Is it true the Amiga nearly DIDN'T use RGB for colour?

[Foebane]

I've got a video of Jay Miner giving a speech about Amiga specs and development, and he says at one point that they only decided on using RGB for the Amiga palette at the last minute, as well as bitplanes as we know them.

For 16-bit computers and onwards, RGB is the most comprehensive, versatile, "natural" and compatible colour space for colour palettes for use on RGB CRT displays (which is the standard) and I can't imagine what other colour space could be used in place of it (AFAIK, Atari 8-Bits used HSV, which was more limited for certain types of colours).

[Hewitson]

I wouldn't be surprised. Remember that the Amiga was created in the US, where RGB compatible TV's don't exist.

[Daedalus]

I'm not sure about the main question, but from memories of my old 800XL, the Atari 8-bits were a little more primitive than proper HSV/HSL, which can still produce a wide palette. Instead they used a partial version, where you had 16 colours to choose from (H) and 16 levels of brightness (L) for a total of 256 possible colours. Special tricks were needed to show them all on screen though, similar to copper tricks on the Amiga. Typically you'd only have 16 of them on screen at any one time.

[robinsonb5]

I'm sure I remember reading somewhere that HAM mode was originally intended to work in some kind of YUV-like mode instead of RGB, and it was nearly removed when the decision was made to go RGB. (Of course I can't find any such reference now!)

[Gorf]

Composite YUV was probably the most used way for a A1000.
But the signal was generated by the Vidiot from the RGB signals (a more sophisticated colour-version of the later A500 b/w signal)

So it would have made sense to stay in the YUV realm during the whole signal chain.
So instead of 12bit RGB values you could simply use 12bit YUV in the registers - bitplanes and blitter-operations would stay the same.

A "reverse-vidiot" could than generate a RGB signal from the YUV values....

[robinsonb5]

Originally Posted by Gorf

So it would have made sense to stay in the YUV realm during the whole signal chain.
So instead of 12bit RGB values you could simply use 12bit YUV in the registers - bitplanes and blitter-operations would stay the same.

Interesting - it does make sense, so much so that I'm now wondering if any computers were actually designed and built that dealt with colour in this way? (In the 80s / very early 90s, at least -I'm not counting YUV video overlays on modern PC graphics cards)

[Toni Wilen]

Yes, HAM was designed because original design used YUV color space where HAM makes much more sense and it was going to be removed after design was changed to RGB.

I think it was mentioned in some old interviews, also in Ron Nicholson and Joe Decuir at Amiga 30th anniversary presentation few years ago and Amiga Years book.

I have never heard of bitplanes being late change. It probably is a misunderstanding because <some other gfx design> to bitplanes would have been very big change. YUV to RGB was easy, no Agnus or Denise changes needed, only needed to rename signal lines coming Denise

[Gorf]

Originally Posted by robinsonb5

Interesting - it does make sense, so much so that I'm now wondering if any computers were actually designed and built that dealt with colour in this way? (In the 80s / very early 90s, at least -I'm not counting YUV video overlays on modern PC graphics cards)

The Atari Jaguar did. The colour-format its labeled CRY there.
It allows easier shading and lighting since only the Y component needs to change...

[Bruce Abbott]

Originally Posted by robinsonb5

I'm now wondering if any computers were actually designed and built that dealt with colour in this way? (In the 80s / very early 90s, at least -I'm not counting YUV video overlays on modern PC graphics cards)

Many 80's home computers produced YUV internally, because that was easier to convert to NTSC/PAL. Of course due to limited graphics memory and bandwidth they mapped a small number of colors into the YUV space. Then the TV had to convert the YUV to RGB to drive the tube!

The MC6847 (used in the Acorn Atom, Tandy Color Computer. Lazer 200/VZ200...) and TI9928/9 (Ti-99/4a, Sega SC3000, Spectravideo, MSX...) were two popular graphics chips that had YUV outputs. Also the original ZX Spectrum ULA produced a form of YUV.

For those of us who wanted sharp displays this was a pain because the analog circuitry required to convert YUV to RGB was quite complex and tricky to set up. RGB was much preferred due to its simplicity and ease of obtaining a sharp picture, but you needed an RGB monitor to get the best out of it (a standard TV could be modified to take RGB in, but the video amplifiers had low bandwaidth and the tube had a coarse shadow mask so the result was usually unsatisfactory).

Having sharp RGB outputs and a matching monitor made the Amiga much more desirable than other home computers which only had crappy composite/rf outputs! That it nearly didn't get it is another example of how lucky we are.

[Paul_s]

Originally Posted by Foebane

I've got a video of Jay Miner giving a speech about Amiga specs and development, and he says at one point that they only decided on using RGB for the Amiga palette at the last minute, as well as bitplanes as we know them.

For 16-bit computers and onwards, RGB is the most comprehensive, versatile, "natural" and compatible colour space for colour palettes for use on RGB CRT displays (which is the standard) and I can't imagine what other colour space could be used in place of it (AFAIK, Atari 8-Bits used HSV, which was more limited for certain types of colours).

Same story with the side bus expansion. I can't remember where I heard him talk about that - possibly on the interviews with Jay on the Cloanto Amiga Forever pack.

[Solderbro]

Amiga RGB is that good, that converters and most RGB equipped flat TV's do work with the hires workbench modes. Even on a cheap Odys TV the highres interlaced workbench is pretty usable, the unthinkable from CRT become true.

[Foebane]

I should correct something I said about HSV used on the Atari 8-Bits: HSV is Hue, Saturation and Value, but the Atari designers didn't use Saturation (the strength of a hue) because of the 8-bit limitation, so all the hues on the Atari are full strength.

[Bruce Abbott]

Originally Posted by Solderbro

Amiga RGB is that good, that converters and most RGB equipped flat TV's do work with the hires workbench modes. Even on a cheap Odys TV the highres interlaced workbench is pretty usable, the unthinkable from CRT become true.

Yes, it's so good that hires is often usable even when converted to Composite. Here's a photo of my Workbench in 640x256 8 colors, connected via the A1200's composite output to my 32" LCD TV. This TV also produces a rock solid flicker-free display in PAL Hires Interlace, which makes IBrowse quite usable in Composite.

Modern TVs have advanced digital video processing which can make a good composite signal look even better. This wasn't so when the Amiga was originally designed. Back then the video processing circuits in TVs were all analog, and most did a pretty poor job (particularly in NTSC, due to its lower bandwidth and lack of alternate line phase correction). Having both composite and RGB put the Amiga ahead of most home computers at the time, and ahead of PCs which only had low color depth RGB outputs.

The Amiga also worked with a genlock to do video titling and picture in picture effects. That was a big deal back then because digital video processing was extremely expensive (the Quantel Paintbox cost $150,000) and standard PCs couldn't do it. Was genlocking another feature that almost didn't make into the Amiga? People who focus on things the Amiga didn't get should be more appreciative of the things it did get.

[amiman99]

I'm actually reading this book "Commodore The Amiga Years" by Brian Bagnall.
On page 210, he wrote that when the plans changed from building game machine to business computer they had to redesign the Daphne chip which use the HSL model for color production to RGB, which made it slower, and changed the name to Denise.
Interesting book.

[pandy71]

Originally Posted by robinsonb5

Interesting - it does make sense, so much so that I'm now wondering if any computers were actually designed and built that dealt with colour in this way? (In the 80s / very early 90s, at least -I'm not counting YUV video overlays on modern PC graphics cards)

As mentioned YUV generation was easier than RGB and supported by many computers - YUV was easy to generate as it can be created purely in digital domain in simple circuits (as colour information is coded in phase of signal and phase can be easily controlled by flipflops). In Amiga times analog RGB was unique - IBM PC used TTL multi-line connection - RGBI - first analog RGB in PC world (large scale) was VGA (MCGA) - from this moment analog RGB begin to be widely implemented.
And Amiga Vidiot was always component and never color composite - Amiga 1000 used discrete video DAC and MC1377 as color encoder - Vidiot is hybrid implementation for Amiga discrete DAC.
It is quite interesting to simulate if HAM can be profitable by using YCbCr (or YCoCg) - it should be not a problem to add YCoCg (YCbCr) to "new Denise" if outcome of simulation shows HAM benefits.

Originally Posted by Foebane

I should correct something I said about HSV used on the Atari 8-Bits: HSV is Hue, Saturation and Value, but the Atari designers didn't use Saturation (the strength of a hue) because of the 8-bit limitation, so all the hues on the Atari are full strength.

Because color can be easily controlled by series of flip flops and simple muxer to select phase delay (when referenced to burst), adding saturation will require DAC (color information is coded by amplitude and phase).