Re: Let's pay a developer to make a VGA output card

["Michael J. Mahon" <mjmahon@aol.com>]

Mark McDougall wrote:
> Michael J. Mahon wrote:
>
> > I understand your preference, but I think you would find it much
> > simpler in reality to build an "Apple scan doubler" that grabbed the
> > dot clock from the Apple (AUX slot?), sampled the composite video
> > signal (it's digital, after all), and "played it back" at 2x the scan
> > frequency for a VGA monitor, using an appropriate RGB conversion.
> > This would produce an "emulator style" VGA display, which, I expect,
> > would satisfy many users--though perhaps none perfectly.  ;-)
>
>
> Ummm, I admit I've not much experience with video processing but I'm not 
> sure it's so "simple" to build a scan doubler.
>
> The chrominance information in a composite signal is encoded via a phase 
> offset from the colour burst. This is not trivial to recover, and 
> certainly can't be done by sampling at the dot clock rate. It is 
> possible to generate a limited number of colours using 'digital' logic 
> but that requires a clock 4 times (for example) the burst frequency.

That is the 14MHz "dot clock" I was referring to.  And since each video
output "state" lasts an integral number of 14MHz periods, sampling at
14MHz *exactly* captures all the information in the Apple II video.

All of the colors of which the Apple II is capable are encoded in binary
levels shifted out at a 14MHz rate.

> In any case, if you're sampling at the dot clock rate, then you can't 
> reproduce any analogue artifacts such as blurring, enlarging due to 
> contrast, etc since you only have a single value sampled at the 'centre' 
> of the pixel.
>
> At least that's how I understand it.

It's pretty straightforward if you take the sampled bit stream and shift
the bits (all states are either 0 or 1) through a shift register.  The
*minimum* length of this shift register is 4 bits, but I suggest using
at least 8 bits to permit "context dependent" or "blurred" color info.

The shift register value can directly index RGB color values to be
output.  I'd guess that 4-5 bits per color would be more than enough
resolution.

This may be a conceptual approach with practical problems, requiring
a video palette RAM that can be cycled at 14MHz to drive the RGB DACs.

I suspect that the existing RGB cards just use a 4-bit window and
a simple logic/resistor matrix to generate the color signals
from the value in the "window" shift register.  (And the //e RGB
card evidently doesn't even shift--it just loads the 4-bit values,
producing an artificially "crisp", non-artifacted display).

> Regardless of all this, my decision to do it this way was heavily 
> influenced by the fact that I have a few other features (video and 
> non-video related) that I'd like to incorporate on an expansion card.
>
> So I'd be mucking about with an FPGA in an Apple slot...

Ah.  I was thinking more of an AUX slot RAM/RGB card with VGA output.
Just trying to keep it simple...

-michael

Parallel computing for 8-bit Apple II's!
Home page:  http://members.aol.com/MJMahon/

"The wastebasket is our most important design
tool--and it is seriously underused."