for 4:3 programming, and it
> works pretty well.  When stretching 4:3 programming to a 16:9 display,
> it gives you black bands on the sides totalling about 11% of the width
> of the screen, it clips about 11% of the height of the source image,
> and it distorts the image by stretching it 11% horizontally.  In
> practice, you can barely see the stretching, you don't really miss the
> clipped part, and you hardly notice the narrow black bands on the
> sides. 

This sounds good (as another option next to non-distorted partial "14:9"
zoom). I suspect my 32" Thomson's "14:9" zoom actually stretches the
image horizontally a bit but I can hardly notice it.

> This approach should work equally well for displaying a 16:9
> program on a 4:3 display, and would even extend to showing a
> wider-screen DVD movie (maybe 2.something:1?) on a widescreen TV.

Yes, good point. 2.35:1 would benefit from this greatly.

Regarding your idea with separating the aspect mode and zoom mode to two
user settings - that's definitely the way to go (Daniel came up with it
anyway) since then the Zoom can be mapped to one key and even
unexperienced user could run through all modes with the "Zoom" button on
the remote without messing with the aspect ratio (that never got
misdetected in my experience thus I never needed to change it plus there
seems to be a bug that the aspect ratio override does not get reset when
changing channels).

Petr




More information about the mythtv-dev mailing list