Re: Epson canned profiles
Re: Epson canned profiles
- Subject: Re: Epson canned profiles
- From: ben <email@hidden>
- Date: Tue, 17 Apr 2018 12:07:47 -0700
- Sun-java-system-smtp-warning: Lines longer than SMTP allows found and truncated.
On Apr 17, 2018, at 11:14 AM, Andrew Rodney <email@hidden> wrote:
> The first URL wasn't set to dE 2000/D50!
...the URL which you supplied, of course. I've been very careful to let you do
whatever math makes you feel good, and simply repeated the various numbers back
to you. That you can't punch the numbers properly into your calculator is
_your_ problem -- as much as is your obsession over a fractional percentage of
a calculation error when we're doing back-of-the-envelope scale checks.
And, all the while that you're obsessing over fractions of a dE, you're
_equally_ obsessing over your bizarre misconception that 1 dE is a magic
threshold, with anything greater a real color and anything less invisible.
The point you're missing is that there are contexts in which a fraction of a dE
is quite significant _and_ there are contexts in which even as much as 10 dE
just doesn't matter at all.
Even in a printing environment!
Get an inversion in your profile of even a very small amount and it'll stick
out like a sore thumb. Indeed, one can imagine a profile with very impressive
average dE fit results -- even better than those that you've been bragging
about -- that was actually a bunch of zig-zag stepwise inversions that would
create truly miserable output, even to the completely uninitiated. Or even a
profile that has 0.1 dE error for every patch but the white and black points,
and had a 0.99 dE error for both, in different directions; that's not going to
be pretty at all. But the average error is 0.1000001 dE, and the maximum error
is under 1 dE!
In the other direction, I've also many, many times made greeting cards on
exhibition-quality stock with matching envelopes on uncoated cotton rag bond --
the type that, in a different era, would have been considered high-end typing
paper. Same photograph printed on both the front of the card and the back of
the envelope. The black point of the card was something with single-digit L*
values; the envelope might even have been in the low 20s, certainly no better
than high teens.
That there was a difference was obvious, as that the card was unquestionably
better, richer, with more contrast. But the envelope didn't look incorrect or
off or bad or anything; the only subjective difference was the paper. Look at
the envelope in isolation and the miracles of adaptation kick in, giving a
roughly similar perception of tonal range. There were zero perceptible hue
shifts. Shadows and highlights still looked perfectly reasonable, only a bit
less detailed. Imagine laying a very thin, very slightly opaque white
transparency sheet on the card, and that's what the envelope looked like. Open
the image in an editor and add a solid white layer on top with 10% - 15%
opacity and you've got a good approximate soft proof.
That's waaaaaaay more of a dE shift than anything we've been arguing about,
and, in the context, it's not merely utterly irrelevant, it's actually damned
impressive.
> But the point is, CT&A is a great application for visually examining what dE
> values can look like on-screen.
>
> http://digitaldog.net/files/dEtwoblacks.jpg
Right. Now remove the space between the two, as if you're printing an image
with some deep shadows where you actually want to be able to get an hint of the
shape of the shadows.
You _do_ understand, do you not, that a vital part of a profile is
characterizing the printer response in such situations so that, when you _do_
have fractional-dE patches next to each other, they get rendered properly? Why
else do you think you need a chart with more than a dozen or so patches? Have
you no idea why somebody would print a chart with several thousand patches,
especially for a printer that's not very well linearized? Why a particularly
linear device such as a camera can be usefully characterized by a single
nine-number matrix?
No, of course not. To you, a dE is a dE is a dE, because you don't even
understand what on earth a dE actually is in the first place or the meaning of
its calculation.
Big hint: _why_ was the original formulation the geometric distance in Lab
space? Why have there been new formulations? Why have there been new color
spaces invented after Lab? And what's the actual scale of the differences
between all these different models? In which contexts do the differences in
scale matter and which do they not?
You can't simultaneously fault me for using this-or-that dE model instead of
this-or-other _and_ claim that sub-dE differences are "invisible."
Unless, of course, you're completely innumerate -- a fact to which you've
repeatedly admitted and demonstrated.
Cheers,
b&
_______________________________________________
Do not post admin requests to the list. They will be ignored.
colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden