🚀 go-pugleaf

One way to do this is rather old technology - there are several on the
market that do just this:

Use a level detector and a couple of counters to count bits from the
horizontal and vertical sync pulses, then when you reach coincidence
(use a comparator) with a register for the current-cursor-position
(stored in a latch), you use a sample-and-hold and digitize that with a
nice slow A/D converter.  When the conversion finishes, you store that
value and update the current-cursor-position latch to the new screen
location you want to digitize.

This is slow for doing a whole screen at a high resolution - about 8
seconds or so for 256 x 256 with a 4MHz Z-80 - but it can give you real
nice control over what area you want to digitize and it can be really
fast if the area is small.  The grey-scale resolution is up to you -
just use whatever size A/D you want.  Faster A/D converters will give
you faster frame conversion.

Only real drawback is that the video image and digitizer clocks must be
rock stable, so real-time work is out unless you have an incredibly
fast processor and a flash A/D converter.

A couple of years ago we bought a product called the "Digisector" made
by The Micro Works in Del Mar, California that works pretty much this
way.  For about $300, we got a 256x256x6 digitizer that takes in RS170
monochrome video, which I think was a pretty good deal!  I'm told they
have a new version that can grab and digitize an NTSC frame in colour
on an IBM-PC card, but I've not yet tried one.

	Brian Kantor	UCSD Computer Graphics Lab
			c/o B-028, La Jolla, CA 92093