Re: U.S. Web Coated (SWOP)v2 vs. SWOP2006_Coated5v2
Re: U.S. Web Coated (SWOP)v2 vs. SWOP2006_Coated5v2
- Subject: Re: U.S. Web Coated (SWOP)v2 vs. SWOP2006_Coated5v2
- From: Todd Shirley <email@hidden>
- Date: Wed, 9 Jan 2008 17:25:41 -0500
First, thanks to Marco and Harold for some truly in-depth analysis on
the differences between the profiles in question. One of the great
things about submitting a problem to this list is that one is
immediately called to task for imprecise language and unclear
procedures. So let me clarify a few things about my original post:
On Jan 7, 2008, at 1:29 PM, email@hidden wrote:
In his initial email Todd reports on differences in Lab for a given
inking
between the 2 profiles. Unfortunately he probably utilized the A2B0
Xform
(perceptual) and interpreted it as being colorimetric which it is NOT.
What I did is ASSIGN (in Photoshop) the two profiles in question to an
untagged CMYK image and noted the difference in LAB values in some
shadow inkings. I didn't actually do any conversion, so I believe this
is a colorimetric reading. Correct me if I'm wrong.
On Jan 6, 2008, at 10:45 PM, Marco Ugolini wrote:
The thing I find interesting/confusing is how the same CMYK values
can
have such significantly different LAB values when comparing U.S. Web
Coated (SWOP)v2 to SWOP2006_Coated5v2.
I am not sure what is meant by Todd as "significant". Without a number
specifying what qualifies as a "significant" difference (a 1.5
DeltaE 2000
value, for example), I am not sure how to weigh that assertion.
The "significant" differences I am talking about really only occur in
the deep shadows, so indeed the average DeltaE between the two
profiles may well not be that large. The issue I was trying to address
in my original post is that ASSIGNING SWOPv2 to to an untagged CMYK
image with a lot of detail in the deep shadows causes those shadows to
plug up and the details disappear. This can cause quite a significant
visual shift in the overall appearance of the image, while at the same
time if one were to average the deltaE for the whole image, it would
seem to be minimal. It is much more visually jarring when most of an
image looks correct but certain parts (like the deep shadows) are
really off.
On Jan 7, 2008, at 1:52 PM, Steve Upton wrote:
At 11:29 PM -0500 1/6/08, Roger Breton wrote:
...I'd wish our friend the color
wizard, Steve Upton, would whip his ColorThink wand and highlight
in his 3D
graph where the difference lies between the two datasets. I'll bet
Steve
will say it already does and I've been sleeping all this time!
For once I am happy to say that this feature is, in fact, available.
Setup the ColorWorksheet with the two sets of data and delta-E
turned on.
Click on the header for the delta-E column twice (once to sort by
delta-E, the other to reverse the sort and bring the worst offenders
to the top)
Then, graph the colors in the 2nd list (by selecting the popup above
the color list, and choosing 'graph')
Then graph the delta-E values (again, select the popup above the
delta-E list)
Finally, select the first color in the list (the highest dE value)
and then shift-select somewhere down the list so you have a group of
the highest dE values selected.
The graph will now have vectors depicting the color differences
between the two lists AND the selected colors with have white cross-
hairs highlighting them.
I more-or-less followed Steve's instructions, and indeed found that
ALL the high delta-E values are in the deep shadows, and pretty much
the darker the color, the bigger the difference! What I was asking
about in my original post (and still trying to figure out) is why the
old profile "kills" shadow detail while the new one doesn't. I will
list the steps I did in ColorThink Pro if you would like to replicate
what I am seeing.
1. Used Colorport to generate a "TC3.5 CMYK" chart. Cropped chart in
Photoshop, duplicated it, assigned U.S. Web Coated (SWOP)v2 to one,
assigned SWOP2006_Coated5v2 to the other, converted both to Lab, saved
as TIFFs.
2. Dragged both charts into a ColorThink Pro worksheet, turned delta-E
on, but then changed to Delta-Lab because I am most interested in
jumps along the L axis.
3. Created a list of all 432 colors using the "Select Target Values to
Create a List" tool, then sorted bey delta-L
There are a lot of ways to interpret this data, but if you do the
above steps, you will see that 5% (22) of the colors have a delta-L
over 4 and they are all in the deep shadow. When SWOP2006_Coated5v2 is
assigned to the chart, these 22 colors have an average L value of 13
and when U.S. Web Coated (SWOP)v2 is assigned, these 22 colors have an
average of 8.4. So while an average of all 432 colors shows that there
is only a 1.63 shift in the L value between the two (SWOP2006 is 1.63
"lighter" than SWOPv2), there is an average jump of 4.6 in the worst
5% which is all right around the visible threshold of detail going
into black.
To (over) simplfy: Assigning U.S. Web Coated (SWOP)v2 to an untagged
CMYK image can kill shadow detail. What I'm curious about is why. I'm
sure others have seen this same thing happen - what is your take on
what is going on? Roger mentioned that the old SWOP is based on the
CGATS TR-001 1995 dataset. Does that data really plug up so badly in
the shadows? And if so, why? I mean really, has web offset technology
really advanced that much in 10 years? Why would TR-001 be so much
different from the 2006 datasets? Even given the differences between
the G7 methodology and the old TVI/density process control, I would
think that the presses behave fundamentally the same as they did back
then.
-Todd Shirley
_______________________________________________
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