Re: Blurry picture

From: Segher Boessenkool <segher_at_kernel.crashing.org>
Date: Mon, 21 Nov 2011 07:05:25 +0100
Message-Id: <93FDA2D1-11F7-4A25-AFBF-85277B143167@kernel.crashing.org>
> I don't currently have a "good" explanation on why the checkboard
> patterns trigger colors. There's a 2-line averaging filter (a "comb
> filter") in PAL displays, that should average out any and all  
> colors of
> opposite phases. Adjacent rows of a checkboard are perfectly in
> "opposite phase". They still trigger colors - which shows that we  
> don't
> (yet) understand something.

The colour burst in adjacent lines is out of sync by 1/2 pi, that's
what you forgot to take into account.  The luma signal on the "stripes"
pattern is the same every line, but the decoded chroma signal from it
isn't!  But of course, after averaging, the end result is the same on
every line.
For the checkerboard, the luma is negated every other line, which means
after averaging the chroma has shifted 90 degrees relative to the
stripes pattern.
You can see what the colours look like without averaging on the very  
first
line of the pattern (it averages with the previous line, which is black,
which is sent out by the VIC as zero amplitude chroma).  You cannot
really see it on the line after the pattern, because it has almost zero
luminance (black again), hard to see anything there :-)


Segher


       Message was sent through the cbm-hackers mailing list
Received on 2011-11-21 07:00:03

Archive generated by hypermail 2.2.0.