Re: Apple //e's RGB Card To Apple IIgs' RGB Monitor

["Bryan Parkoff" <nospam@nospam.com>]

"Michael J. Mahon" <mjmahon@aol.com> wrote in message 
ct-dnStXWsxnpBXYnZ2dnUVZ_vC3nZ2d@comcast.com">news:ct-dnStXWsxnpBXYnZ2dnUVZ_vC3nZ2d@comcast.com...
> Bryan Parkoff wrote:
>> "mdj" <mdj.mdj@gmail.com> wrote in message 
>> 1166499011.152258.3620@73g2000cwn.googlegroups.com">news:1166499011.152258.3620@73g2000cwn.googlegroups.com...
>>
>>>Michael J. Mahon wrote:
>>>
>>>
>>>>I'd suggest using three 4-resistor networks to do the three 4-bit D-to-A
>>>>conversions--four resistors beats a D/A chip for just 4 bits.
>>>
>>>Agreed. It might also be possible nowadays to use a uC to replace the
>>>GAL idea. GAL/PAL seems to be going the way of the dodo, and using one
>>>small cheap chip instead of two seems a good move.
>>>
>>>
>>>>And buffer amplifiers may be needed for the RGB signals as well, if the
>>>>RGB monitor's inputs are 75 Ohm.
>>>
>>>Without this, you'd see some pretty strange colour artifacting.
>>>
>>>Also Bryan - your table seems to be quite wrong. Care to post a
>>>corrected one?
>>>
>>>Have you done a comparison of the IIgs's colour palette to the actual
>>>composite colours? It's possible that 12 bit RGB only provides an
>>>approximate of some of the colours, although admitedly it's probably
>>>close enough.
>>
>>
>> Matt,
>>
>>     Sorry, I made mistake.  It should be the correct table to be shown 
>> strange colors below.
>>
>>          IRGB  RED   GREEN BLUE  Hex
>> Color 0  0000  0000  0000  0000  000
>> Color 1  0001  0000  0000  1010  00A
>> Color 2  0010  0000  1010  0000  0A0
>> Color 3  0011  0000  1010  1010  0AA
>> Color 4  0100  1010  0000  0000  A00
>> Color 5  0101  1010  0000  1010  A0A
>> Color 6  0110  1010  0101  0000  A50
>> Color 7  0111  1010  1010  1010  AAA
>> Color 8  1000  0101  0101  0101  555
>> Color 9  1001  0101  0101  1111  55F
>> Color A  1010  0101  1111  0101  5F5
>> Color B  1011  0101  1111  1111  5FF
>> Color C  1100  1111  0101  0101  F55
>> Color D  1101  1111  0101  1111  F5F
>> Color E  1110  1111  1111  0101  FF5
>> Color F  1111  1111  1111  1111  FFF
>>
>>     This table above is 16 CGA colors.  Thanks to Michael's correction 
>> when I made mistake to label RGBI as bit reverse.  I believe that RGB, 
>> GREEN, and BLUE data are to be correct because 24 bit data and 12 bit 
>> data are 99% identical when you set "1111" on low nibble or duplicate 
>> from high nibble to low nibble.  Filling "1111" on low nibble gives 
>> little brightness.  You can use like (12 bit data) A50 as 24 bit data) 
>> AA5500 or (24 bit data) AF5F00.
>
> Just for the record, the exact conversion from 12-bit to 24-bit RGB data
> (or any other kind of 8-8-8 data) is: 12-bit XYZ --> 24-bit X0Y0Z0.
>
> Anything besides low-order zero fill is creating data that isn't in the
> 12-bit representation.  Duplicating the hex digits (e.g.: XXYYZZ) is a
> way of expanding the 24-bit values so that they still run from zero to
> full-scale, so it produces the correct relative colors.

Michael,

    I agree.  It can be X0Y0Z0, XXYYZZ, or XFYFZF according to your 
preference when low nibble gives less, normal, or more brightness.  Think of 
8 colors (3 bits).  RGB has only one bit.  How do you translate one bit to 
24 bits?  You put one bit into 8th bit as MSB by filling "1111111" on 7th 
bit through 1st bit as LSB.  Think of 64 colors (6 bits).  RGB has only 2 
bits.  You put two bits into 8th bit and 7th bit as MSB by filling "111111" 
on 6th bit through 1st bit as LSB.  It does the same for 512 colors (9 bits) 
with 3 bits and 4,096 colors (12 bits) with 4 bits.
    My point is to fill zero or one on LSB according to your best judgment 
when you agree what you think color to look good or bad.

    Please try to answer my one question.  Do you think it is worth to use 
GAL16V8 chip or ROM chip so Apple //e's RGBI port can be translated to Apple 
IIgs' 16 colors using 12 bit look-up table in the GAL or ROM chip.  Then D/A 
chip or resistor network will give correct 16 colors (not 16 CGA colors).

Bryan Parkoff