subliminal color images
Kragen Javier Sitaker
kragen at pobox.com
Mon Oct 16 14:40:55 EDT 2006
On Mon, 16 Oct 2006 12:17:50 +0200, Dave Long wrote:
> > It would be cool to print out a black-and-white image on, say, a laser
> > printer, which contained an unobtrusive embedded "watermark" or
> > "barcode" that contained chroma information for the image --- rather
> > like what the Apple II did to get NTSC color just by producing a
> > pattern of 1's and 0's.
>
> Come to think of it, the Apple II scheme turns a well known broadcast
> bug (there are dress codes for on-air appearances, as houndstooth and
> similar high-frequency fabrics alias into color) into a feature. I
> wonder how much the use of gradients in web graphics is due to
> recent-featurism, and how much is in imitation of broadcast graphics,
> which have to have smooth gradients -- they'd bleed if one tried to
> make a crisp transition.
That's an interesting question. My preferred hypothesis is that it's
the age of calm technology --- one of the disadvantages of the Apple
][ scheme was that it could only produce bright, garish colors with
sharp boundaries between them, and gradients are much more soothing.
> > Then you could point, say, a cellphone camera at the image, and push a
> > button, and see the image in color.
>
> Unfortunately (fortunately?) the information between a cellphone camera
> and a printout isn't mediated by NTSC, but by regular photons, so it
> seems unlikely this would work.
I didn't mean it would work by accident --- I meant that software in
the cellphone could decode a chroma signal from a barcode "hidden" in
the image, for example in the angle and spacing of patterns of
parallel lines used to approximate a grayscale as in an engraving.
Venezuelan paper money has areas that appear from a distance to be one
solid color, but consist of many areas of fine parallel lines at
different angles. Presumably the idea is that each area would appear
a different brightness by raster aliasing if scanned with a
poor-quality scanner, or printed with a poor-quality printer.
But the same scheme could be applied to encode more interesting
information subliminally into the print; perhaps the frequency of the
lines could represent saturation, and their angle could represent hue.
More information about the Kragen-discuss
mailing list