Dan Margulis wrote:

No, I read the article, and Bruce has unquestionably decided that stacking
multiple adjustment layers is somehow better than consecutive curves...

It is also interesting to note that the image in question can be very
effectively corrected with a trip into LAB and "one" curve adjusting the
"a" and "b" channels.

--
regards,

Lee Varis
varis@...
www.varis.com
888-964-0024

Herbert writes,

Does using Adjustment Layers reduce image degradation? I have never heard

of
this before>>

Nor have I, so I tested it on one pair of RGB and one of CMYK files,
applying three sets of massive curves consecutively to the base layer of
one image and applying the same curves as three adjustment layers to the
other, then flattening the image. On a third CMYK image, I applied one set
of curves and one selective color correction, again comparing doing it
consecutively vs. two adjustment layers.

In all three cases, the resulting pairs of images were identical, pixel for
pixel.

Bob Smith writes,

All of the experimenting is done by altering the adjustment layer a

number of times, not the image. I don't think Bruce was talking about the
effect of stacking multiple adjustment layers.>>

No, I read the article, and Bruce has unquestionably decided that stacking
multiple adjustment layers is somehow better than consecutive curves. As
Lee Varis has also pointed out, this is not correct. Here is the actual
quote. (The scenario is: he is correcting an image with a blue cast. He
recommends using the gray eyedropper to do this. Not surprisingly, this
moves the image in the right direction but is not quite accurate enough,
hence he needs a second curve to correct the first one.)

"You may be tempted to adjust the curve created by the eyedropper to fix
this. My advice is to use a second set of curve adjustments: If you're
using Adjustment Layers on an 8-bit-per-channel file, you don't need to
worry about successive rounds of curves degrading the image, and if you're
working on a high-bit file, you have more than enough data to withstand two
rounds of curves. It's simply a lot easier to fix the problem using a new
curve than it is to tweak the old one."

Dan Margulis

Lee Varis wrote:

I don't know where Bruce got this idea but it's not true. Basically the
layer "stack" is calculated sequentially from the bottom up - it has to
work that way.

My reading of Bruce's use of an adjustment layers simply means that an
adjustment layer is applied to the image once whereas someone (like me!) who
wants to fiddle endlessly with an image might other wise might darken, then
lighten, then change gray balance, etc etc. All trying to figure out which
version I like best. With an adjustment layer you're really only making one
adjustment to the image. All of the experimenting is done by altering the
adjustment layer a number of times, not the image. I don't think Bruce was
talking about the effect of stacking multiple adjustment layers.

Bob Smith

If you are on a Mac you can use Applescript to have Photoshop run Actions. It
should be fairly easy to set up a loop to perform an action x number of times.

I haven't actually tried any scripting of Photoshop myself, but it should be
pretty easy as far as Applescript goes, since the only command is to 'do script'
with a specified action.

RJay

http://www.2540dpi.f2s.com

Dave wrote:

How do I make an action in Photoshop repeat automatically?
Is it possible?

-------------------------------------------------
Everyone should have http://www.freedom2surf.net/

How do I make an action in Photoshop repeat automatically?
Is it possible?
Thanks
Dave

--

One of these days I'm going to get the hang of sending mail to the group
instead of just hitting reply:

"Ripka, Herb" wrote:

... when you use
Adjustment Layers to edit the image, all the edits are calculated at the
same time when you flatten the image. This approach degrades the image much
less than burning successive rounds of corrections into the image one by
one. . . ."

Does using Adjustment Layers reduce image degradation? I have never heard of
this before

I don't know where Bruce got this idea but it's not true. Basically the
layer "stack" is calculated sequentially from the bottom up - it has to
work that way. Now the final pixel rendering happens either when the
file is flattened or when it is "ripped" to print but the calculations
are still
done sequentially so there is bound to be successive degradation - the
more layers, the more degradation. Most of the time this isn't such a
big deal. You still wouldn't want to pile on multiple layers of
hue/saturation, channel mixer and curves if one curves layer could do
it, however,

The only app I know of that did these sort of complex calculations in
one step was Live Picture. It's layer structure was/is completely
different and image transforms were recorded into one complex equation
that kept getting more complex as you added layers of transforms. The
result was calculated in 16 bits only once when the final file was
"built". All masks were calculated at 12 bits so gradations were
smoother and less likely to alias if multiple blends were stacked on top
of each other. In the end, the advantage of the high bit transforms in
Live Picture were not that noticeable and not enough people used Live
Picture to insure it's survival. So, Live Picture is now Dead Picture.

--
regards,

Lee Varis
varis@...
www.varis.com
888-964-0024

on 21/6/01 10:58 PM, J Walton at j.walton@... wrote:

I read this on http://www.creativepro.com/story/feature/13486-2.html
Out of Gamut: Color-Correcting Photographs in Photoshop by Bruce Fraser

BTW, anyone who says that their tool of choice is ALWAYS curves needs to
master a few more tools, famous author or not.

Teaching granny to suck eggs?.... :-)

Bruce is my hero...if he says he ALWAYS uses Curves adjustment layers
(except in 16bit/channel mode) then that's what he does. If I don't agree
with him I don't have to work the way he does. Each to his own.

BTW, at a guess I'd say Bruce Fraser knows more about Photoshop than most.
If I was compiling a list of Real World Photoshop Gurus I'd put him in the
first ten!

If you read Real World Photoshop then you'll know he is, indeed, the master
of "...a few more tools."

--
Respectfully,

Shangara Singh.

Adobe Certified Expert/Photoshop 6.0
------------------------------/---------------------------------
Graphics Photography Multimedia Websites
\_______\__________\________\_______
www.oxyopia.co.uk www.photoshopAce.co.uk

Not if you pile adjustment layer on top of adjustment layer. 10 adjustment
layers, or 10 curves applied in the same order would, in my view have the
same effect. I have not tested this out, of course.

However, if you were to simply make adjustments to the same adjustment layer
over and over again I suppose you would effect less image degradation. But
how many huge giant corrections are you doing so that your image turns to
mush after even 10 adjustment layers? One bad move is infinitely more
damaging to an image than 10 or even 20 sensible moves, but adjustment
layers are very useful anyway, IMO.

BTW, anyone who says that their tool of choice is ALWAYS curves needs to
master a few more tools, famous author or not.

J

I read this on http://www.creativepro.com/story/feature/13486-2.html
Out of Gamut: Color-Correcting Photographs in Photoshop by Bruce Fraser

"Photoshop offers many different ways to fix this problem of too much blue,
but my tool of choice is always Curves. However, if I'm working on a
high-bit file, I can only apply the Curves adjustment to the actual image
data as Adjustment Layers are unavailable (for more on this topic, read "The
High-Bit Advantage"). For 8-bit-per-channel images, I always use a Curves
Adjustment layer, both because it's more flexible, and because when you use
Adjustment Layers to edit the image, all the edits are calculated at the
same time when you flatten the image. This approach degrades the image much
less than burning successive rounds of corrections into the image one by
one. . . ."

Does using Adjustment Layers reduce image degradation? I have never heard of
this before

--Herbert Ripka
Greendale, WI

Andrew Rodney <andrew@...> writes:

Lastly, scanner and camera manufactures have to
decide what they want to do with the color of their devices; funnel it into
a monitor like space (so any bonehead likes the color they see), funnel it
into some known, standard space like sRGB (good for a definition of color,
bad if you want something more than sRGB) or lastly they can provide the
widest gamut color and supply a profile (if they dont, the image looks

ugly

in Photoshop 6 since there is no description of this color).

I just saw a preview of the new Nikon D1X, and it now gives you a
choice of sRGB or Adobe RGB for images processed in the camera
(as opposed to RAW files which are processed later).

Nikon says the best input colorspace for the original D1 is
probably NTSC, because the engineers who designed it used NTSC
monitors. Has anyone here tried it?

The D1X also has a playback mode that highlights any blown-out
pixels, so you can easily see if the shot is overexposed. Some
cameras provide a histogram for this purpose, but it can be hard
to read on that little LCD.

Lee Varis writes,

As far as Adobe 1998 goes, sometimes it its worth assigning for creative

effect. If the image has a fair amount of red tones in it I find that
they tend to render over saturated and kind of orange but certain images
perk up quite nicely. The advantage of "assigning" a different profile
is that you can get a different color rendering without pushing the data
in the file. Try "assigning" wide gamut RGB to an image with very
pastel, muted tones - it's surprising how much color you can extract
from an image this way - convert to your output and fine tune.>

This underrated method will be the focus of my August column in Electronic
Publishing, which is called "Fate and the False Profile." The problem with
Wide Gamut and Adobe RGBs in this context is that while they may give you
the color you want, they both have a 2.2 gamma, which results in an overly
dark image in many cases. Solution: save several false profiles using the
same primaries, but with lower gammas. Very often nowadays we get handed
digital captures that are ridiculously dark. When this happens, their
colors are usually flat also. I find that opening them into Wide Gamut RGB
with a false gamma of 1.0 saves a lot of subsequent headaches.

I think one can get too fixated on "correct", or "true", whatever that

is. What clients really want is an image that "looks" better than
reality. Jeff Schewe often says that the reason Photoshop exists is that
"reality sucks". Using simple work space profiles gets you into a good
position from which you can "depart" from reality.>>

This wisdom continues the extremely useful content of Lee's posts in this
thread, which should be required reading.

Dan Margulis

Bob Smith wrote:

...Mike makes the case that building a profile
based on a simple RGB space definition (like a custom RGB working space) is
going to be better suited for using the camera under a broad range of
conditions than a profile based on LUTs as most input profiles are. With
the D1, I believe that what Mike has done is simply come up with a custom
working space that more closely fits the D1's usual RGB than any of the
built in spaces in Photoshop. Files are then moved from that space to a
more normal space upon opening in Photoshop. Comments?

I definitely think this is the way to go. In fact, I use something very
similar - I use a workspace that is basically ColorMatch with a gamma of
2.2, 5500k white - it's a real plain Jane monitor type of space that I
can convert out of pretty easily and it seems to work well for the
digital camera files that I use.

As far as Adobe 1998 goes, sometimes it its worth assigning for creative
effect. If the image has a fair amount of red tones in it I find that
they tend to render over saturated and kind of orange but certain images
perk up quite nicely. The advantage of "assigning" a different profile
is that you can get a different color rendering without pushing the data
in the file. Try "assigning" wide gamut RGB to an image with very
pastel, muted tones - it's surprising how much color you can extract
from an image this way - convert to your output and fine tune.

I think one can get too fixated on "correct", or "true", whatever that
is. What clients really want is an image that "looks" better than
reality. Jeff Schewe often says that the reason Photoshop exists is that
"reality sucks". Using simple work space profiles gets you into a good
position from which you can "depart" from reality.

You still might need a custom LUT based input profile if you are trying
to really accurately track colors for some scientific reason.


--
regards,

Lee Varis
varis@...
www.varis.com
888-964-0024

on 6/20/01 6:05 AM, Bob Smith at rmsmith@... wrote:

Mike makes the case that building a profile
based on a simple RGB space definition (like a custom RGB working space) is
going to be better suited for using the camera under a broad range of
conditions than a profile based on LUTs as most input profiles are. With
the D1, I believe that what Mike has done is simply come up with a custom
working space that more closely fits the D1's usual RGB than any of the
built in spaces in Photoshop. Files are then moved from that space to a
more normal space upon opening in Photoshop. Comments?

IF one can push and pull the RGB from such a camera into a definition of a
custom working space (which is a rather simple description that only needs
to specify chromaticity, white point and gamma), then I see no reason why it
couldn¹t work.

Andrew Rodney


[Non-text portions of this message have been removed]

Dan Margulis wrote:

Depends on the camera. Bob Smith has reported previously that certain
cameras seem to do better with Adobe RGB

must have been one of those other Bob Smiths... this one hasn't experienced
a digicam who's images go directly into Adobe RGB well. Most I've worked
with fit better into one of the spaces that more closely matches a typical
monitor like Apple, sRGB or ColorMatch. Maybe the confusion is that I
typically use Adobe RGB for my workflow, so I've probably stated that the
first move I make on one of these digicam files is from one of those spaces
to Adobe RGB.

I've seen a number of posts on other lists where someone is complaining
about the native color of a particular camera. Closer examination often
reveals that they are opening the files directly into Adobe RGB space
because someone told them that Adobe RGB was a good versatile space.
However, they've made no compensation to the files to get them into Adobe
RGB. That usually results in at least an over-saturated image. It looks
great for some images as it puts some added pop into an otherwise slightly
dull image so the user is unaware of the error of this process. On other
images though, the over-saturation just exaggerates what might otherwise be
a very minor color problems in the file and the user blames the camera.
Adobe RGB is a nice working space, but in the hands of someone who doesn't
fully understand how to use it, its a problem.

still on this camera profiling thread... I wonder if someone who does a lot
of camera profiling (Andrew?) could comment on a post by Mike Chaney on the
dpreveiew site.

http://www.dpreview.com/forums/read.asp?forum=1022&message=1111242

Mike is the author of one of the third party pieces of software that's often
used for processing Nikon D1 files. I believe he also sells a very low cost
(under $10?) generic profile that he's made available for a D1. I've seen a
number of D1 users speak very positively about the benefits from using it.
In the post referenced above, Mike makes the case that building a profile
based on a simple RGB space definition (like a custom RGB working space) is
going to be better suited for using the camera under a broad range of
conditions than a profile based on LUTs as most input profiles are. With
the D1, I believe that what Mike has done is simply come up with a custom
working space that more closely fits the D1's usual RGB than any of the
built in spaces in Photoshop. Files are then moved from that space to a
more normal space upon opening in Photoshop. Comments?

Bob Smith

Chris Murphy wrote:

HOWEVER, there is a missing component. I'm not hearing if there is a gray
or black balance setting. It only sounds like there is a white point
setting. Ideally one would have both a gray and a white setting to gray
balance the camera to the conditions you're currently in, and then you
have an even better chance that one profile will be helpful more often
than not.

With this particular camera, the Oly E-10, it uses on-board software to
locate and sample the brightest highlight (which may or may not be a
specular highlight). The raw files attest to this; they are very dark and
contrasty.

As for a mid-tone, it doesn't seem possible for an algorithm to locate a
"neutral" gray. Not as well as an operator who can ID the proper spot in any
given scene.

Chris Brown

Would you need seven different profiles, or does it depend more
on having a good *match* between the white balance setting and
the actual lighting conditions?

You might need seven differet profiles - actually you might find it's
just not doable at all.

If you used the "manual" setting
all the time (ie, point the camera at something white and push a
button that says "define this as white"), could you use one
profile for everything?

Manual is better because it allows you to "perfectly" adapt the camera to
the specific conditions, instead of hoping a generic setting will work.
Once white is set, then you have a better chance of one profile helping
out more often than hurting.

HOWEVER, there is a missing component. I'm not hearing if there is a gray
or black balance setting. It only sounds like there is a white point
setting. Ideally one would have both a gray and a white setting to gray
balance the camera to the conditions you're currently in, and then you
have an even better chance that one profile will be helpful more often
than not.

It seems to me it would be better to shoot a color target each
time and use that to generate a source profile on the fly
(I assume there is software that could do this without manual
intervention).

One might think so. There are problems with this method however. If you
shoot under a sunset, you are doing this for a reason. You want
everything to have the strong orange-pink cast that you get from shooting
under a sunset. If you produce a profile under this condition, the
profiling process is going to say "wow, this camera has a really nasty
orange-pink cast to it, I'm going to remove it." This conversation
doesn't really occur obviously (and it's not exactly technically accurate
either); BUT the idea is valid. Basically the process is going to remove
the sunset because it doesn't know the difference between a desired
ambient lighting effect and a device caused color cast.

Even in a non-ICC workflow, this would give you a lot of
information about the lighting conditions. Since the
highlight/shadow/neutral numbers were traditionally considered
sufficient data for color correction (with the rest being
subjective), does this imply that the manual white balance
setting is sufficient to normalize the camera's behavior?

Probably not. It's better than no setting at all, but it's possible for
the whitepoint to have a minor cast and for gray to have a nasty cast,
possibly even vice versa (although that would be peculiar).

Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932

When I open the images in PS, how would I implement/use a camera profile?

If there are profiles for the camera already you would go to
Image:Mode:Assign Profile and select profiles until you find the preview
on screen sufficient. The preview will change with each selection of
profiles, but the RGB values will remain intact. Once you find a suitable
preview, then click OK. Next go to Image:Mode:Convert to Profile and
convert the image to your preferred RGB working space.

Chris Murphy
Color Remedies (tm)
Boulder, CO
303-415-9932

Stephen writes,

I have had very limited experience with these images - but I think that I

have been able to spot a 'trend' in this new form of input. Whitepoints,
blacks and tones which one would presume to be neutral/near neutral agree
'by the numbers'. Whites are balanced 255 RGB, blacks may be balanced
20-10 RGB and neutral greys agree in RGB or LAB readouts.>>

Most of these cameras are autobalancing the highlights and shadows, which
accounts for the above.

But running the cursor over a known value such as a skintone is a totally

different story. The common trend seems to be a somewhat weak blue channel
in the quarter/midtones (for skin)...which translates as a low yellow value
in a CMYK conversion or fixed sampler reading.>>

I don't see this trend in the digicam captures I see, but certainly it
happens from time to time.

The more I deal with these digicam images, I initially conclude that

skintones are a major problem without profiles of some basic description.
I have not drawn any conclusions about other colours, but if the
white/black/neutrals are right and skin is
wrong - then it can be hard to know what is going on with colour in the
rest of file (unless other colour 'anchors' can be found to latch onto).>>

There are almost always anchors if one looks for them. In spite of the fact
that there's an enormous range of possible skintones, if the order from
darkest to lightest isn't B,G,R there's something wrong. There's an
enormous range of possible colors for a tree's leaves, but if they're more
yellow than they are green, or if they're to the cyan side of green,
there's something wrong.

I agree with the general consensus - Apple RGB, ColorMatch RGB and other

'weak' flavours of RGB often produce better skintones than sRGB or Adobe
RGB or wider spaces...when no profile is available. This is a big crap
shoot.>>

It is a big crapshoot, because the shots are taken under such a variety of
conditions that one can only guess as to what will produce the best color.
For most cameras Apple RGB or ColorMatch RGB will give the best results
most of the time, but more than every once in a while images pop up that
will open more favorably in Adobe RGB or even sRGB, let alone some custom
variant.

I am amazed that this subject does not get more traffic on this list than

it does. Are these images not the new 'revolution'?>>

Yes, but when you get right down to it it's just the same old song. Client
provides inadequate image, wants to make a work of art out of it and not
pay.

There's nothing new about having to correct inadequate fleshtones. The real
color issues with the cheap digicams that I see are a) having to cope with
images that have been "corrected" once already because the white and black
points have been neutralized; b) being confronted with ridiculously dark
images far more often than ever before.

Dan Margulis

John Denniston writes,

We have been totally digital with Kodak NC2000's for 6 years. Last year

we
purchased one Nikon D1 and have found that unlike the NC2000's 95% of the
pictures from this camera can go into the paper with little colour
correction. We have never profiled the camera.>>

Nor is there any need. Your photographers shoot in random lighting
conditions and make subjective exposure decisions. Under these
circumstances, any method of acquiring the images will work well on certain
images and less well on others. Unless you can detect some type of
consistent pattern of inadequacy when you open the images now (e.g.
generally too light, generally have green cast to the shadows), adding an
input profile is useless.

The picture on the front page was not crap and my question is why?

1. Our photogs process their pictures in ps3.01 - would this have an
effect?>>

No.

2. The RGB CMYK conversion was made in ps5.5 which is set to ignore

profiles but does use a colourmatch RGB space. Is the ideal for Jpegs?>>

Depends on the camera. Bob Smith has reported previously that certain
cameras seem to do better with Adobe RGB, and so, apparently, does the
camera described earlier this week by Kiki. The majority, however, seem to
open best in ColorMatch RGB or Apple RGB, by all reports.

3. Is your, and please don't take this unkindly, definition of "crap"

different from my mine; ie purely technical but not visual?>>

"Crap" equates to "not requiring expenditures on a custom profile."

I ask these question because more D1's are on the way, plus an upgrade to

PS6, and I don't want to spend a lot of money on the photoshop upgrade and
change a simple and quick process into a slow and complicated one.>>

In this and other areas of imaging, beware anyone offering to make things
more complex, and at the very least insist on proof that the complex method
provides better quality than the obvious one.

Dan Margulis

Andrew Rodney <andrew@...> writes:

In reality, ALL the camera provides is a grayscale file. The rawest
file produced hasnt had color created yet! You dont get that data (nor do
you want it).

Actually some cameras can give you this data, but it's in a
proprietary format, so you have to use their software to do the
transformation to RGB (which includes the adjustment for white
balance). But in at least one case there is now software
available from a third party that will do it...

And it's not really a grayscale file in the normal sense-- it
only has one channel, but each pixel is a different "color".
The color is "created" by the filter on the CCD, it's just not
in a readable format yet.

Before Kodachrome, Agfa had a color film that worked this way
(but without interpolation-- it had a shadow mask like a TV screen).