10

The Amiga 1000, 2000, 500, 1200, and 600 all had composite video output ports in addition to their preferred RGB connectors. While I know that the A1000 will produce color output over the composite port, I was always disappointed and a bit curious as to why the later A2000 only produced monochrome / grayscale output from its composite port. Not that I would expect great quality output, but the lack of color is curious.

As I only really have experience with the 1000/2000/500 models, I can't speak to the others' color capabilities. However, I suspect the mono output was possibly "fixed" in the 600/1200.

So the question is, what was the reason for (at least) the A2000's composite video output being grayscale and not full color?

3
  • 2
    The A500 has mono too. A600 and A1200 has color. Technically composite is always color, the A2000 and A500 are just plain old monochrome video.
    – pipe
    Commented May 17 at 14:50
  • It was 1987, so I wonder whether it's as simple as "Macs are doing the monochrome schtick so let's follow suit."
    – Theodore
    Commented May 17 at 15:09
  • Yes, the A1200 had colour "aerial" output.
    – nsandersen
    Commented May 17 at 20:37

4 Answers 4

19

I don’t know that we’re going to do much better than speculation here, but assuming the same reason as the Acorn machines amongst others: if you offer monochrome by composite out then almost any monochrome monitor of the era can be used with full pixel fidelity.

If you offer colour composite out then the image gets less sharp, as the colour channels occupy the high-frequency part of the image and are filtered out either by the display or perceptively by you.

In NTSC land there are only 227.5 colour cycles across the entire line (i.e. even less in the visible part), so any image with more detail than that is going to be downgraded if output in colour.

So the decision was likely: that it’s one of the professional models of Amiga, so people will most-likely use a monitor, which means they’ll either have a colour RGB connection or we can give them a monochrome composite.

5
  • Chrominance carriers shouldn't interfere with the clarity of luminance-only display. Certainly my Spectrums worked very well with B&W TV - equally as sharp as a ZX81. I'm from a PAL region, but surely other colour systems should have the same property, otherwise they would be totally unsuitable for broadcast to mixed monochrome and colour audiences? Commented May 21 at 12:47
  • @TobySpeight if you had a 48kb then you should have seen at least dot crawl, with a 128kb just a softness that won’t have subtracted much at 256px across. But, yeah, NTSC is a bit worse here; its colour subcarrier is at 227.5 cycles per line whereas PAL moves it up to around 283.75. In both cases the intention on old black and white sets is that the high-frequency noise perceptively cancel itself out, but the high-frequency part of the luminance channel is still missing.
    – Tommy
    Commented May 21 at 13:37
  • @Tommy: If an NTSC colorburst signal is present, luminance patterns that repeat at a 3.579545Mhz rate will appear as saturated colors; if horizontal sweep is chroma/227.5, a checkerboard pattern would cause the same color to appear on every line of a frame (an opposing checkerboard would show a different color). Depending upon whether the Amiga outputs 262 or 263 lines in non-interlaced mode, the colors may stay constant or may flip every frame, and depending upon the circuit design software may or may not be able to predict what the colors are (on an NES, it's possible to...
    – supercat
    Commented May 21 at 15:26
  • ...use chroma artifacts to synthesize colors (e.g. produce a 27-color display using only three colors plus black), but the hardware chroma reference powers up randomly in one of twelve phases, so any game which wanted to exploit chroma artifacts would need to ask the user to indicate it, e.g. by showing twelve squares and asking the user to select whichever square appears darkest.
    – supercat
    Commented May 21 at 15:30
  • 1
    @supercat although true, we’re now in very digressive territory. But if we’re here anyway, noting that the emulator I write has been getting these details correct thanks to doing a genuine NTSC or PAL decode for almost a decade now. Tying all the comments here together, it displays Chromatrons Attack correctly, the only known ZX Spectrum title to deploy artefact colours deliberately.
    – Tommy
    Commented May 21 at 16:11
14

As the Amiga was sort of high performance serious computing device, it was assumed that it is being used with an analog RGB color monitor or in high-res monochrome on standard video monitors and TVs, instead of inferior quality color composite signal on video monitors or TVs.

Since the A2000 already has three 4-bit DACs for converting the RGB to analog, it would have been more expensive to design in the colour composite video encoding circuitry and add the parts cost and circuit board area, while many would never use the colour composite output.

The monochrome output is easy, just add a weighted sum of the analog RGB outputs and the composite sync together as the luma output.

If you wanted to have a color composite output, it was possible with the addtion of an external adapter. Just like it was done with A500 which was often bundled with A520 adapter, to allow color composite output and RF modulator.

The point might be that the monochrome composite output is very sharp as it does not have colour information, i.e. no colourburst means no display tries to decode the colors, no colour carrier means it does not have colour pixel artifacts and can utilize higher bandwidth. So effecively composite capable monitors switch off their colour decoding and allow for higher fidelity monochrome video.

9
  • Uh? A520 has nothing to do with video. It is a SCSI controller + RAM expansion. I think you have the type-number wrong.
    – Tonny
    Commented May 19 at 10:20
  • 1
    @Tonny I took the model from here. Maybe it's wrong then. amiga.resource.cx/exp/a520
    – Justme
    Commented May 19 at 10:24
  • No, you're right all along. My mistake. I had it confused with the A590 expander. Somehow I had A520 locked in my mind, but I just checked my own rig in the basement and the SCSI/RAM expander is indeed the A590.
    – Tonny
    Commented May 19 at 12:04
  • 1
    FWIW, the A2000 composite output was quite crisp. It used slightly odd conversion weights: L = .3*R + .6*G + .1*B. That gives you 151 grey levels, a big improvement on the usual 16.
    – PM 2Ring
    Commented May 19 at 12:49
  • 1
    It would seem less odd had they rounded to something quick to carry out (and/or compact to represent) in binary even at the expense of being further from standard values. It seems almost like they were working from a heavily-rounded source (@PM2Ring)
    – Chris H
    Commented May 20 at 9:09
2

In simplest terms the Amiga 2000 only outputs greyscale on its composite output because it's nearly architecturally identical to the the Amiga 500 (Rev 5) it's based on¹ which also only outputs greyscale. Why the Amiga 500 only outputs monochrome over composite is probably a matter of speculation though.


¹ The original Amiga 2000 "A" model (Rev 4) is based on the Amiga 1000 but unlike it (or the later cost reduced 2000 "B" model) it doesn't feature any composite output at all.

1

If a machine outputs pixels at a dot rate precisely 2x chroma (as is done in the 320-pixel modes of the Atari 8-bit computers, IBM CGA, or Amiga, as well as the 280-pixel modes of the Apple II), patterns of alternating pixels will show up as colors which aren't present in the original. When using 4x chroma modes, repeating 4-pixel sequences will be tinted similarly. This can be a useful trick for showing more colors than would otherwise be possible, but only if programmers are planning for it.

This point may be well illustrated by considering how games are designed on the IBM PC Color Graphics Adapter. Many games allow the user to select between composite and RGB output. In order of quality:

  1. Composite option, shown on composite monitor; image shows up with nice colors.
  2. RGB option, shown on RGB monitor; colors are generally limited to dithered combinations of red, green, and yellow on a blue background, but make sense within those limitations.
  3. Composite option, shown on RGB display; image shows up in black and white
  4. RGB option, shown on composite monitor: weird and garish colors that make no sense.

I haven't actually seen an Amiga used with composite output, but I expect the typical experience was like #4 above. Eliminating the chroma signal changed it to #3.

15
  • I don't think the Amiga is in-phase with the colour subcarrier, but can find only weak supporting evidence for that belief; e.g. Page 23 of the Hardware reference — "All lines are not the same length in NTSC. Every other line is a long line (228 color clocks ...), with the others being 227 color clocks long".
    – Tommy
    Commented May 17 at 16:13
  • 1
    @Tommy: Given that description, I'd guess the time between horizontal sync pulses is 227.5 chroma clocks long, consistent with broadcast video. That would result alternate scan lines having opposite color phase, but software which uses checkerboard dithering would end up putting all of one color on one color phase and all of the other on the opposite phase. Depending upon whether the Amiga outputs 262 or 263 lines/frame in non-interlaced mode, each scan line may have either a fixed color phase, or a phase that alternates every frame, but either way the color artifacts...
    – supercat
    Commented May 17 at 16:21
  • 1
    ...would be very bad when showing content that isn't designed their existence.
    – supercat
    Commented May 17 at 16:24
  • @Tommy Please read the next two sentences after the quoted section. It explains that each line is 227.5 color clocks long. In other words, exactly 455 dot clocks, or exactly 910 oscillator clocks, which matches the NTSC standard. The lines are only 227 or 228 clocks long to the copper because it does not work with half clocks.
    – Justme
    Commented May 18 at 11:01
  • 1
    @Justme that’s useful, disjoint information. The question is: why is the composite output monochrome. Given that the output is purely monochrome your statement that “if a machine outputs pixels at a clock rate … patterns of alternating pixels will show up as colours” is either false or, more charitably, irrelevant. What you’ve posted is unrelated to the question asked, just as your previous reply is to my previous reply.
    – Tommy
    Commented May 18 at 15:08

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .