The above writing motivated me to write a bit more about my current project,
Retrotron, and its real-world counter-part.
Recently I said that I'm going to build a real counter-part of Retrotron using
a micro-controller and a mini CRT while generating all the signals in
software. This, basically, is the same as what the Pixie or any other CRT
controller did back then, yet I try to do it in software, meaning I'm going to
simulate a simple CRT controller in software which, given some RAM and a dot-
clock, should output a composite video signal to drive a real TV/CRT.
Basically, one should be able to take the controller and connect it with the
composite video input of ones own TV getting whatever cool graphics on the
screen the controller generates at that instance. But not only do I want to
generate the video signal myself, I also want to control the TV/CRT on its
own. In the end this should be my little video hacking computer fully
simulated in software with a real CRT displaying all the cool stuff.
One thing that got me hooked since ever is about generating a different scanning
pattern for the beam of an CRT. Instead of generating a linear scanning
raster, what about letting the beam scan the screen in a spiral from the
inside out? With the project described as is, it should become possible to do
it some day. This would be a new way of scanning a TV (perhaps 30 years late
xD) which becomes possible by generating all the complex signals in software/
digitally. I don't know if this will be of any huge advantages over linear
raster-scan, but I may imagine that the flickering will be less pronounced
(perceptual-wise) and that we can produce higher-resolution composite video
signals, because there will be no horizontal blanking interval resp. retrace,
only a retrace of the beam from the outside to the inside, i.e. half the
screen width. So the spiral-scan vertical retrace time compared to the
standard linear raster vertical retrace time reduces to just half the time of
the standard horizontal retrace time. Hence, this will be utterly fast. One
may also try to scan the screen using space-filling curves. Lots of
possibilities here. But there is more. Being able to scan arbitrarily would
allow to combine raster and vector displays, for, you can have vectors
displayed like on a vector display but you also can display raster (shaded-)
graphics at the same time because you can define a scanning pattern for such
(shaded) objects and raster along that pattern. Sounds crazy, does it? xD
However, that's for the future. I need to make the first step first. I finally
now got together some very small CRT tubes to work with.
I ordered a couple of controllers like an RPI, some Arduinos, etc. The one in
the picture is a Teensy++ 2.0. Don't know which controller to use, yet. I want
to keep it small, very small, should all run from a just a battery cell.
However. This real-world counter-part of Retrotron is a side project of
mine. Don't expect cool result too soon. I think I will need to burn quite
some time getting the controller to generate a proper (real) composite video
signal. Anyhow, I can't wait displaying some games or animations on it!