RE: Definitive apple ][ lo res (and hi res) mode colors (RGB values)

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

Hello,

    Back to the date: September 20, 2005.  My comments is below.

>Inspired by this thread I did the math to find out what the NTSC Apple
>II/II+/IIe/IIc colors actually are. (The IIgs produces noticeably
>different colors because of its RGB operation). Unfortunately, most
>emulators don't get the colors right. Maybe somebody feels like patching
>the correct color values as given here into some emulator(s)?
>
>The PAL Apples produce almost identical colors to their NTSC cousins,
>but unfortunately unlike for NTSC there's no pretty mathematical way to
>calculate their theoretical values, since the PAL Apples use some
>not-so-well-defined analog circuitry to add together luminance and
>chrominance signals, while the NTSC Apples produce both luma and chroma
>in a single digital process.
>
>I'm assuming that the A2 logic circuits produce a perfectly rectangular
>output on the video out, alternating between 0 (black) and 1 (white),
>not counting the sync (which is added afterwards). This is pretty close
>to what they actually do. The only possible error is a very slight one
>in color saturation, hue and brightness should not be affected by this
>assumption.
>
>I'm also assuming that the NTSC decoder at the receiving end is
>completely to (modern) spec. The original "NTSC 1953" norm called for
>YIQ decoding and had different color primaries than modern sets, but
>almost all TV sets actually produced use YUV decoding and the modern
>primaries, which were then made the standard in "SMPTE 1979", the modern
>NTSC norm. I follow the 1979 norm since the old norm is mostly of
>theoretical value only, and when the Apple was released in 1977 the
>modern norm was de facto already in place.
>
>Then it is easy to mathematically derive the correct color values. All
>it takes is a fourier transform of the waveform. The zeroeth coefficient
>  (which is 0.0 for black, 0.25 for the 4 "dark" colors, 0.5 for grey
>and the four "mid" colors (which are also the hires colors), 0.75 for
>the four "bright" colors, and 1.0 for white) of said transform gives the
>luminance, the first coefficient (0 for black, grey, white, 1/(2*pi) for
>the "mid" colors, and 1/(2*sqrt(2)*pi) for the "dark" and "bright"
>colors) gives the color saturation. Phase angels (hues) are multiples of
>90� for the 4 "dark" and the 4 "bright" colors, and multiples of 90�
>plus 45� for the 4 "mid" colors.
>
>The lores colors 4, 7, 8 and 11 are affected by clipping, i.e. when
>these colors are used the Apple is telling the TV to produce a color it
>can't actually produce since it's somewhat outside the gamut - none of
>the three electron guns in the picture tube can be turned down below
>zero power, or above maximum power. This clipping is taken into
>consideration in the tables.
>
>The resulting RGB values (in hex) in the SMPTE color space for the 16
>lores colors are as follows:
>
>NTSC Apple II Colors RGB Values in SMPTE 1979 color space
>--------------------------------------------------------- 
>R, G, B [ 0] = 0x000000 * Hires 0 & 3
>R, G, B [ 1] = 0x901740
>R, G, B [ 2] = 0x402ca5
>R, G, B [ 3] = 0xd043e5 * Hires 2
>R, G, B [ 4] = 0x006940
>R, G, B [ 5] = 0x808080
>R, G, B [ 6] = 0x2f95e5 * Hires 6
>R, G, B [ 7] = 0xbfabff
>R, G, B [ 8] = 0x405400
>R, G, B [ 9] = 0xd06a1a * Hires 5
>R, G, B [10] = 0x808080
>R, G, B [11] = 0xff96bf
>R, G, B [12] = 0x2fbc1a * Hires 1
>R, G, B [13] = 0xbfd35a
>R, G, B [14] = 0x6fe8bf
>R, G, B [15] = 0xffffff * Hires 4 & 7
>
>The numbers in square brackets are the lores color numbers (COLOR= in
>Applesoft). The hires colors (HCOLOR= in Applesoft) are marked with an
>asterisk.
>
>Note that these RGB values are valid only for the SMPTE 1979 color
>space. Converting them to sRGB color space (which is what modern PC
>monitors *should* use when set to 6500K color temperature, most are at
>least close; this is also the color standard for HDTV sets and for Web
>pages) and linearily scaling their brightness so that the maximum
>coefficient is 0xff, gives these slightly different values:
>
>NTSC Apple II Colors RGB Values in sRGB color space
>--------------------------------------------------- 
>R, G, B [ 0] = 0x000000 * Hires 0, 3
>R, G, B [ 1] = 0x8a2140
>R, G, B [ 2] = 0x3c22a5
>R, G, B [ 3] = 0xc847e4 * Hires 2
>R, G, B [ 4] = 0x07653e
>R, G, B [ 5] = 0x7b7e80
>R, G, B [ 6] = 0x308fe3 * Hires 6
>R, G, B [ 7] = 0xb9a9fd
>R, G, B [ 8] = 0x3b5107
>R, G, B [ 9] = 0xc77028 * Hires 5
>R, G, B [10] = 0x7b7e80
>R, G, B [11] = 0xf39ac2
>R, G, B [12] = 0x2fb81f * Hires 1
>R, G, B [13] = 0xb9d060
>R, G, B [14] = 0x6ee1c0
>R, G, B [15] = 0xf5faff * Hires 4 & 7
>
>Hope this helps,
>-- 
>Linards Ticmanis

    I am curious.  How did you get NTSC colors to be translated into RGB 
values?  Do you have formula?  I like Apple IIgs' 16 colors better than NTSC 
colors according to my preference, but some people prefer NTSC colors. 
Emulator authors allows people to choose primary Apple IIgs' 16 colors or 
secondary NTSC colors.

Bryan Parkoff