How to get rid of speckles from matte finished photos?
Jeanie Sumrall-Ajero
I have a project where I'm needing to get good digital versions of
25-year-old photos that were processed with a matte finish. So, in addition
to having to fix faded, yellow colors, I'm having a major problem with the
consistent grid of white dots all over the photos. The negatives don't
exist anymore, so I'm stuck having to work from the prints. My scanner (HP
6300) seems to pick up the matte finish as white dots. (The matte finish
from 25 years ago is decidedly different than the matte finish of today. It
looks like a uniform fine mesh imprinted into the photo. I don't know how
else to describe it.)
Anyway, I think what is happening is that the scanner lamp is reflecting off
of the edges of the "mesh" and creating a uniform grid of white dots over
the entire photograph. I've tried scanning with the "no sharpening" option,
but that doesn't seem to make any difference.
My current solution is to use the Gaussian blur filter over the whole
photograph (working on the lightness channel in Lab mode). At a radius
around 2.5, the dots are sufficiently "removed", but of course the entire
picture is now blurry. Then, I use the Unsharp filter to sharpen the entire
photo (again working on the lightness channel). At a radius of 2.0-2.2,
sharpening of about 180% and threshold 0, I'm able to bring the photo back
to the sharpness of the original photo with the grid of white dots greatly
faded. However, the original scan is slightly blurry, so I'd really like to
be able to sharpen from the original photo - not just get back to that
point. Unfortunately, sharpening from the original just makes the grid of
white dots more pronounced.
Has anyone else run into this problem before? Is there a better way to
handle this, or am I stuck b/c of bad input to start with?
Thanks for any tips. If this has been discussed before and there's info in
the archives, please give me some keywords to search on. I suppose there's
a formal term for what I'm dealing with, but I'm fairly new to this whole
game. Just learning as I go...
Thanks, Jeanie
Jeanie Sumrall-Ajero
archives. First thing to try is increasing the threshold in the unsharp
mask. However, if anyone has any great ideas, I'm still open to
suggestions.
Thanks, Jeanie
Gregor Munro
"Jeanie Sumrall-Ajero" <lotus...@home.com> wrote in message
news:LLcu6.467975$ge4.16...@news2.rdc2.tx.home.com...
"If you are scanning photographs on paper that has a distinct surface
texture, then you are seeing the texture, and reflections of the scanner
lamp from the irregularities of that surface. "
and
"If your SLR camera has a macro lens for closeups, the best solution is to
photograph the original print, and then scan the copy. This allows you to
control the direction and angle and diffusion of the lighting to reduce the
reflections. "
This Site is an excellent source of info on scanning !
--
Gregor Munro
Miramichi, N.B.
John Boy
good point is that it hides fingerprints. I've heard it suggested to coat
the scanner bed with some kind of oil prior to scanning, to 'fill in' the
pattern, but this would be pretty messy. If the prints are all on the same
kind/brand of paper, they would probably have the same size pattern. If you
scanned all of them at the same resolution, you could make a similar pattern
in Photoshop, and fill it into some sort of adjustment layer, or an Alpha
channel, which you could then load as a selection, to help work on the
image. This would take a lot of experimentation, but once you have it you
could save it as an action.
- John
Jane Krate Duda
more than is desirable. I have had success masking out faces and other items
I don't want to lose detail in, then running D&S.
Jane
Peter Wollenberg
...
>
>My current solution is to use the Gaussian blur filter over the whole
>photograph (working on the lightness channel in Lab mode). At a radius
>around 2.5, the dots are sufficiently "removed", but of course the entire
>picture is now blurry.
...
The median filter might be superior for your purpose to begin with.
After either filter - median or Gaussian - set the blending mode to
"darken" in the fade dialogue and decrease the opacity of the effect
as much as possible.
See a comparison of the two approaches at
http://www.med-rz.uni-sb.de/med_fak/pharma-toxi/smooth.html
A loss in the highlights is inevitable (as seen at the nose tip in my
sample image), but by sparing the shadows, degradation of image
quality is minimized.If you have a lot of time to spend, you might do
the filtering on a layer copy and recover relevant highlight parts of
the original by selectively masking the top layer away as required.
HTH, Peter
Mike Russell
There are a few ways you may get rid of the silk finish on your photographs.
One way is to play with the scan angle. With silk, which consists of
clothlike vertical and horizontal lines, perhaps scanning with the print
rotated 45 degrees will reduce or eliminate the finish.
Another trick is to scan the images twice, rotated 180 degrees (or perhaps
in this case 45 and 225 degrees), rotate the images to a matching
orientation in Photoshop, and combine one image over the other as a layer in
darken mode. This sounds simple, but the devil is in the alignment, and you
may find yourself spending many minutes on each image. For really accurate
lineup, you may need to scan at a higher resolution, and pick an area of the
scanner bed that minimizes dimensional distortion - and there will be some
even on a flatbed. Accurate rotation will probably be easier if you cut a
card accurately, and tape each image to it for scanning.
Finally, you may eliminate the texture completely if you photograph the
images, using a Polaroid filter and Polaroid gels on the light sources. If
you have a digital camera this will have the added benefit of being even
faster than scanning.
I don't really favor using only software filtering to get rid of the
pattern, but it might be minimized by using curves or levels to create a
selection mask from the specular pattern reflections, and using that as a
layer mask on a gaussian blurred version of the original. Add a bit of
noise afterward until the pattern becomes invisible.
"Jeanie Sumrall-Ajero" <lotus...@home.com> wrote in message
news:nN7u6.467187$ge4.16...@news2.rdc2.tx.home.com...
Neil Jones
"John Boy" <jrn...@mcs.net> wrote:
>pattern, but this would be pretty messy. If the prints are all on the same
>kind/brand of paper, they would probably have the same size pattern. If you
>scanned all of them at the same resolution, you could make a similar pattern
>in Photoshop, and fill it into some sort of adjustment layer, or an Alpha
>channel, which you could then load as a selection, to help work on the
>image. This would take a lot of experimentation, but once you have it you
>could save it as an action.
>
>- John
I don't there are enough minutes left before the Universe ends to get this
approach to work. I'm sure chaos theory holds the key...
Neil
LA
Mike Russell wrote:
> Jeanie,
>
> There are a few ways you may get rid of the silk finish on your photographs.
>
> One way is to play with the scan angle. With silk, which consists of
> clothlike vertical and horizontal lines, perhaps scanning with the print
> rotated 45 degrees will reduce or eliminate the finish.
>
> Another trick is to scan the images twice, rotated 180 degrees (or perhaps
> in this case 45 and 225 degrees), rotate the images to a matching
> orientation in Photoshop, and combine one image over the other as a layer in
> darken mode. This sounds simple, but the devil is in the alignment, and you
> may find yourself spending many minutes on each image. For really accurate
> lineup, you may need to scan at a higher resolution, and pick an area of the
> scanner bed that minimizes dimensional distortion - and there will be some
> even on a flatbed. Accurate rotation will probably be easier if you cut a
> card accurately, and tape each image to it for scanning.
>
> Finally, you may eliminate the texture completely if you photograph the
> images, using a Polaroid filter and Polaroid gels on the light sources. If
> you have a digital camera this will have the added benefit of being even
> faster than scanning.
>
ITYM Polarizing Filters for both lens and lights. A completely different animal
than Polaroid... and it will work.
Mike Russell
...
> ITYM Polarizing Filters for both lens and lights. A completely different animal
> than Polaroid... and it will work.
Not so different - although the patents are up, Dr. Land invented Polaroid filters and started his
first company to manufacture them:
http://web.mit.edu/invent/www/inventorsI-Q/land.html
LA
Mike Russell wrote:
Whatever, I just meant by what they do! Polarizing light has nothing to do with one step processing. I
was just pointing it out so that the question asker would be shopping for the correct thing. Thanks
for the information though. I didn't know that.
LA
