Digital BW, The Print

John,

You wrote:

>I am attempting to calibrate my printer output using a calibrated
>X-Rite 810 densitometer.  I am using MIS FS inks through an Epson
>1160 onto PhotoRag 308 paper; I am on a Mac computer. ...

>...how shockingly non-linear they are.  I would have expected,
>for example, that the 50% reading would have been close to
>1.7/2 = .85.  The Measured value of around 0.6 is WAY off!

>...

You're densitometry reading are correct.  Whether the system is "WAY off" is
a question that I've never found a really good answer to.

In general, I've been told that human visual response is logarithmic --
which is what those densitometry readings are.  However, whether this is an
exact relationship or just a close approximation is not clear to me.

As a practical matter, in designing the variable-tone system, I used Piezo
outputs as the standard, because I wanted to be able to use the same file
for both the Piezo driver and the Epson driver, using the VM curves.  When I
took a number of Piezo ink samples (throwing out the clearly bad inks),
printed them only from newly-filled cartridges (thus avoiding the "greenies"
and other problems that infect a number of CIS systems), and used the latest
profiles, I got an average 50% density of 0.61 -- just what you report.

So, I have asked both here and elsewhere, "What is linear" -- how does one
find the theoretical middle of our visual response.  I have received no
adequate responses.

I assume that Cone and his software designers had some printing industry
background and expertise that went into their system (or not -- who knows).
Again, as a practical matter, that system works and many of us use it.  So,
for me it became more of a practical issue of consistency with a system that
worked well and was widely used by those likely to use the variable-tone
system.

Have you tried printing a "black ink only" 21-step test file?  On the ones I
tried, the "50%" is closer to 50% of the reflection of the paper white --
yet another definition, and even further "off" than the Piezo definition of
the "50%" value.

Paul
http://www.PaulRoark.com

on 4/23/02 8:02 PM, Paul Roark wrote:

<Snip>
> So, I have asked both here and elsewhere, "What is linear" -- how does one
> find the theoretical middle of our visual response.  I have received no
> adequate responses.

> I assume that Cone and his software designers had some printing industry
> background and expertise that went into their system (or not -- who knows).

Paul

I don't know where you've asked this question but these are the best two
lists I know for such topics:

colortheory@yahoogroups.com
colorsync-users@...

The first one is obviously a yahoo group, so easily subscribed to, and is
hosted by Dan Margulis. The Colorsync list is subscribed to through the
Colorsync section of Apple's website (apple.com).

Both lists love arcane theoretical topics (or so they often seem to me <g>),
but the nose in that department probably goes to the Colorsync list. Both
lists have a fair amount of press and prepress people on them. If you've
tried these lists already it may be worth another shot - you know how lists
are WRT timing...

Also, Stouffer offers many "targets" that probably represent industry
standards. 

http://www.stouffer.net

Todd

on 4/23/02 8:02 PM, Paul Roark wrote:

> So, I have asked both here and elsewhere, "What is linear" -- how does one
> find the theoretical middle of our visual response.  I have received no
> adequate responses.

As a follow up to my last post, this might be all you need, at least WRT
determining what is a linear ("industry standard") step wedge:

http://www.stouffer.net/R2110spec.htm

Todd

In a message dated 4/23/02 5:25:21 PM, chervinsky@... writes:

>If you look at the numbers
>carefully, you will conclude as to how shockingly non-linear they are.
> I would have expected, for example that the 50 % reading would have
>been close to 1.7/2 = .85.  The Measured value of around 0.6 is WAY
>off!

You mention that you don't think the paper will make much difference, but the 
chief coulprit in all this is dot gain, which is a paper related function. 
When you cover the entire area, then dot gain does not matter; its all 
covered already. When you place no dots and leave the paper white dot gain 
does not matter either. But at fifty percent dot gain matters hugely. I have 
worked with systems that had a 46% dot gain at the midpoint... meaning that 
printing 50% gray would get you a 96% gray...basicly black. Densitometer's 
specialty is determining dot gain. Print that same ramp on several widely 
differing papers, and see what midpoint values you get... variations of 30% 
are possible.

C. David Tobie
Design Cooperative
CDTobie@...