Newbie Scanning 35mm slides with CS4200F

  • Thread starter Thread starter Den
  • Start date Start date
D

Den

Just bought a Canon CS4200F and I started scanning my old 35mm slides.
The purpose is to have an archive, make corrections, etc.

I want to make CD or DVD and look at them on TV and maybe make a few
print not larger then 8x10

Right now I'm scanning at 3200 resolution and I'm saving as Jpeg with
a high quality around 9-10 which give me a file around 2meg. At
maximum quality the file is 35meg.

Are those settings Ok for my needs?
Any suggestions?
Did I forget something?

Den
 
2 MB seems too low for decent 10x8 prints. Better use the entire scanner
capacity so you can have better pics. I scan at 4000 dpi (about 50-55 MB
files) and can produce very decent A3+ prints. A4 prints (which is close to
10x8) can be made at half these figures but 2 MB doesn't seem adequate for
what you want.
Vasilis
 
Thanks for your comments Vasilis

It means that I've got some learning to do before going any further
with my project. The problem is I'm trying to get a small file size
and still have a good to excellent picture quality.

Thanks for reminding me that sometime small is just to small :-)

Den
 
Just bought a Canon CS4200F and I started scanning my old 35mm slides.
The purpose is to have an archive, make corrections, etc.

I want to make CD or DVD and look at them on TV and maybe make a few
print not larger then 8x10

Right now I'm scanning at 3200 resolution and I'm saving as Jpeg with
a high quality around 9-10 which give me a file around 2meg. At
maximum quality the file is 35meg.
Click to expand...


You said "Jpeg with a high quality around 9-10".

Scanning 35 mm film at 3200 dpi is about 38 MB, uncompressed. If that
9-10 JPG setting is on a 1-10 scale, it doesnt seem like 9-10 maximum
could compress 38 MB to 2 MB. Seems it should be more like 20 MB (a
wild guess, but 1/2 size might be in the ballpark for maximum setting).
Or if on a 1-100 scale, 9 couldnt be called high quality by any
imagination. So it isnt very clear what the situation might be?

Regardless, the file size after JPG compression is simply a quality
matter. Higher quality is less compression, and more compression is
lower quality, referring to visual image quality. Compression does not
affect the image size in pixels, it affects only quality and the file
size in bytes. Image size in pixels is not changed. You always get
back the same pixel dimensions out of the JPG file (and it will be 38
MB again in memory after it is opened and uncompressed), but you dont
get back the same quality from JPG that you originally had.

If you really meant 2 MB, then try an experiment to understand quality.
Scan a new image, fresh and perfect, unaltered. First save that new
image as a TIF file. Then, second, also save that same image as a JPG
file with these same settings giving 2 MB. Then close the image
totally. Then open both files, to be certain that you are seeing the
actual contents of the files. Then zoom both to about 400%. You will
have to scroll around, but examine the same area of both images side by
side, several same areas. All that garbage you see in the JPG that is
not in the TIF is JPG artifacts caused by JPG compresson. I am assuming
a less than maximum quality setting was used.

There are usually at least two types of artifacts - one type is sort of
dark smudges around all sharp edges of detail. Hard to describe, but you
will know it when you see it now, as compared to the TIF which is still
perfect. Bad cases make the image be very "unclear". One type is
larger 8x8 pixel blocks in the empty or blank areas, like skies and
walls. The more the JPG compression (the lower the quality setting),
the worst it is.

After you first learn to see these JPG artifacts at 400%, then you will
have learned to recognize them at 100%, and the point is to understand
the down side of excessive JPG compression. This experience will help
you decide if this small file size suits your goal or not.

If you must use JPG, and sometimes we must, then for purposes not
involving modems, I think you'd appreciate the higher image quality of a
JPG that was more like 1/4 size than this 1/18 size. 1/18 is a lot of
compression and it will affect quality. 1/4 size is phenomenal in
itself. However this actual compressed size does depend on the image
content, images with much empty plain space do compress a lot more than
images totally full of fine detail everywhere. There is not one
numerical answer. You might see a 2:1 range at extremes.

But instead of the compressed file size, it is instead the image size in
pixels that affects how we would use the file for a specific purpose.
Generally the goal is that we want enough pixels to print at about 300
dpi. So if we want to print 10 inches at 300 dpi, then that image
dimension should be around 3000 pixels. It need not be exact, a 3200 dpi
scan of 35 mm film should be fine. If we only want to print at 6
inches, then we only need around 1800 pixels, maybe a 1600 dpi scan. We
can archive the larger image suitable for our largest (realistic)
purpose, and then we can resample a smaller copy of it for a smaller
purpose.

Digital images are dimensioned in pixels. It is the dimension in pixels
that counts, that is meaningful when contemplating image size.

The JPG compression file size only affects if we can still bear to look
at it. :)
 
Fact of Life No. 4279:
Photographic Image files are large. There is *nothing* you can do about it.
 
Hi Wayne,

Very helpful comments that help clear part of my very cloudy sky;

The key phrase was about the JPG compression; now I understand the
problem and since my first goal is to preserve those slides I should
stick to a TIFF file. No need to experiment you were "sadly" :-) very
convincing.

Now, for many reasons, we need a smaller file size and this is
related, as I understand, to what I want to do later with those
digital files.

Showing on TV is one part, and printing *some* file on a 8x10 format
(maybe bigger) is an other.

From your comments & on your website you write: "But note that
lowering scan resolution to reasonable values for the purpose is often
the best file size improvement you can make"

So I decided on the target of 300dpi on printing and a 5"x7" size for
most of the slides. I will scan at the calculated resolution; I figure
a scan resolution of 1700 that would result in a 10meg files.

I'm still fighting with figures ;-) dpi, pixels, screen, print, in
scanning even mixing that with figures related to my Canon A70.
(megapixels)

Lots of reading to do on your web site.

Thanks your help Wayne

Den
 
Just a correction to my previous post :-((

I intend to scan most all the slides at 2800 dpi minimum and keep the
TIFF file as is. In any circontances the file size will be at least 20
to 30 meg.

Den
 
Saving as TIF with LZW compression will cut the file size by a large
percentage, yet it will retain all of the image data---TIF/LZW is lossless.
 
So I decided on the target of 300dpi on printing and a 5"x7" size for
most of the slides. I will scan at the calculated resolution; I figure
a scan resolution of 1700 that would result in a 10meg files.

I'm still fighting with figures ;-) dpi, pixels, screen, print, in
scanning even mixing that with figures related to my Canon A70.
(megapixels)
Click to expand...


The image from the camera is exactly the same thing, in that its image is
also dimensioned in pixels. Your camera menu likely has choices of
creating a few different image sizes (pixels), like full size, 1/2 size, 1/4
size, etc. The scanner also creates images dimensioned in pixels, and its
scan resolution (and size of scanned original) is the menu that determines
the image size. Scan 3 inches at 300 dpi, and you will create 900 pixels.
But after you have either one of these images dimensioned in pixels, then
it's the same thing, regarding use of that image.

Printing the 5 inch width (in the 5x7 inch goal) at say 300 dpi on paper
implies the need for 1500 pixels in that dimension (in whatever way these
might be created). Creating 1500 pixels by scanning say 0.9 inches of 35 mm
film width implies 1500/0.9 = 1667 dpi scanning resolution. Which is
precisely correct (ignoring any cropping issues). This method does of course
work, and it will likely appear as a fine result, good enough.

Specifically, this is exactly what your scanner software does where you
specify to scan at 300 dpi at say 555% scale... it scans at 300x5.55 = 1667
dpi to create sufficient pixels, and outputs this image scaled to print 300
dpi 555% size (the 5x7 inches), which is what you asked it to do. This
little calculator is very convenient.

However there are other considerations too, which some people like to worry
about. One of those is the belief that the scanner can do best when
scanning at integer divisions of its maximum optical rating (in the same way
the camera offers its size choics). For a 3200 dpi scanner, that would be
3200 dpi, 1600 dpi, 1067 dpi, 800 dpi, etc. The scanner only has the 3200
dpi sensor, and so must resample this 3200 dpi horizontal scan line to
create any smaller value, and the integer divisions are an easier resampling
job, which comes out best. Some scanners offer ONLY those integer division
choices for this reason.

So if you scanned this case at 1600 dpi (3200/2), then print the five inches
at (1600 dpi x 0.9 inches) = 1440 pixels, divided by 5 inches is 288 dpi
printed on paper at 5 inch size. And 288 dpi is plenty, it will be fine.
We only need to get near the ballpark of 300 dpi, there is no significance
of 300.000 dpi. We might debate if we can even detect the difference in
plus or minus 30%, but it really wont be any big deal. Normally when you get
to half, it may become apparent.

That often works plenty well enough, and it is often convenient. However
this method sometimes leaves some uncomfortable situations when we cannot
get near enough to the 300 dpi goal (however we may decide it). For example,
scanning at the next higher integer division might come out too high above
300 dpi for our case.

Choices then are still a few:

1. ignore it, and print the larger image anyway, at say 462 dpi. You will
hear a few say their inkjet does best at these large values, but I have
never been able to detect that myself. I cant even imagine it. So I aim for
300 dpi as a reasonable target value for photo quality. It is a comfortable
value (I'm speaking of color, not line art). Anyway, the number is your
choice, whatever you believe that you can see.

2. Scan large and resample the image smaller (to appropriate size, the 1500
pixels) in a photo editor. The photo editor is generally better at this
resample than the scanner (more resources, cpu, memory, software algorithm,
time to do it, etc). A bit of USM sharpening works great after this
resample.

3. Revert back to the first way, scanning at value like 1667 dpi anyway.

It is pretty much personal opinion at this point.
There are two main schools of thought,
A) if I can't see it anyway, pick the convenient way, and
B) I could care less if I can never see it, I want to fight to maximize
every little detail in the most tedious way.

Fortunately, I am very unbiased in this :) Actually, I do feel I work
towards a middle ground about it, and do go for those results that I can
actually see. I probably do 2, and sometimes 3, but very rarely 1.
 
Back
Top