[Coco] screencast presentation re: NTSC 8-bit Color on the Color Computer 3

[John W. Linville]

On Tue, Apr 29, 2014 at 10:03:41PM -0500, Joel Ewy wrote:
> On 04/29/2014 06:23 PM, John W. Linville wrote:
> >I made a screencast version of the presentation I gave at the 23rd
> >Annual "Last" Chicago CoCoFEST! about the CoCo3's NTSC-only 8-bit
> >color mode.  If you didn't get to see it (or want to see it again),
> >then now is your chance -- enjoy!
> >
> >	https://www.youtube.com/watch?v=gc6Eu7IansE
> >
> >John
> Thanks for posting this, John.  This was helpful in understanding
> artifact colors a little better.  In addition to emulator support
> (which would really encourage the wider use of 8-bit artifacts),
> being able to calculate artifact colors programmatically would be
> useful for, say, GIF display programs on the CoCo 3, though a
> look-up table would probably be faster.  But if 16 palette registers
> and the 'A' bits yield different palettes than just the 'S' bits
> alone, then you could select the most representative palette for a
> given image algorithmically, assuming all that can eventually be
> worked out.

There definitely are other palette variations available with the
non-zero 'A' bits.  Trying to calculate the best palette while running
on the CoCo3 itself sounds a bit painful to me, but if it sounds cool
to you then I hope to see you do it! :-)

> One thing I'm still not clear on is this business of 160 color
> transitions per line.  If that is inherent to the NTSC signal, then
> the effective horizontal resolution of all NTSC video is only the
> analog equivalent of 160 "pixels" (plus overscan?).  Somehow my
> video cassettes look better than that, even as low-def as they are.
> Is it just that the Luma can vary quicker and is more visually
> important?  Or am I misunderstanding something.

It is a bit complicated, but as Gene confirmed there are only those
160 color transitions in a line.  Black and white output could be
higher resolution than that.  Also, keep in mind that a 'real' video
signal might contain essentially an infinite number of color levels.
Higher 'color resolution' can have similar improvement effects on
image perception as higher 'spatial resolution' has.

> Regarding the relative importance of luma over chroma, I have
> considered combining 4-color 640 pixel wide grayscale images derived
> from the Y component with 160 pixel color images mapped to one of
> the artifact palettes in order to get more visual detail in these
> 8-bit artifact color images.  Undoubtedly because of the artifacting
> there would be some additional color error, but I think it's worth
> trying.

I'm not entirely clear on the details of what your are proposing.
But keep in mind that the Y signal is one of the three components
of a color signal.  So any Y variation will be reflected in the
output color.  Given that, I would think that any attempt to further
manipulate Y would be hindered by the resulting disturbance in the
displayed colors.

John
-- 
John W. Linville		Someday the world will need a hero, and you
linville at tuxdriver.com			might be all we have.  Be ready.