Notes on the Cambodia Portfolio
The original note describes how the Cambodia Portfolio was created.
However, there are many changes to this process that I would make if
doing this task today. Here are a few.
* PhotoShop and the CanoScan FS2710 work with Win 2000P with the
addition of the 'aspi32.exe' driver from Adaptec. We no longer have
to scan using Win NT 4.0 and LE 5.0.
* Scans should be done at the maximum bit depth (12 bits/color with
this scanner). Many operations can be done in Photoshop using the 16
bit mode. One should maintain 16 bit mode in Photoshop as long as
possible. A color scan produces a 59 MB file, 19 MB for
monochrome. Luckily, CDs are cheap enough for this not to matter.
* The images were copied to CD-ROM, but we need to rescan them at the
greater depth. Saving to CD-ROM is good. Saving poor quality scans to
CD-ROM is a waste of time. Knowing at which point in the process to
save to CD requires some consideration.
* I have since added another 256 MB of RAM memory which speeds up the
scans providing the size of the Photoshop cache is increased
Original Note 7/6/01
This is the first effort to produce a portfolio using old pictures. It
is not optimum, but provides a starting point for further
The original pictures were Kodachrome slides, taken with a Canon EOS
and a Canon zoom lens EF 35-105 mm, f3.5-4.5. It is a very modest
camera that does not produce very sharp images, but it made a good
The process to make the portfolio images:
1. Scan the Kodachrome slides with the CanoScan FS2710 using Win NT
4.0 and the twain driver through Photoshop LE 5.0. The images were
captured at the maximum resolution of 2720 dpi. Each image is about 29
MB in size. Scanning images is a laborious time intensive problem made
particularly painful because my old PC only has 256 MB of memory and a
450 MHz processor. Adding more memory would help more than a faster
processor. The scanning operation spends most of its time swapping
memory to disk. Changing from LE to the full Photoshop may also help
because the scratch disk and memory could be optimized.
2. Copy the images to a CD-ROM in hopes that we won't have to re-scan
them any time soon.
3. Change to the Win 2000 disk which has Photoshop. We've set up the
PC to use interchangeable primary hard disks, each with its own
operating system. Hard disks and the replaceable trays are cheap and
we don't have to fight with multiple OS on one disk. I copied the
images from the CD to the hard disk. The CD remains as an archive of
the original images.
1. Process the images in Photoshop.
Each of the images had the same general processing flow in
Photoshop. The goal was to make images with a richer color than
straight black and white - kind a of a sepia tone. We are using
Photoshop 6.0 for these results. See the excellent
article by Michael
H. Reichmann on the technique used.
1. It is good to have a clean image, free of dust and lint. Blowing
the negative or slide off with a
Besler DustGun 100
before scanning helps. In Photoshop you can use the 'clone stamp
tool' set at the fuzzy 21 pixel setting and scanning the image at
2-300% magnification seems to work well. This is tedious to do, so it
is well to save the image at this point without any other processing
so you don't have to do it again.
1. Crop the image as required. How can cropping be done to standard
sizes? These have aspect ratios that are determined by what I wand to
eliminate from the edges and will seldom be a 4:5 or 5:7 final aspect.
It doesn't help that the original images are 2:3. Maybe we used to be
slaves to standard print sizes to the detriment of good
image. However, I do not have any capability to make custom mattes.
So 8x10 or 11x14 prints are nice sizes because they fit standard sized
1. Image->Adjust->Auto Level
Often I'll try the Auto Level control as a first check toward
optimizing the tonal range. It does not always work best, so be
prepared to back out of it. Manually, check the level (Ctrl-L
control) of the image. Not worried about color balance here but want
the image to have full tonal variation. Just the RGB control was
used. First the two end pointers are adjusted as needed to eliminate
any flat area at each end of the scale. Then the center control was
moved to achieve the nicest overall image contrast and balance.
1. Image->Adjust->Channel Mixer
Use the Chanel Mixer control to convert to monochrome.
Most of the images were converted using 50% red, 0% green and 50% blue.
Convert to gray scale. This eliminates the color information in the
images and reduces the size considerably.
1. Now look at the image to see what additional processing is
needed. I found that the original images are too contrasty. Faces in
the shadows needed to be lightened.
Add the quadtone. I reproduced Michael Reichmann's suggestion for the
pictures shown in the Cambodian portfolio.
The four colors and curves in the quadtone were:
PANTONE Process Black CVP
PANTONE 732 CVP
PANTONE 727 CVP
PANTONE 649 CVP
2. Image->Mode->RGB Color
Convert to RGB. This is to allow the images to be saved under
JPEG compression for the WEB.
3. Image->Image Size
Change image scale to 72 dpi and a maximum width of 650 pixels or
height of 500 pixels. My Porsche 912 site uses smaller images 550 max width and 400 max height
in an effort to speed the loading of images. This site should not see
much, if any, traffic so the download times are not much of an issue.
1. Use the unsharp mask at settings 100%, 1.0 width and 2.0.
1. Save the images using ImageReady at a quality factor of 30.
1. Make a 130 pixel wide thumbnail image to be used in a navigation
It is important to calibrate your monitor. This seems to be browser
dependent, so if you use more than one browser as I do, check each
for color balance.
My original images have a green tone on his monitor which was not the
intention. (He has calibrated his monitor). This was apparent with
another browser (Mozilla) on my system compared to the normal one
Last modified: Mon, 15-Oct-01 08:14:42 PDT