Re: [stella] How many NTSC colors?

Subject: Re: [stella] How many NTSC colors?
From: Chris Wilkson <ecwilkso@xxxxxxx>
Date: Wed, 13 Jul 2005 19:03:37 -0400
On the 2600, all saturations are the same.  You have 7 non-adjustable
luma (brightness) values.  For color (hue), you have "black", "gold"
and 14 other values which are adjustable via the motherboard pot.

NTSC hue is encoded as a phase delay between a reference signwave and
the actual chroma sinewave.  On the 2600, "gold" is the reference.
The motherboard pot adjusts the delay between gold and the other 14
non-black colors.

It is possible to make the delay too large, so that the phase delay
for color 15 is equal to or more than 360 degrees.  When the delay
equals 360 degrees, the effect is that color 15 = color 1 = gold.
But there will be a 1 pixel delay to the right for color 15.  If the
phase is > 360 then color 15 could be equal to color 2, 3, etc...

Ideally, the pot is set so that there is 24 degrees of delay between
each color, so that color 1 = 0 degrees and color 15 = 336 degrees
and you have 15 unique hues spaced evenly around the colorwheel, plus
black (color 0) which is the absence of color.

In practice, many hobbiests and low-end repair shops and hackers will
adjust the pot so that color 15 = color 1 when compared on the screen
with the naked eye.  That's too bad, but it happens more and more as
people start following instructions that they find on the internet.

BTW, when adjusted correctly, there are twice as many yellows/golds
as when it's done the lazy way.  ;)


On Tue, 12 Jul 2005, Glenn Saunders wrote:

> Mathematically speaking none of the colors are
> identical as far as I know.  But after going through
> NTSC, some may seem to be.
> One reason I think has to do with the NTSC colorspace,
> which is "compressed".  It can show more unique colors
> in certain ranges than others.
> So even if the 2600 technically generates different
> colors for all 128, then after going through NTSC,
> some colors might appear so similar as to be virtually
> the same.
> The 128 colors don't really represent an evenly spread
> out choice out of all the possible 24-bit RGB colors,
> or even the available colors that NTSC can express.
> RGB can be expressed as (h)ue, (s)aturation, and
> (l)uminance.  The 2600 deals with hue and luminance
> but has fixed saturation.  The Amiga's palette
> introduced saturation.
> I think some of the colors that come out the other end
> looking the most similar are in the blue range.
> Meanwhile, there aren't really enough yellows/golds.
> This depends on how you adjust the internal hue pot in
> the 2600 and your TV's color settings.
> I'm sure someone like Chris Wilkson can give a more
> accurate explanation.
> --- Thomas Jentzsch <tjentzsch@xxxxxx> wrote:
> > I just thought there might be some technical reason,
> > why some of the
> > generated colors have to be identical (e.g. like
> > some of the generated
> > audio frequencies). No?
> __________________________________________________
> Do You Yahoo!?
> Tired of spam?  Yahoo! Mail has the best spam protection around
> Archives (includes files) at
> Unsub & more at

Archives (includes files) at
Unsub & more at

Current Thread