Dan Margulis Applied Color Theory
In the Wake of the 2007 Advanced Classes
In the wake of the ACT advanced classes
Posted by: Dan Margulis
Mon Dec 17, 2007 6:19 pm (PST)
My arrival home late Friday night culminated almost
four months of nonstop travel. I get three weeks off, and then it all
begins again.
This is the time of year when I do advanced courses,
open only to those who have already suffered through the three-day ACT
course. These are the most valuable classes of the year for me because of
the skill level of the students. The first class exposes a lot of of errors
in technique. In these advanced classes it's unusual to find students
making outright errors, so it's more of a comparison of techniques without
introduction of extraneous silliness.
Also, the basic ACT course doesn't change much from
year to year. The fundamentals of color correction haven't changed in a
long time. The advanced course is a new adventure every year, because it
features tricks that I've learned since the previous year. That was
particularly the case this time. The images that get worked on are turned
over much more frequently than in the basic course. This year's advanced
classes had to work 26 images, of which 9 were new. One of these was the
"Giselle" image by John Ruttenberg that provoked a lengthy thread
on the list earlier this year. The classes got an explanation of what was
desired, a high-res JPEG, and also the raw file, with free rein to do
whatever they wanted. The image, as you may recall, is shot for promotional
use by a ballet company. It portrays a ballerina who is supposed to be
dead, risen from the grave, and dancing by moonlight, represented by
extreme blue lighting. The different takes on the question of how far to go
in eliminating the cast and how much variation to incorporate in the
skintone were interesting. At some point I'll post some of them for
discussion.
When I'm in doubt about how well a certain method will
work on a certain image, there's no better way of getting a definitive
answer than by assigning 20 experts to try to correct it with their own
methods. That's what I did this year with the "picture postcard"
workflow that I showed at Photoshop World. The results indeed validated the
technique. OTOH, that 17 of the images are *not* new means that not only
can the classes' work be compared against mine, but also against that of
past years, because I save the better results.
People who are so committed to image quality that they
suffer through two (and sometimes more) three-day sessions with me are
likely to be sophisticated enough that we should listen to their message,
so each year I've summarized what the advanced classes had to say about
industry trends. This year it was a little difficult to do this, because
the groups were not typical of those who had attended in the past. But,
here goes.
We usually schedule two of these advanced classes,
always late in the year so that I can develop a new curriculum and repeat
it. This year, we had enough advance signup for three courses (these
classes aren't open to the general public, only those who have taken ACT
already). We scheduled them for San Diego, Washington, and Toronto in that
order. Unfortunately, the San Diego session got smoked out. We were able to
reschedule it for week before last, which accounts for why this message
comes late in the year.
We had one last-minute cancellation, so there were 20
people--and the trend was obvious. The following five persons represent
groups that, five years ago, would have constituted the majority of
students.
*One person works in the production department of a
daily newspaper.
*One owns a fine-art output service in a geographically
remote area.
*One person is a full-time retoucher in a New York
advertising agency.
*Two people work as retouchers for companies that
produce a lot of printing, but do not accept graphic work from clients.
Then came the the folks who would not have been seen
five years ago.
*Four might generically be described as professional
photographers, but this is for want of a better phrase. Each is heavily
involved in doing image manipulation, usually for print purposes, that is
not typical of past practice.
*Four came from the biggest growth area over the last
two years--successful, mature professionals in other fields who happen to
be deadly serious about reproducing their own photography. There's doctors,
and there's lawyers, and business ex-ecutives, but they aren't all made of
ticky-tacky and they definitely don't look all the same, particularly their
images.
*Seven (count 'em, I had to) are independent. They take
assignments wherever they find them and exploit their imaging skills
usually in conjunction with something else, like design or writing, so that
they can offer a whole package to clients.
An additional wrinkle. The Toronto class was
international. In assigning seats to these classes, I tried to divert
everyone whose first language was not English to this class. Our lone
Russian was forced to cancel and was replaced by an American. This left us
with two members of the class hailing from different regions of Italy, one
each from Portugal, Spain/Puerto Rico, and Japan, and two from the United
States. No Canadians. On a personal level this was one of the most
enjoyable classes I've ever conducted. The class went everywhere as a
group, including a special preclass dinner where we sampled Quebecois
cuisine.
Back to trends. First, although there were exceptions,
by and large these 20 were significantly stronger than advanced classes in
recent years. I think I can say why, too. I am not sufficiently
ill-mannered as to inquire who is footing the bill. However, one would
guess that 16 of these 20 people paid their own way, some in an indirect
fashion. Such people are apt to be highly committed.
That trend is actually disturbing. For the past several
years I've been noting a wider and wider gulf between the skills of those
taking my ACT (not advanced) courses in the United States, and in Canada.
At first I suspected that since Canadians can't even *spell*
"color" they probably aren't capable of working well with it,
either.
The actual reason, I'm convinced: many U.S. companies
have cut way, way back on training their staffs. In Canada, companies
usually put a higher value on developing the individual. Consequently, in
typical Canadian classes I'd guess that about three-quarters of the
students are being bankrolled by a company, whereas in the U.S. it's now
less than half. With great respect to the many truly talented people whose
companies have invested in them, if on the one hand you have an individual
who is being sent for training at his company's behest, and on the other
someone who is paying his own way or else really had to put the screws to
his boss to get the funding, the chances are that the second individual may
be a bit better motivated and likely more skilled as well.
I've pointed out previously that commercial printers
rarely send people to my classes, which is an extreme case of penny-wise,
pound foolish. The people who *did* attend this year had to figure out
whether the benefits outweighed the cost over the long term. A printer (and
to a lesser extent, the guy who works for a large newspaper) all he has to
do is pick up one tip that saves one job or one ad from being remade, and
the course is more than paid for. OTOH, dissatisfied clients are extremely
expensive for everybody, not just printers.
The other striking trend was prosperity--and a blurring
of job description. By my count, at least 11 of these 20 people, probably
12, are in the business of marketing an improved quality of image to their
clients. Some of these clients are companies, others are artists or
entrepreneurs, and surprisingly, many are photographers. That's
right--these attendees (including a couple who are nominally photographers
themselves) are offering their services to quality- oriented photographers
who can talk the talk with their own clients but are unable to walk the
walk.
The photographers in the class may disagree with my
characterization, but I'm not sure that "photographer" is an
adequate description any more for what they do. At best one of the four
makes most of his money by the act of taking photographs. The rest, IMHO,
are competing with the likes of the seven "independents". Each of
those independents, I'm fairly sure, can provide photography services on
demand, probably of the point and shoot variety, but good enough for many
purposes. Similarly, many of these photographers can provide page layout or
web design services, nothing too fancy, but good enough in many cases. It's
somewhat hard to distinguish one imaging specialist from another at this
point: one may be an imaging specialist who happens to be an experienced
art director; another may happen be an expert web designer; a third may be
an expert photographer; and a fourth may be able to offer specialized types
of printing--always in addition to their basic imaging orientation.
For reasons of manners, I was not about to go into the
particulars of anybody's business affairs. However, we did have some
roundtable discussions from which I got the distinct impression that almost
all of these 11 or 12 independent people are doing well. While this is no
more of a verification of the old saw that if you can provide something
that the client likes at a quality level he's satisfied with and at a price
he's willing to pay, you have a recipe for commercial success, it should be
heartening news to members of this list, who tend to be quality-oriented
themselves.
Each year, I ask several questions of the group to see
if any trend can be detected. For example, I ask those that deal with
commercial printers whether they find them to be more sophisticated, less
sophisticated, or about the same as two years ago. I ask the same of those
who receive pictures directly from photographers. And I ask whether their
company finds it more or less difficult to hire qualified persons than two
years ago, or about the same. None of these questions got satisfactory
answers this year because the group didn't have the experience with them.
Few deal with multiple commercial printers (those that did thought they
were about the same). The guy who works for a newspaper said that it was
his impression that photographers were more sophisticated, but all of the
independents who are dealing with photographers are biased because the
photographers use their services precisely because they aren't
sophisticated enough themselves. And nobody in all the classes AFAIK is
involved in the hiring process.
Clear consensus, though, on three things: a) clients
are demanding faster turnarounds than two years ago; b) clients are often
insisting on the use of nonprofessionally shot images in cases where two
years ago a superior original would have been provided; c) clients tend to
believe that more repair can be done easily in Photoshop than is actually
the case.
One photographer told a story that takes the prize in
illustrating this last point. It seems he was on a fashion shoot where the
male model was wearing a sport jacket. The written instructions for the
shoot stated that the jacket was to be open, unbuttoned. The photographer
shot it that way, but remarked to the art director that it looked a little
horsy, and perhaps they should play it safe by shooting a few exposures
with the jacket buttoned, just in case the client changed his mind. The art
director replied, "don't worry, if they decide to do that we'll just
button it in Photoshop."
Getting back to Photoshop technique, it's hard to be
specific without the ability to post images. The skill level of the classes
was such that it really highlighted the importance of certain steps. We're
always comparing eight or more corrected versions of the same image (the
seven class members, plus mine, plus, if applicable, particularly good
renditions from past years). Of these, maybe only one or two would have
some kind of technical defect that made it obviously noncompetitive with
the rest. (In basic ACT, the number of defective images is at least twice
that, probably more). So, we're comparing five or six images that are all
technically fine and trying to make a decision as to which is the best.
When people won decisively (the sort of image where you
can take one look and see that it's better than the other contestants) it
was usually because of better color variation, of the kind introduced in
the Man from Mars Method, which I have since updated. Also, people were
able to put together clear wins by creative blends of channels to gain
contrast. Often this required multiple applications, e.g. in the case of a
portrait, apply the green to the RGB in Luminosity mode, flatten, and then
apply the *resulting* green again in Luminosity mode. Plus, we had a couple
of very intelligent uses of painting.
Workflow-wise, I noted more use of Smart Objects than I
was expecting. Particularly, the advertising agency, whose final output is
almost always CMYK, retouches heavily in LAB. They save all these LAB steps
as Smart Objects in case they wish to edit them. I had been thinking that
Smart Objects are primarily useful if you want to edit filters such as USM
or blurring, but there's certainly no reason not to use them for curves as
well.
It was also interesting to see how many people are
deliberately going for overkill and then using layer opacity to achieve the
desired effect, as opposed to trying to hit perfection with a single pass.
This is especially valuable when we know we have to go in a general
direction but are uncertain of exactly how far to go. It's long been my
practice, before embarking on any LAB adventures, to immediately save and
set aside a conservatively-corrected version. Then, I proceed to the LAB
correction. If, at the end, I consider that the correction has made the
image too loud, I can blend the conservative version back in at, say, 10%
opacity.
An unusual variant of this practice showed up in the
Giselle image, which is, you may recall, about as clear an example as can
be found of the kind of image where the optimal result is hard to visualize
without experimentation. The lighting is deliberately overwhelmingly blue.
For a human observer, both chromatic adaptation and simultaneous contrast
have a fierce effect. We will perceive far less blueness than the camera
does. Also, we will see all kinds of color variation in the face where the
camera just captured a flat blue.
The only question is, how far to go? We know that
evolution has adapted us to compensate for lighting variations and to
emphasize small variations. But this aberrant lighting is so impossibly
strong that we can't ignore it altogether. The client certainly doesn't
want us to correct it so that it appears to have been shot in sunlight, and
it would be ridiculous, as well as tasteless, to do that.
Or *would* it? That's exactly what a couple of class
attendees did. They had a version that blew out the blueness completely,
yielding a white gown and a normal fleshtone. This was not attractive in
and of itself, but turned out to be just the thing for blending. Everybody
in the class had a different idea of what result to shoot for, but the
majority of them, including mine were improved when, at the end, this crazy
nonblue image was blended in at around 20% opacity. The lighting-is-normal
version is such a radical correction that it may not be evident that it has
the same role as my conservative version does when I do extensive LAB work.
It certainly wasn't evident to me, but the next time I have at that image,
I'll know better.
Dan Margulis
___________________________________________________________________________
Re: In the wake of the ACT advanced classes
Posted by: "Ron Kelly"
Tue Dec 18, 2007 5:43 am (PST)
Dan:
Thanks for the lengthy description of the action; I
wish I was there, even though I don't know how to spell "color".
Maybe that's why that word is always underlined in my emails to
colortheory.
Those of you out there who have never taken a class
from Dan, I urge you to do so if you are serious about your work. What you
learn won't go stale in a hurry, as Dan has pointed out - the image
correction basics haven't changed.
The method of the class, by competition and comparison
between everyone's results is very effective and also fun. Dan's not always
the humblest of people, and it's entertaining to see his image bested, and
doubly satisfying when *you've* done it.
I learned a lot in a hurry and I'd like to come back
for another sometime, loaded for bear.
Cheers,
Ron Kelly
___________________________________________________________________________
Advance Class Question
Posted by: Andy Adams
Wed Dec 19, 2007 5:18 am (PST)
I am curious, has anyone taken the Advance Class more
than once? I took it in 2004 and would love to take it again. Last time I
took the class I came in feeling like a "top dog", and by time it
was over I had my tail between my legs realizing my tech wasn't so high. I
think I am ready for more abuse.
Andy Adams
____________________________________________________________________________
.
Re: Advance Class Question
Posted by: Dan Margulis
Thu Dec 20, 2007 5:03 am (PST)
I can think of eight people off the top of my head who
have taken the advanced course twice, going back to 2004, and so far two
people who have taken it once and two who have already taken it twice have
pre-registered for 2008. This suggests that around 7 or 10 percent of the
people who take it once later take it again.
The basic organization of topics hasn't changed
drastically between 2004 and 2007 (it may next year). The lectures are
structured the same but around 50% of the content changed in that time.
That's a guess, but the following isn't: of the 26 images that the 2007
group worked on, 8 were introduced in 2004 or earlier. One of those eight
is definitely being replaced for 2008 because I have a more instructive
exercise now.
The people who take it again don't mind hearing me
repeat myself but are more interested in the comparisons of work (as am I).
In the basic class it's valuable to watch other people make the mistakes
that they inevitably do. In an advanced class you don't get too see too
many mistakes from the others, but you do get to see a lot of intelligent
planning that may well introduce some ideas that you hadn't thought of.
My plan is to revise the basic ACT course next year to
accommodate more of the workflow I developed this year. That revise is
harder than it sounds because the ACT curriculum is time-tested and doesn't
have a lot of obvious weaknesses that can be easily cut out.
Depending upon the decision with the basic course, I
may or may not have some kind of radical revise for the 2008 advanced
classes. Presumably I'll know the answer by the time people finalize their
registrations in late February.
Dan Margulis