Quote:
Quote:
The S-Video/composite board should have the color carrier rotate 180 degrees every scanline and switch off every frame
That's actually the point of 20th scanline being short by one pixel. (I'd point you at the wiki but it's down for me).
Actually, being short by one pixel only rotates the phase by 240 degrees, not 180, since the CPU and the color carrier oscillate at a ratio of 2 color carrier cycles to 3 PPU cycles. 3.58/5.37 = 2/3. (2/3)*360 = 240. Visually:
Code:
Original color carrier:
--- --- --- ---
___ ___ ___ ___
CPU clock:
-- -- -- -- -- --
__ __ __ __ __ __
Color carrier -1 CPU cycle:
(4 dashes)
- --- --- --- --
___ ___ ___ ___
Original color carrier:
(again) --- --- --- ---
___ ___ ___ ___
Notice how the original and phase-shifted color carriers' clock edges do not exactly match up.
Look at Nestopia with the NTSC filter on and you'll notice that color and luminance decoding errors are still noticeable on still images, which they wouldn't be if the phase rotated by 180 degrees. For instance, if one white pixel surrounded by black pixels created interference in the chroma bandwidth that caused it to be yellow tinted, it would be blue tinted in the next frame if the color carrier shifted 180 degrees. Since yellow and blue are exactly opposite each other, when averaged out by flickering, the pixel looks colorless (white). However, if the color carrier shifts 240 degrees, the pixel would be cyan tinted in the next frame, and would look light green tinted when averaged out.