On Tue, 13 Jan 2004 15:46:41 -0000
Jim H wrote:

Well I consider them so. To me the 'colour' of an object is defined
only by the light coming from it.

That definition is at odds with multi-colour reproduction though. on a
computer screen a 'pink' is defined in terms of its red and blue (and to
a lesser extent green) content.

In 'real life' pink has a given wavelength and intensity.

Btw, can the human eye distinguish between yellow light and a similar
combination of red and green?

No. thats the whole crux of how RGB (and of course CMY) works.

You could describe RGB (or CMY or XYZ) colour reproduction as a
perceptual encoding, in a similar way to the way we think of mp3.

To describe a 'pink' we throw away the raw data and pretend its really a
combination of a far smaller set of wavelengths.

On a frequency/intensity plot the two situations would look very
different, but the assumption that they appear same seems fundemental
to rgb representation of colour.

Indeed. And thats before we really get onto the subject of surfaces that
emit / reflect a multi-wavelength pattern...

--
Spyros lair: http://www.mnementh.co.uk/ |||| Maintainer: arm26 linux

Do not meddle in the affairs of Dragons, for you are tasty and good with
ketchup.

On Tue, 13 Jan 2004 15:46:41 -0000
Jim H wrote:

Well I consider them so. To me the 'colour' of an object is defined
only by the light coming from it.

That definition is at odds with multi-colour reproduction though. on a
computer screen a 'pink' is defined in terms of its red and blue (and to
a lesser extent green) content.

In 'real life' pink has a given wavelength and intensity.

Btw, can the human eye distinguish between yellow light and a similar
combination of red and green?

No. thats the whole crux of how RGB (and of course CMY) works.

You could describe RGB (or CMY or XYZ) colour reproduction as a
perceptual encoding, in a similar way to the way we think of mp3.

To describe a 'pink' we throw away the raw data and pretend its really a
combination of a far smaller set of wavelengths.

On a frequency/intensity plot the two situations would look very
different, but the assumption that they appear same seems fundemental
to rgb representation of colour.

Indeed. And thats before we really get onto the subject of surfaces that
emit / reflect a multi-wavelength pattern...

--
Spyros lair: http://www.mnementh.co.uk/ |||| Maintainer: arm26 linux

Do not meddle in the affairs of Dragons, for you are tasty and good with
ketchup.

On Tue, 13 Jan 2004 13:53:33 +0000 (GMT), Jim Lesurf
wrote:

In article , Keith G
wrote:

It's not subtle with me - it's glaringly obvious. The difference between
SS and Valve amps is profound - I notice instantly and always.

I will bet you any amount of money you like, that I can put up a pair
of amps, one SS and one valved, that you can *not* distinguish
audibly, when used below clipping.

I have also noticed audible differences between valve amps and SS amps in
the past. Like youself, on occasion I have found these very obvious.
Although in other cases they have seemed small or unnoticable.

Quite so.

However to me this seems quite different to agreeing that SS amps 'kill'
the imaging, etc. This is certainly not my general experience.

No, it's just propaganda put about by the 'valvies', who refuse to
admit that these are just artifacts *added* by valve amps, and were
never in the input signal.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

On Tue, 13 Jan 2004 13:53:33 +0000 (GMT), Jim Lesurf
wrote:

In article , Keith G
wrote:

It's not subtle with me - it's glaringly obvious. The difference between
SS and Valve amps is profound - I notice instantly and always.

I will bet you any amount of money you like, that I can put up a pair
of amps, one SS and one valved, that you can *not* distinguish
audibly, when used below clipping.

I have also noticed audible differences between valve amps and SS amps in
the past. Like youself, on occasion I have found these very obvious.
Although in other cases they have seemed small or unnoticable.

Quite so.

However to me this seems quite different to agreeing that SS amps 'kill'
the imaging, etc. This is certainly not my general experience.

No, it's just propaganda put about by the 'valvies', who refuse to
admit that these are just artifacts *added* by valve amps, and were
never in the input signal.
--

Stewart Pinkerton | Music is Art - Audio is Engineering

Hi,

In message , Ian Molton
writes
On Tue, 13 Jan 2004 15:46:41 -0000
Jim H wrote:

Well I consider them so. To me the 'colour' of an object is defined
only by the light coming from it.

That definition is at odds with multi-colour reproduction though. on a
computer screen a 'pink' is defined in terms of its red and blue (and to
a lesser extent green) content.

In 'real life' pink has a given wavelength and intensity.

True for pink (desaturated red), but it gets complicated with purple.
The texts refer to 'lines of purple' as it doesn't have a single
'frequency' and doesn't appear on a classic colour wheel, but you can
sure as heck see it. The same perceptual problem accounts for the
'bands' that appear in rainbows.

I stand by what I said originally. It's a bag of worms :-)

--
Regards,
Glenn Booth

Hi,

In message , Ian Molton
writes
On Tue, 13 Jan 2004 15:46:41 -0000
Jim H wrote:

Well I consider them so. To me the 'colour' of an object is defined
only by the light coming from it.

That definition is at odds with multi-colour reproduction though. on a
computer screen a 'pink' is defined in terms of its red and blue (and to
a lesser extent green) content.

In 'real life' pink has a given wavelength and intensity.

True for pink (desaturated red), but it gets complicated with purple.
The texts refer to 'lines of purple' as it doesn't have a single
'frequency' and doesn't appear on a classic colour wheel, but you can
sure as heck see it. The same perceptual problem accounts for the
'bands' that appear in rainbows.

I stand by what I said originally. It's a bag of worms :-)

--
Regards,
Glenn Booth

Hi,

In message , Ian Molton
writes

Of course theres no real reason why either class of device should be
*incapable* of this - its just not within the bounds of *current*
technology... (of course, this presupposes a good 'white' light source
for the transmissive case)

Sure, but it's those real world limitations that led to the need for
different colour spaces and the CIE in the first place, so I guess it
brings us full circle :-)

Well, I am limiting my CMY values to those can be printed, yes. Find
me a printer that can give me a true black with no K, and I'll be
impressed (and probably shocked at the cost).

If I had one that could do that I'd be keeping it ;-)

The same goes for a CRT
that can reproduce the 'white' of an equitorial summer sun.

Give it time ;-)

I'd love to see the power supply for that thing...

I have to ask, why would they? If you can calculate and output the
correct voltage in the DAC on the fly at the required speed, why would

its far quicker to reprogram 3 256 byte (or 1024) LUTs than to
recalculate and redisplay an entire screenfull of output.

But then surely you can't do meaningful operations directly on the frame
buffer? Maybe I'm missing something here, as I've never worked on a
graphics system that works like this. You can kind of think of a RAMDAC
as having a lookup table of it's own, since each bit in the incoming
word will contribute an associated amount of voltage at the DAC output.
The over-simple analogy is 'buckets of charge', doubling in size with
each bit from least significant to most significant. A given binary
input will always generate the same corresponding voltage at the DAC
output.


you need LUTs? It's only a voltage varying with time, after all. It
doesn't become a 'colour' until the electrons excite the phosphors. I
can't really see the advantage of doing a memory access to get a
voltage value every time when a good DAC can do it so fast, and so
accurately.

I dont know of any systems that store the LUT in the card RAM. AFAIK
its built into the chip.

Ok. I'm on unknown territory in that case. I guess this system relies on
having the DAC as part of the graphics chip, which is very common, but
not universal. Matrox has several cards that use outboard DACs, so I'm
not sure how it would work.

I always quite liked the VD3. the DAC may not have been stunning, but
the accelleration was great (at the time) and the output was razor
sharp (easily the most jitter free images I have seen either now or
back then, including matrox cards. (the matrox did have better colour
output, but jittered more).

I think we're using different terms again. Jitter (to me) is a term used
to describe the DAC clock stability, which impacts quality of output,
since it dictates how closely the electron beam will hit the right bit
of phosphor as it scans with varying intensity. I don't remember the VD3
being that great, but it's a long time ago now. I was working on S3
chips for Videologic back then.

Having said that, it was a very fast 3D accelerator, and not a bad
effort considering it was their first go at a combined 2D/3D card. No
match for an SLI pair of Voodoo 2s though :-)

--
Regards,
Glenn Booth

Hi,

In message , Ian Molton
writes

Of course theres no real reason why either class of device should be
*incapable* of this - its just not within the bounds of *current*
technology... (of course, this presupposes a good 'white' light source
for the transmissive case)

Sure, but it's those real world limitations that led to the need for
different colour spaces and the CIE in the first place, so I guess it
brings us full circle :-)

Well, I am limiting my CMY values to those can be printed, yes. Find
me a printer that can give me a true black with no K, and I'll be
impressed (and probably shocked at the cost).

If I had one that could do that I'd be keeping it ;-)

The same goes for a CRT
that can reproduce the 'white' of an equitorial summer sun.

Give it time ;-)

I'd love to see the power supply for that thing...

I have to ask, why would they? If you can calculate and output the
correct voltage in the DAC on the fly at the required speed, why would

its far quicker to reprogram 3 256 byte (or 1024) LUTs than to
recalculate and redisplay an entire screenfull of output.

But then surely you can't do meaningful operations directly on the frame
buffer? Maybe I'm missing something here, as I've never worked on a
graphics system that works like this. You can kind of think of a RAMDAC
as having a lookup table of it's own, since each bit in the incoming
word will contribute an associated amount of voltage at the DAC output.
The over-simple analogy is 'buckets of charge', doubling in size with
each bit from least significant to most significant. A given binary
input will always generate the same corresponding voltage at the DAC
output.


you need LUTs? It's only a voltage varying with time, after all. It
doesn't become a 'colour' until the electrons excite the phosphors. I
can't really see the advantage of doing a memory access to get a
voltage value every time when a good DAC can do it so fast, and so
accurately.

I dont know of any systems that store the LUT in the card RAM. AFAIK
its built into the chip.

Ok. I'm on unknown territory in that case. I guess this system relies on
having the DAC as part of the graphics chip, which is very common, but
not universal. Matrox has several cards that use outboard DACs, so I'm
not sure how it would work.

I always quite liked the VD3. the DAC may not have been stunning, but
the accelleration was great (at the time) and the output was razor
sharp (easily the most jitter free images I have seen either now or
back then, including matrox cards. (the matrox did have better colour
output, but jittered more).

I think we're using different terms again. Jitter (to me) is a term used
to describe the DAC clock stability, which impacts quality of output,
since it dictates how closely the electron beam will hit the right bit
of phosphor as it scans with varying intensity. I don't remember the VD3
being that great, but it's a long time ago now. I was working on S3
chips for Videologic back then.

Having said that, it was a very fast 3D accelerator, and not a bad
effort considering it was their first go at a combined 2D/3D card. No
match for an SLI pair of Voodoo 2s though :-)

--
Regards,
Glenn Booth

On Tue, 13 Jan 2004 19:20:08 +0000
Glenn Booth wrote:

In 'real life' pink has a given wavelength and intensity.

True for pink (desaturated red), but it gets complicated with purple.
The texts refer to 'lines of purple' as it doesn't have a single
'frequency' and doesn't appear on a classic colour wheel, but you can
sure as heck see it. The same perceptual problem accounts for the
'bands' that appear in rainbows.

Im wondering if this is the same 'problem' as hearing a 50 Hz tone when
you play a 400 and 450 Hz tone ;-)

I stand by what I said originally. It's a bag of worms :-)

Shant disagree there :-)


--
Spyros lair: http://www.mnementh.co.uk/ |||| Maintainer: arm26 linux

Do not meddle in the affairs of Dragons, for you are tasty and good with
ketchup.

On Tue, 13 Jan 2004 19:20:08 +0000
Glenn Booth wrote:

In 'real life' pink has a given wavelength and intensity.

True for pink (desaturated red), but it gets complicated with purple.
The texts refer to 'lines of purple' as it doesn't have a single
'frequency' and doesn't appear on a classic colour wheel, but you can
sure as heck see it. The same perceptual problem accounts for the
'bands' that appear in rainbows.

Im wondering if this is the same 'problem' as hearing a 50 Hz tone when
you play a 400 and 450 Hz tone ;-)

I stand by what I said originally. It's a bag of worms :-)

Shant disagree there :-)


--
Spyros lair: http://www.mnementh.co.uk/ |||| Maintainer: arm26 linux

Do not meddle in the affairs of Dragons, for you are tasty and good with
ketchup.