Re: [stella] Optimized color reduction approaches

Subject: Re: [stella] Optimized color reduction approaches
From: "Andrew Davie" <atari2600@xxxxxxxxxxxxx>
Date: Wed, 12 Feb 2003 20:07:32 +1100
> I think this is how the COLRVIEW technique as used with GR8 on the
> Atari8-bit.  Is that what you are trying to emulate?  It just like a 1-bit
> version of COLRVIEW.

It may be similar, though I am not specifically trying to emulate this.
I've been interested for a long time in ways of improving the colour
capability of '2600 bitmap screens - as far back as my PUSH demo in '98 or
'99 I was experimenting with time-based interlacing to give more colours

As with many things in technology, sometime it's simply the right TIME for
something to happen - and several people may independantly be working on the
same thing without knowing about each others' efforts. In this case, I was
attempting colour imagery with 6-sprite routines before I became aware of
the "high resolution bitmap" thread on AtariAge which asked if such a thing
were possible.   I contributed to that thread for a while, but am rather
annoyed at the moment with how it degenerated to a personal thing.

It's highly likely that I got some of my ideas from that thread - but bear
in mind that I was already thinking of techniques and discussing the same
with others.  My efforts did not flow as a direct attempt to copy the
techniques mentioned in those discussions - I just did what I thought might
work.  In particular, the separation to separate colour planes and the
stippled 2-colour reduction of each is probably something anyone tackling
the problem would eventually decide to do.

Likewise, the multiplexing/interlacing of the RGB lines isn't a particularly
tricky technique - just a few lines of code - and something just about
anyone would eventually try.  I don't claim to have invented the technique -
but nor do I admit to copying it.  Probably among the many bits of
information I picked up during my thinking about the problem lead to my
solution/technique.  It's rather pointless saying who invented it, or who
deserves credit.  Look, for example, at the recent "interlacing" triumph.
Many of us had a significant hand in the development of that technique.

Only one person ACTUALLY produced the first demo showing the technique.

> Another thing you might want to emulate is GED on the Atari8.  GED

And with my efforts to date, my interest in this technique is now completed.
I have *absolutely no* interest in the sort of GED techniques - and I'll
happily leave THAT challenge to Thomas :)


----- Original Message -----
From: "Glenn Saunders" <mos6507@xxxxxxxxxxx>
To: <stella@xxxxxxxxxxx>
Sent: Wednesday, February 12, 2003 6:17 PM
Subject: [stella] Optimized color reduction approaches

> At 05:06 PM 2/12/2003 +1100, you wrote:
> > > Looks great.  I really like to weird look of the very wide pixels.
> > > interlacing looks nice and really reduces the flicker.  It just kinda
> >shimmers.
> >
> >It's not actually interlaced - at least, it's not an interlaced TV frame.
> I knew something didn't look quite right about it ;)
> >It's kind of double-interlaced in that each individual scanline
> >alternatively displays RGB, GBR, BRG... and the lines are out-of-synch
> >each other, so consecutive lines display RGBRGBRGB etc....  That is,
> >one line is displaying the red values for that line, the next line is
> >displaying it's green values, and the one after that displaying blue
> I think this is how the COLRVIEW technique as used with GR8 on the
> Atari8-bit.  Is that what you are trying to emulate?  It just like a 1-bit
> version of COLRVIEW.
> Another thing you might want to emulate is GED on the Atari8.  GED
> blocking out zones of color on the horizontal and doing color register
> changes in mid screen almost like HAM on the Amiga, but obviously you can
> only change registers a few times.  On the Atari you might only be able to
> change it once or twice per line.  But this tends to work well for cartoon
> style artwork where you might have one character on one end that is mostly
> blue, let's say, and the other character on the right being mostly
> red.  They could both use the same playfield color register if you changed
> it in mid screen over black.
> The other thing you should do is come up with a routine that will optimize
> the RGB triad.  Like let's say the top part of the screen was generally
> darker than the rest.  Instead of relying on dithering alone (over black
> which is rather flickery to begin with) you could reduce the luminance of
> the red green and blue frames.  And if the top of the screen is, let's
> a sky, you could optimize the RGB triad to be 3 shades of blue instead of
> G and B.  And as the sky gets generally lighter or darker, you can adjust
> the 3 shades of blue to approximate this and try to minimize the
> This approach wouldn't work well unless the image naturally had zones
> each zone strip had a somewhat dominant, monochromatic color scheme.  I'd
> really like to see a landscape like a beach scene or something using this
> method, or a photo of a real sunset rather than the usual Activision one
> --------------------------------------------------------------------------
> Archives (includes files) at
> Unsub & more at

Archives (includes files) at
Unsub & more at

Current Thread