Now came the step of combining the scans and removing
defects. Karl's playfield has an overlay over the middle star
base medallion that is slightly different from the original.
There was also a fair amount of fine alligator cracks in the finish as
can be seen from the test scans above. These will be removed
during this step of the project.
The scans arrived in ten files, and I started to combine two of
them. The resolution was 600 dpi, and I quickly found this
resolution to be impractical. The Photoshop file grew to 344
MB,
and I
had only combined two of the ten segments! Also, rasterising
and
saving the image took several minutes. To overcome this, I
cut
the image resolution down to 300 dpi. Merging two segments
resulted in a Photoshop file size of 93 MB, a more manageable size.

The merged set of ten scans from Karl. Note that the center
medallion is a decal, and is why it looks brighter.
The combined file with
all ten scans
in separate layers was a size of ~450MB. I then used a 36 inch wide
inkjet to print a full scale test print (Test Print 1). This
test
print was a
special version that was converted to monochrome and then run through
Photoshop's edge detection algorithm. This allowed the
features
and lines to stand out. I then backlit the playfield to check
how
the inserts shine through the test print. This is an
important
test to check the location of the inserts on the merged file.
It
took a few iterations and tweaks, but I was able to line up everything
very precisely. Once the file was flattened to a single
layer, it
had a size of 180MB.

Test Print 1 overlaid onto the junk playfield. This playfield
will be the first target of the overlay.
I started touching up
the defects,
and quickly realized that the only good way to handle this was to
completely redraw the artwork. This completely eliminates all
defects as well as color and pixel noise from the scanning
process. One example is shown below. Although it is
possible to redraw in vector graphics, I think it would be very tedious
to do it for these example areas with intricate lines and natural
features. It took one full day to redraw the upper fourth
part of
the playfield (above the Hubble and the 'T' target). I only
used
the following ten colors: Light Blue, Blue, Orange, Red, Yellow, Light
Peach (playfield background), Light Gray, Dark Gray, Black,
White. These discrete colors will be easy to replace in one
operation with Photoshop tools.

Example of the redrawing of the playfield in progress.
Redrawing the playfield
in Photoshop
took almost two weeks of of intermittent work. Work
progressed
steadily until near the end, when I had to figure out a way to
reproduce the gradient dots that are on the playfield. These
occur in two locations. The first is near the in and out
lanes. The color there changes from red to yellow in a
gradual
fashion using a mesh of dots. I could easily have replaced
this
with a uniform gradient, but it is of course not entirely
accurate. After several days of experimentation and learning
how
to use Photoshop, I hit on a way to do it convincingly (see
below). The second area where gradient dots are used is in
the
background, where the color changes from black (space) to blue (sky).

The original scan of the playfield along the left outlane.
Note
the gradual change from yellow to red using coarse diamond shaped dots.
Reproducing this proved more difficult than I thought, and it took
several days to develop the technique.

The redrawn playfield using the gradient dots as a back
layer.
This allows me to replace it later if needed.
Note that the blue is solid. The grainy texture is due to JPG
compression in this excerpt.
The completed artwork. Every pixel you see here has been
redrawn.
Note that the holes in the playfield have wood texture. In
the
finished
overlay, they can be cut out, or left in place to cover the wood.
I chose to setup four
layers in the
main Photoshop file. The bottom layer is the calibrated
merged
photographic scan. The next layer I called 'sky' and is the
black
to blue gradient using the dot pattern. Putting it on a
separate
layer allows the background to be replaced in the future. The
next layer is the main artwork, consisting of the upper playfield and
the material on the extreme left and right. Finally, the
center
of the playfield is its own layer. Small sections were
cropped,
and then redrawn on their own. Each of these subfiles has
many
more layers, which are then merged for one paste into the main
file. These smaller files allow easier handling by the
computer,
and leads to faster and safer saves. I can also in the future
open them again if I need to shift things around, and replace them in
the main file. At the conclusion of the redraw, the main file
was
about 300MB (~62 Mega Pixels). The individual sub-files were
about 20MB each.
Test
Print
2
I ordered a full-scale test print (TP2) on the same
translucent
backlit material that I used for the
translites.
The colors were a little muted as a result of the medium, but still
very vibrant. The intention of using a translucent material
was
to be able to most easily check the registration of the
artwork.
Overall, the colors were the right hue, except the background blue sky
was too dark (lower part of playfield background), and the dark gray
was also too dark.

Test Print 2 next to the junk playfield. The dark colors are
a
bit muted
because of the choice of translucent medium.
I made a
simple light table
by propping up the wooden playfield plank up on
soup cans, and then
inserting a large fluorescent shop light underneath.
As you can see, the
inserts shine brightly through, and I can
check the registration much more accurately than with
Test
Print
1.
With the above setup, I
could much
more accurately check the registration. The errors proved
interesting. Overall, the right hand side of the playfield
had to
be shifted up, and the left side had to be shifted down. It
was a
sort of 'racking', or parallelogram-shaped correction. The
highest error was 3 mm, but most errors were in the 1-2 mm
range.
Interestingly, there were many inserts that did not need
shifting. Possible explanations are a slight distortion of
the
original scan, or differences between Karl's playfield and
mine.
After a few days work, the artwork file was adjusted and shifted to
line up with the playfield.

April 2011 update. This test print now hangs in the window of
my
NASA office and functions as stained glass.
Afternoon sun lights the print up very nicely.
For this test, I decided
to print on
high gloss photo paper for the best color rendering.

Test Print 3 turned out real beautiful. The colors were very
vibrant and
matched quite closely. Only the blue background on the
lower playfield needed to be tweaked.
Although not really
visible in the
above image, the colors turned out very vibrant and beautiful this time
around due to the choice of printing medium. To my relief,
the
shifts to the artwork made from Test Print 2 were verified, and all the
inserts matched to well within one half millimeter. The light
and
dark grey (adjusted for this print) also matched. However, I
thought the blue background turned out too red. Suspecting
this,
included a color wheel along with the artwork for Test Print 3 so that
I could adjust and calibrate the printer. At this point, the
Photoshop file is 280 MB. The print file is 18 MB.

Application of 476MP adhesive to the back of the print to make it
self-adhesive.
E-mail from a recipient of Print 3:
Hi Ed,
sorry for the
delay. I got the poster, many thanks. You have done
some
excellent work there.
Thanks
again Mick
|
Test
Print 4
Now that I was confident that the artwork file would line up with the
playfield, it was time to take the next step, and to print onto
vinyl. One problem with playfield overlays is that they have
areas that are transparent, yet have color and black and white in other
areas. After lots of searching and phone calls, I found a
printer
that was able to prepare an overlay to
do that. The print proof is shown below, with the parts in
pink
signifying the clear (transparent) areas.
As one can see, there are certain challenging areas. For
example,
at the top of the playfield, the letters "U-S-A" are in white
color. They float in the middle of an insert. The
next
example is in the center of the playfield, where there are six circular
inserts with the white words "Space Shuttle" crossing ink/clear
boundaries. The printer needs to be able to print white ink
on
clear, or use a multi-layer process, where white vinyl is first used,
and then the clear areas cut out and removed. Afterwards, a protective
clear layer is applied. In the case
of Test Print 4, the latter process will be used.

A major step forward is printing on vinyl. The parts in pink
will
be clear to allow the inserts below to show.
The file with the color
artwork was
imported into Adobe Illustrator, and the lines to be cut were then
drawn in that software. The finished file can be sent to the
printer.
The advantage of using vinyl are two fold:
- It allows the use of a liquid to 'float' the overlay while
it
is drying so that it can be precisely positioned to align with the
inserts. See a link below for such a product. If
paper were
used for the overlay, it would probably absorb this liquid, and swell
and distort.
- The clear parts would not have a 'ridge' that could affect
ball
movement. In addition, the clear areas will allow the inserts
to
be visible with their full brightness.
The alternative to the latter point is to have the inserts be covered
by colored vinyl, but this would darken them in my opinion.
Click
here for photo of an
Addams Family
overlay, also printed on white adhesive vinyl. However, you
will
note that the inserts are not clear. The producer of this
overlay
is most likely
"RnR"
(defunct
company).
Here is
another overlay for
High Speed
that is of inferior construction. It was on Ebay in August
2009,
and sold for about $190.

Print 4 on adhesive vinyl with clear and color parts.

Close-up of the center of Print 4. Compare with
this view
of my restored
playfield.
This test print had some
small
defects: 1) some lettering was on the top of the clear vinyl, which
means it would not be protected, 2) the size was 0.8% too large
(about 5/16"). This means that some inserts on the ends of
the
playfield would not line up, and 3) the blue sky background turned out
a bit purple in this printing. A subsequent test print will
address
these problems.
Test
Print V
Test Print V corrected most of the issues above. The size was
perfectly
aligned with the test playfield, and the insert text is applied
separately. An important benefit of that is that the user no
longer needs to sand the entire playfield to bare wood. If
the
inserts on the subject playfield are reasonable, they can be preserved,
and after proper sanding to smoothen out the old paint, the overlay can
be applied. The user can also apply the insert text only on
the
inserts that need them.

Test Print V. The insert text is now separate.
Test Prints 6A and 6B
were done in
miniature to try and get the blue sky color right.
I was still not
satisfied with the
blue sky background, and ordered two
small test prints to check the color accuracy. This was not
fully
successful as can be seen in the photo above. The one on the
left
was
too dark and purple, while the one on the right was too
light.
The
printer then had the good idea of sending me a sample of blue colors on
the actual white vinyl with the RGB values written into the
samples.
This would allow me to nail the right value to use.

Color sample with the blue Pantone colors and their RGB values printed
with the same printer on the same substrate. This is much
more
accurate
than the color wheels I was using in the corners of the artwork.
This print is the
culmination of a
lot of effort to get things right.

Betcha can't tell which one is the real playfield 8-).
I no longer have a playfield to apply the overlay, but
below are
several users of the overlay that sent me their photographs.