Scanner res and enlargements

  • Thread starter Thread starter Barn owl
  • Start date Start date
B

Barn owl

Could anyone tell me the max size enlargements advisable from a scanner with
a 2400 x 4800 dpi scan resolution (Would 10" x 8" be advisable). Scanning
35mm film negatives and slides

Also same question regarding a scanner with a res of 3200 x 6400

I have two scanners in mind the Epson RX700 with a res of 3200 x 6400 or the
Epson RX620 with a res of 2400 x 4800.

I suppose a stand alone scanner would be better but for practical and space
considerations it will have to be an all in one scanner...

I would be most unlikely to want to enlarge beyond A4 size. However I would
probably want to do some cropping of 35mm based images.

Bottom line is would the Epson RX620 be sufficient for my needs...

Also please how do I work out the relationship between resolutions given in
dpi and those resolutions stated in pixels. (is that pixels per inch??? I
don't understand!!) Thanks......
 
...
Could anyone tell me the max size enlargements advisable from a scanner
with
a 2400 x 4800 dpi scan resolution (Would 10" x 8" be advisable). Scanning
35mm film negatives and slides

Also same question regarding a scanner with a res of 3200 x 6400

I have two scanners in mind the Epson RX700 with a res of 3200 x 6400 or
the
Epson RX620 with a res of 2400 x 4800.

I suppose a stand alone scanner would be better but for practical and
space
considerations it will have to be an all in one scanner...

I would be most unlikely to want to enlarge beyond A4 size. However I
would
probably want to do some cropping of 35mm based images.

Bottom line is would the Epson RX620 be sufficient for my needs...

Also please how do I work out the relationship between resolutions given
in
dpi and those resolutions stated in pixels. (is that pixels per inch??? I
don't understand!!) Thanks......

Figure that you want 200-300 dpi for a good quality print, trying to be
closer to the 300 than 200. Even at that, either scanner will do 8x10"
prints even allowing for some cropping. However, for best prints, it's not
going to be which of these two scanners you choose, it'll be your ability to
use photo imaging software to improve the images: color correction,
sharpness, curves (brightness and contrast), etc., etc., etc.

For on-screen displays, it's 72 pixels per inch so directly scanned images
at your proposed resolutions will need substantial cropping or resizing to a
smaller resolution via software before displaying (or email-attaching).

The scanners you're looking at can do a fine job with scanning prints, but
will be near-horrible for scanning negatives or slides. And even these poor
image scan from a negative/slide will require major despeckling time and
effort on your part. A dedicated slide/negative scanner with dust-removal
algorithms is almost mandatory for any adequate scanning of slides/negs. In
any event, negatives will be easier to scan than slides since the
substantially increased contrast in a slide can cause grief to less pricey
scanners. I haven't priced scanners recently but from my past searches, I'd
say that while many can live well with a $100 scanner for print scanning,
they may need to spend nearly $1000 for slides and negs to maintain their
overall happiness.

Just my 2 cents.

Craig
 
Many thanks Craig. Actually the all in one scanners I mentioned do advertise
as having dust removal sofware on board. They also print directly from
negatives/Transparencies and memory cards without the need for a
computer......Although that is not a function I crave for.....Most of my
printing these days wil be from a digital camera source. However i do have
some old negatives that would be nice to make prints from......I also have
photoshop LE given so called free with my digi cam......So restoring damage
and dust marks from negs although tedious would not be a major
problem......It is the basic resolution capabilities of the scanners that I
am mostly concerned with and you have I think re-assured me with regard to
that and 8" x 10" printing. Regards...
 
pixels are what the scanner creates and you see on your screen. dots are
what the printer creates from the pixel color information (R,G,B). There
are a variety of ways for a printer to create a color dot from a color
pixel, depending on the printing technology. For ink jet, this can be
anywhere from 3 "micro( defined as the smallest blob of ink the
printhead can deliver on the paper)" dots (cyan, magenta, yellow) to 6
or more. And then there are half tones (similar to what you see in a
newspaper image) used to create shading, by varying dot size,
patterning, etc. The printer driver has a "native" ppi that it wants to
be fed. For most that is around 300 ppi, although some will want as
high a 720 native. Any other PPI resolution fed to the driver will be
upsampled (pixels created by interpolation - creating something out of
nothing, so to speak) or downsampled (merging adjacent pixels by some
algorithm) to get to the native resolution the driver expects. The
driver will then figure out how to produce that color with a combination
of inks and dot size. So 300 ppi input for an 8 x 10 inch print , you
would need an image file with 7.2 million pixels. A slide is roughly 1
inch on the short side, so you would have to scan that slide at 2400 PPI
to get the required 8X enlargement (300 ppi input to the printer) so
that it can print 2400 dots on the 8 inch side of the print.

hope that helps.
 
The actual "resolving" capability of most of the flatbeds you mention is
around 2000 ppi or less (usually much less). Even the Nikon coolscans
only resolve around 2700 ppi. What the scanner vendors claim and what
the optics and sensor are capable of are two very different things.
 
degrub said:
The actual "resolving" capability of most of the flatbeds you mention is
around 2000 ppi or less (usually much less). Even the Nikon coolscans
only resolve around 2700 ppi. What the scanner vendors claim and what
the optics and sensor are capable of are two very different things.

So when you scan at 3200 ppi you are simply dividing the same image
information among more pixels.
 
Many thanks for the further info. I am beginning to see the light...or
perhaps I should say pixels.....:-)
 
Barn said:
Could anyone tell me the max size enlargements advisable from a scanner with
a 2400 x 4800 dpi scan resolution (Would 10" x 8" be advisable). Scanning
35mm film negatives and slides

Take 2400 ppi as the actual optical scanning resolution. According to
digital samply theory, the maximum possible resolution obtainable from
such a scanner would be half that or 1200 line pairs per inch. That
translates to 1200/24.5 ~ 47 lp/mm. The human visual system can
resolve at least 5 lp/mm at 10-12 inches, which would be nomral close
viewing distance for an 8 x 10 print. Some people claim to be able to
see better, even up to 10 lp/mm, but lets go with the lower figure.
47/5 ~ 9.5 X. The dimensions of a 35 mm frame are 24 x 36 mm. This
has a different aspect ratio (1:1.5) than 8 x 10 (1:1.25) so you have to
decide which dimension to use. If you don't crop, you should use the
long dimension to be 10 inches = 254 mm. (The resulting print will only
be 6 2/3 inches in the short dimension.) 254/36 ~ 7X.

As an exercise, you might try the calculation using 3200 ppi. You would
get 4/3 the theoretical resolution of about 63 lp/mm, allowing an
enlargement of about 12 X.

The result of all that calculation is that a PERFECT scanner with that
resolution would easily produce a scan which can produce a good 8 x 10
print. Unfortunately, real scanners are not perfect and they don't
deliver the actual theoretical resolution. Better scanners come
closer. For example, Epson flatbed scanners such as the 4990 and
similar scanners don't deliver anythi9ng higher than 30-35 lp/mm
allowing a maximum enlargement of at most 6 X, at least for close
viewing. Dedicated film scanners tend to do better. For example, my
Minolta SD IV seems to be able to yield adequate 8 x 10 prints. Higher
priced film scanners will do even better.


Also same question regarding a scanner with a res of 3200 x 6400

I have two scanners in mind the Epson RX700 with a res of 3200 x 6400 or the
Epson RX620 with a res of 2400 x 4800.

I don't have any experience with these printers, but I doubt if they can
deliver better than the 30-35 lp/mm I described above. That would make
them marginal at best for producing 8 x 10 prints from 35 mm negatives,
but it would of course depend on your standards.

It should be noted that there is a separate question of whether or not
the scanner produces enough pixels for a printer, in order to produce a
smooth print without color distortions. If you assume you are aiming at
sending the printer 300 ppi, and you intend a 7 X enlargement, than you
have to start with at least 2100 ppi. either of you choices would
provide that. But the resolution of fine detail is really a separate
issue, and that is based on different arithmetic as outlined above.
I suppose a stand alone scanner would be better but for practical and space
considerations it will have to be an all in one scanner...

Most dedicated film scanners take up relatively little space. If you
are scanning only 35 mm film, you would be better off with as good a
dedicated film scanner as you can afford and an inexpensive flatbed
scanner for doing reflective scanning.
I would be most unlikely to want to enlarge beyond A4 size. However I would
probably want to do some cropping of 35mm based images.

Bottom line is would the Epson RX620 be sufficient for my needs...

Again, it depends on your needs, but my guess is that it wouldn't do.
Also please how do I work out the relationship between resolutions given in
dpi and those resolutions stated in pixels. (is that pixels per inch??? I
don't understand!!) Thanks......

DPI stands for "dots per inch", but it is somewhat misleading when
applied to scanners. You really should use the term PPI which stands
for "pixels per inch". In any event it is the same number either way
for a scanner. These terms are different when appplied to printers.
The printer produces color by printing very small dots, each of one
specific color. The distribution of these dots determines what you see.
In your computer, the image is stored as pixels, which are really
triples of numbers, one for each of the primary colors Red, Green, and
Blue. Each pixel is capable of representing a very large number of
different colors. When you send a pixel to the printer, the printer
software/firmware must translate that color information into a
collection of dots as indicated above. You usually need many dots for
each pixel to get an accurate color rendition. Most modern printers can
handle fairly high pixel input counts and do a good job of producing
accurate color. But if the number of pixels is too small, you will
actually see them as discrete blocks in the print. If you send the
printer fewer than 200 ppi and the print is viewed close-up, most people
will find the print unsatisfactory. There are some differences of
opinion about the pixel density that is sufficient. Some people say 240
ppi is adequate, but most people use a figure like 300 ppi, and some go
even higher. If the print is going to be viewed from further away, then
it is okay to reduce the pixel density of what is sent to the printer.

As noted above, if you are enlarging the image, you have to multiply the
ppi sent to the printer by the enlargement factor.
 
Barn owl said:
Could anyone tell me the max size enlargements advisable from a scanner
with
a 2400 x 4800 dpi scan resolution (Would 10" x 8" be advisable). Scanning
35mm film negatives and slides

Also same question regarding a scanner with a res of 3200 x 6400

Almost every print I make on my Canon S900 photo printer is done at 300 DPI
and in truth I can't see with the naked eye any really noticeable
improvement to even 400 DPI -- though the printer's supposed to be capable
of doing well over 1000 DPI. If you can be sure that people aren't going to
stick their noses right up to the print, you can sometimes -- depending on
the textures and edges in the image -- get away with 250 DPI as well. Of
the image of a distant island "appearing" out of the fog [a al Brigadoon],
even 200 DPI would be quite usable and might even enhance the image as well.

Assuming 300 DPI for the print, you can make 8"x16" from a 2400 x 4800
original. Certainly 8-1/2"x11" is quite acceptable too (2400 / 8.5 = 282
DPI). You can sometimes stretch it to a borderless 11" x 14" since 2400 /
11" = 218 DPI; a 1/4" border all around gets you 2400 / 1.5" = 228 DPI.

Summary: for really super high quality prints, you might not want to go
beyond 5x7 but the vast majority of viewers would find no fault with an
8x10. Depending on the tonality, textures, and edges in the image, you may
be able to get away with an 11x14 too.

Norm
 
degrub said:
The actual "resolving" capability of most of the flatbeds you mention is
around 2000 ppi or less (usually much less). Even the Nikon coolscans only
resolve around 2700 ppi. What the scanner vendors claim and what the
optics and sensor are capable of are two very different things.

I hear this, but where's the actual evidence that applies to this actual
situation? "What the scanner vendors claim and what the optics and sensor
are capable of are two very different things." is too generic a statement to
be of any worth. _All_ scanner vendors? _All_ scanners?

What evidence do you have that the mentioned scanners are 2000 ppi or less
(usually much less) -- the bracketed statement means that the ones mentioned
can be divided into two groups, those that are close to 2000 ppi and those
that are much less than that but not always. That just doens't make sense
unless you know the production models differ so widely. In which case,
please provide some way for the hapless public to identify the really bad
ones. Serial #s perhaps?

What I'm saying is please feel free to stand conventional wisdom on its head
and point out the flaws and deficiences, but please do it with a couple of
actual facts.Otherwise, all it is is an opinion -- and they are plentiful
(and usually useless) on UseNet. :-)
 
I hear this, but where's the actual evidence that applies to this actual
situation? "What the scanner vendors claim and what the optics and sensor
are capable of are two very different things." is too generic a statement to
be of any worth.

Unless it's based on fact in which case it's not a generic statement
but a summary. ;o)
_All_ scanner vendors? _All_ scanners?

Yes, and yes. ;o)

That's easily verified with the so-called "slanted edge" test. A sharp
object, usually a razor, is scanned at an angle. The resulting "pixel
staircase" is then mathematically evaluated to deduce the actual
resolution i.e. what the scanner actually resolved.

Several reliable contributors to this group have confirmed that
manufacturer's claims are just wishful thinking dreamt up by their
marketroids (check the archives for details).

Googling for "<insert scanner model here>" and "slanted edge" should
return quite a few hits with actual data. There are some people who
prefer to do this rather than scan images. ;o) Googling for "scanner
bakeoff" should also return a few hits.

I don't know what's on this page but it sounds like it contains
exactly such data, although I have no idea how reliable it is:

http://www.jamesphotography.ca/bakeoff2005/results.html

Don.
 
Don said:
Unless it's based on fact in which case it's not a generic statement
but a summary. ;o)

Agreed. _If_ it's based on facts. Without the facts, I have to presume it's
an opinion. That said, I'm more than willing to evaluate actual facts.
Yes, and yes. ;o)

That's easily verified with the so-called "slanted edge" test. A sharp
object, usually a razor, is scanned at an angle. The resulting "pixel
staircase" is then mathematically evaluated to deduce the actual
resolution i.e. what the scanner actually resolved.

That's what I'm looking for. Obviously I'm not in a position to evaluate
all, not even some, scanners. I'll follow your suggestions below and see
where it leads me. Appreciated. :-)
 
That's what I'm looking for. Obviously I'm not in a position to evaluate
all, not even some, scanners. I'll follow your suggestions below and see
where it leads me. Appreciated. :-)

My pleasure.

BTW, I also like looking "under the hood" but I must confess I'm not
so concerned with the actual resolution for pragmatic reasons.

I mean, "it is what it is" and there isn't much I can do about it
other than to sulk that my 4000 dpi scanner is actually considerably
less. And ever since I started scanning I've got lots of other things
to sulk about. ;o)

Seriously though, there's the Minolta with 5400 dpi but it has other
problems (apparently the hardware is quite flimsy with plastic cogs
and whatnot). And the improvement is really not that great, far less
if we take actual resolution.

The only real quantum leap would be to use a drum scanner. However,
they are not only very expensive but they need a competent operator.
Using one is apparently not at all comparable to a consumer scanner.

Don.
 
The last several generations of film-only scanners will readily make
sharp, enlarger-quality images to large sizes.

Both the Minolta 5400/5400II are in that category, as is the older IV
(a bargain, used, especially if someone mostly scans traditonal silver
B&W film). The Nikon Coolscan V and older IV (which also has Ice and
also a bargain used) readily produce razor sharp enlargements to 13X18
(I do that with V) and beyond (beyond because the 13X18 are grain
sharp).

The "4000ppi" and "3200ppi" Epsons are plenty sharp within reasonable
limits, irrespective the real accuracy of their ppi assertions. My old
3200 probably resolves half its claimed 3200ppi in reality, maybe
1600ppi, but with medium format negatives it makes wonderfully sharp
11X14 and from 35 it was plenty sharp at 6X9, though it's distinctly
less sharp than my Nikon beyond that.

PPI is NOT a sharpness measure and does not directly indicate
enlargement potential. Flatbed scanners have issues with focus that
cause them to perform less well with film than with reflective art (eg
photographic prints). My Nikon does an incredibly good job of autofocus
on film, but if I scan an extremely curly piece of film (eg heat
damaged film) or intentionally defocus on good film, the scan will be
unsharp, even though it's still 4000ppi. Minoltas may do less well than
Nikon with autofocus, but they can be made equal with manual override.
 
The last several generations of film-only scanners will readily make
sharp, enlarger-quality images to large sizes.

The enlargement potential is really a side effect of resolution. Many
people (me included) don't print at all and even if they do the
primary purpose of scanning (in most cases) is to archive i.e. freeze
film deterioration and transfer the data into the lossless digital
domain.

So, resolution is important only in the sense to extract as much data
from film as possible.

Obviously, that's an "elastic" goal because unlike digital where all
pixels are the same size, grain varies dramatically regardless of film
speed. To really get all grain the consensus is that requires about
10,000 PPI.

Of course, that opens up another can of worms, film depth. At such
levels of precision the film has a distinct depth (so-called grain
clouds) which then begs the question where the focus should be. Front,
back or somewhere in the middle of this cloud?

Personally, I consider the nominal 4000 resolution to be quite
sufficient. Again, the consensus is that to benefit from any higher
resolution a number of additional requirements must be fulfilled such
as: fine grained film, use of tripod, perfect camera focus, etc. And
even then the benefits of higher resolution would be very marginal.

Therefore, in practical terms and most cases, at 4000 ppi the
resolution stops being an issue and other things such as dynamic range
become much more important otherwise all that additional resolution
would only be scanning noise.

Don.
 
Back
Top