Digital BW, The Print

Jon 

This is very helpful and I think we are talking apples to apples!  I don't
know what gamma is implicit in a linearized L* greyscale from, say, L*=15ish
to L*=96ish but I suspect it is a lot less than 1.8 (and especially 2.2).
Someone smarter than me can figure that one out.  So tackling the issue of
greyscale gamma, as you are doing, is, I think, something that needs to be
addressed.  

BTW if (1) CIELab depicts the nature of human vision well and, specifically,
L* the eye's reaction to luminance change, (2) the human eye is good at
adjusting for black and white points and (3) as Bruce Lindbloom demonstrates
the best fit gamma for Lab to be 2.2ish, then I would have thought that an
"ideal" gamma for a greyscale would have been 2.2 and not 1.8 but what do I
know on this...  ;-)

This issue of "what to linearize" or "what to linearize to" comes up
periodically.  I am glad you are taking an active approach to the subject
because I do think "linearizing L*" fails on its own when there is no colour
management involvement.  I really like though Roy's approach of profiling
the greyscale with an ICC profile and using that plus a CMM to manage the
density/luminance transformation that's necessary from document space to
print space.  I'm not sure that we have really got that approach fully
figured out though, yet the methodology is full of potential because it
could ultimately be device specific ie an individual could measure and
profile the actual greyscale of their own printer.  Short of this, the other
alternative is, as you appear to be doing, managing the gamma of the
greyscale during the curve construction process into a visually appealing
gamma.  

As a matter of interest, where are you setting 50% mid grey, or rather where
does it fall out, with your 1.8 gamma and standard ink black and paper white
points?


Steve

> From: piezobw <piezobw@...>
> Reply-To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> Date: Thu, 16 Jun 2005 14:35:59 -0000
> To: <DigitalBlackandWhiteThePrint@yahoogroups.com>
> Subject: [Digital BW] Re: How many shades of grey?
> 
> Steve,
> 
> I hope we are talking same apples and apples. If we were to linearize the
> process to 1.0 
> Gamma, it would be scientifically linear - but humans would see it as "flat".
> The reason we 
> choose to linearize PiezographyBW ICC and iQuads to a Gamma of 1.8, was that
> we wanted 
> a system which would behave the same from printer group to printer group. In
> other 
> words, the output of an 1160 would closely replicate the output of a 7000. Now
> the end 
> points may be different. dMax in one may be greater than the other (and no one
> would 
> want to cut back on dMax!) But when printed - would the same image have about
> the 
> same relative contrast from printer to printer? And could that contrast be
> very very similar
> between printer families. So that is Gamma in an inkjet useful way, and we
> bring Gamma 
> into our linearization process. Perhaps that is where our process differs.
> 
> On the initial Beta of K7 we did not linearize with our intended Gamma - and
> as we 
> suspected we had a range of comments about the output in relation to either
> highlights 
> (mostly) and to some shadows.
> 
> So part of our final process is to linearize the process so that there are
> nice even steps 
> between the grays. But as I mentioned, if we make that linearization 1.0 - no
> one is going 
> to "see" it as a visually logical progression from light to dark. Humans
> prefer contrast. So
> which contrast? If we darken and lighten the near-to-the-ends to create a
> contrast - we 
> need to specify one. We choose Gamma 1.8 for several reasons.
> 
> So a Gamma is applied in our linearization process that makes sense to the
> average 
> human eye. Another way to think of Gamma of course, is as a compensation for
> the 
> human visual system.  Linearization on its own may not always make the best
> visual sense 
> - unless you linearize to a system of tone that a human responds. I can't
> apparently 
> emphasize Gamma enough when talking about linearization. Sometimes scientists
> when 
> they begin producing drivers and the like for humans - forget all about the
> human 
> condition, or worse yet they are unaware completely of the human condition.
> They then 
> tend to make tools which are not consistent from printer to printer, but
> satisfy their own
> scientific process. EPSON too, obviously has computer scientists making their
> printer 
> drivers and not artists or psychologists. There are other creators of software
> too that have 
> complete disregard for the human condition - and yet humans are the intended
> target 
> audience. The problem here is that inkjet printing is now so very much about
> photography 
> and image interpretation - whereas earlier in its infancy it was about
> scientific 
> visualization and graphics.
> 
> We built a software to enable this process in-house (relating output to how
> humans see). 
> So we're supplying our own curves to Roy for his QuadTone RIP. But a user can
> make their 
> own with his process as you know. We think our supplied final curves will be
> best. 
> However, not everyone will agree. So they can be tweaked and or replaced. I
> suppose we 
> could make custom curves for users - do not know if we will go down that road
> though.
> 
> Why not Gamma 2.2 you say? or you might say next!! :)  I believe that humans
> respond 
> better to a Gamma of 1.8.  And any Mac or PC can be calibrated to 1.8. Just
> preference.
> 
> best,
> 
> Jon Cone