|Most of the Photoshop
printed material has, until recently, been aimed at producing
files that will be sent to a printer and reproduced as
half-tones. The recent availability of high-quality inkjet
printers requires changes to these procedures for the person
making his own prints.
The steps here are designed to help make the printed page look as close to the screen as possible and take some of the confusion out of color management and image sizing.
The first step is to obtain an ICC profile for your scanner or digital camera. You can use a commercial product such as Monaco EZ-Color to make your own or obtain one from the manufacturer. Without this information all the remaining steps will be much more difficult.
The second is to set up your monitor using the Adobe Gamma tool or a similar utility.
The steps below assume you are scanning a transparency; you will have to make some adjustments if you are starting with a digital camera image.
Original image from scanner
The first step is to set
up the scanner so that all of the image detail is captured. Place
a piece of clear film and a piece of fully exposed film in the
scanner and note the highlight and shadow settings that produce a
full-range of values. Use these settings for every slide rather
than the automatic values the software may try to use. If you
have a profile for the scanner you must use the values that were
set when the profile was created. In this example, the shadow is
set to 1 and the highlight to 191. Since we will make all
adjustments in Photoshop there is no need for expensive scanning
software. Any adjustments made by the scanning software is done
on the pixel values after they are captured by the scanner. This
is exactly the same as what is done in Photoshop or another
Scanner Density Settings
|Next set the
resolution to the maximum optical resolution that the scanner
provides (this model is 1600 dpi) and set the size to 100%. We
will make all the adjustments after the image is brought into
Photoshop. This avoids resampling by the scanning software.
Also set the bit depth to the maximum your scanner allows (this model is 12 bits/color or 36-bit image).
We are trying to capture every bit of information that we can so we will have the most flexibility when editing the image.
You can make an exception to using the maximum scanning resolution if you are certain that you are not going to enlarge to the limits. The rule for most inkjets is to aim for 300dpi in the final image so a 35mm film scanned at 3000x2000 would yield a 10x6.7 inch print.
On the other hand a 4000dpi scan of a 4x5" slide will yield an image capable of a 13x enlargement. Since a 16x20" image is only a 4x enlargement there is little value in using such high resolution in the scanner. I have found no difference between resampling in the scanner software or afterward in Photoshop.
Some users of digital cameras are satisfied with output in the 200 to 240dpi range.
Scanner Image Size Settings
Image after scanning without profile
The first step in Photoshop is assign the scanner profile. This is found under menu Image->Mode->Assign Profile. Pick the profile for your camera or scanner as established before. Notice that the colors have shifted towards the green and the most of the magenta bias is gone. The image is also somewhat darker. This is close to what the slide looks like, but we still have to take into account the effects of the printer.
Image after scanning with profile assigned
Unlike most advice on workflow, I advocate using the unsharp mask as the next step. The unsharp mask makes light tones lighter and dark tones darker, so that if we push these too far with brightness adjustments later in the process and then sharpen, we run the risk of clipping the values.
Prior workflow advice
assumed that the final output would be to a printing press and
thus the sharpening was done near the end to compensate for
losses in the half-tone process. Inkjet printers can produce much
finer detail than a printing press so we won't need to
compensate for press losses.
Quick Mask of sky
Here is the unsharp mask
dialog box. We want to choose a radius so that the physical size
is about 1/100 inch. This would be in the range of 2 to 3 pixels
(3 pixels /300dpi). Because of the bright window trim I chose a
value toward the low end so as not to produce halos around the
window frames. The threshold is set to avoid sharpening the
noise. I usually use a value of 7, but this image has a lot of
low contrast grays so I lowered it to 4 to enhance the roof
detail. I find an Amount of 150 to 170 is usually correct. The
unsharp mask will not be applied to the sky and the edge of the
Unsharp mask settings
|To show that the
unsharp mask alters the overall color range compare these two
histograms. One histogram is from before the unsharp mask and one
after. Notice the change in the height of the peaks in the
highlight and shadow region regions.
Since we don't have any values at 0 or 255 yet we will not having clipping from the unsharp mask step.
Unsharp histogram vs sharpened histogram
step is to fix any remaining brightness and color issues. I find
using the curves tool the best for almost every adjustment. Here
is the curve I applied to the image to brighten the dark and
midtone areas slightly.
Overall brightness adjustment
the image after the curve shown above was applied. In order to
maintain the dramatic look we have not brightened as much as
possible so there are practically no full white areas.
The curves effect applied
|The dark greens of the
trees are lacking in detail so we will treat these next.
Using the selective color tool we select the dark greens. Here is the selection shown as a quick mask. Notice that not all of the tree area is fully selected. We don't want the effect to be too uniform. See my other tip on how to make selections when editing 16 bit images.
Dark green quick mask
the curve for fixing the dark greens. Notice the boost from 52 to
62. Because of the selection, the upper part of the curve will
not do anything since there are no bright pixels
Dark green curve
the image with the dark green opened up a little.
Until now we have been making adjustments without regard to the output. In order to fine tune the image for the intended use we must now switch to preview mode.
Along with the scanner profile and the monitor adjustment we need a series of profiles for the printer, ink and paper combination we are planning to use. Once again it is possible to create these yourself or to get ones that are close to your print conditions. Ideally you want to have a profile made for each combination of paper and ink you plan to use on your model printer.
Image with boosted greens
preview mode: View->Proof setup->specific paper. You can
toggle between preview on and off with Control-Y. This paper
preview shows that the image will be printed lighter and with
more contrast that the screen image indicates.
We will continue all the following steps with the preview turned on. We are "targeting" the image for our specific output device.
Same image with print preview on
|As shown in one of my
other tips, fixing "leaning building syndrome" is one
of the ways to make architectural pictures more professional
looking. Here is the image while using the perspective tool. We
have added some guide lines to make it easier to check for
parallel vertical lines.
Before this step we converted from 16 bit to 8 bit mode so that we can use this tool. From this point on we will try to limit the amount of image correction that is done so as to avoid posterization of the image caused by quantization errors when adjusting RGB values.
|Remember we have done
all prior editing in 16 bit/color mode. Notice that the file size
is now 43.5M and the image size is still as it was when we
scanned it in at 100%. When the image was in 16 bit mode the file
size was 87M.
Original image size
|To aid in
cropping/sizing we change the image resolution to 300dpi without
resampling image. This changes the document size values without
altering any of the image data. We are trying to avoid lossy
changes as much as possible. The image will now print at about
11x14 inches. This new size will allow us to visualize the final
size when cropping the image.
Resized image size
|While cropping we can
look at the info palette to keep an eye on the image size. The W
and H indicators show the cropped size in the current
Info palette during crop
|We are removing most
of the street which we had to include since we photographed the
church from across the street. The nominal image is now about
Crop preview during crop
|Now that we are
viewing the image with the printer preview on we find that the
contrast is a little too low and the blacks and whites are not as
intense as they might be. We add a curves layer which spreads the
values out to compensate for the printer
|Another change we wish
to make is to remove some of the excess blue in the shadow side
of the church. While shadows are supposed to be bluish, this
film exaggerates the
effect. We made a selection of the shadows, shown here in quick
mask mode. We then added another curves layer to lower the value
of the blues. See my other tip on how to do this in
Shadows quick mask
|Here is the layers
palette showing the overall curve and the shadows curves. In
principal we could have made these changes in 16 bit mode using
the selection tricks mentioned in my other tip, but by using
layers we can alter the amount of the effect and see how the
multiple layers interact. In addition, if we need to use this
image for another purpose later, we can turn these layers off and
target the image for the new purpose.
Final layers palette
|Here is the final
image just prior to printing. We now make one more adjustment to
image size to set it for printing.
|We want to print this
image on 8½x11 inch paper. So we resize the image to a
width of 8 inches. We don't resample so the resolution goes
up. This way we have not thrown away any data. We are sending too
much data to the printer, but it will be resampled down to the
printer's preferred resolution. This slows down the printing
slightly, but we have limited our resizing to just a single step
from scan to print.
Image size for printing
|Here is the print
dialog showing the print preview. Note that Photoshop does not
color manage this dialog so don't pay any attention to how
the dark the image appears, just its placement.
The important issue is to not use any color management in the printer. Since we have edited the file with the print preview on we don't want any adjustments added during the printing step.
Also note that the document is still tagged with the scanner profile, so we have avoided any data loss by converting from one profile space to another. If you think you will need to use this file again make sure to save it with the profile attached.
|To summarize the main
1. Scan at max resolution and 100%.
2. Capture all data at highest bit depth.
3. Apply profile for scanner or camera.
4. Sharpen select areas.
5. Adjust brightness, contrast and density.
6. Fix problem areas or colors.
7. Fix perspective, if necessary
8. Crop as needed
9. Add layers to enhance image while viewing in preview mode.
10. Resize to print size without resampling.
11. Print without printer color management.
|For reference the two
images are shown again, below.
Most of the changes on this image are relatively subtle.
It is the small changes that change a snapshot into a picture
that reflects your artistic intentions.
Nothing was done except to compensate for the failures of
the recording and reproducing media.
You are, of course, free to make deliberate distortions if that
is your goal.
For the frugal all changes were made without purchasing any
extra scanner software or fancy plugins.
© 2003 Robert D Feinman