Scanned film vs. digital

What has a better image quality? The new, full size 12meg chips in slr's or film shot 35mm and then scanned at a very high resolution? I am only interested in B&W. Also what is the finest B&W printing process available?

Gil Junger said:

What has a better image quality? The new, full size 12meg chips in
slr's or film shot 35mm and then scanned at a very high
resolution? I am only interested in B&W. Also what is the finest
B&W printing process available?

Click to expand...

For B/W, use film. It has far greater dynamic range and somewhat greater resolution. The best B/W printing process is still traditional silver printing, but high-quality inkjet isn't bad either -- the Epson 2100/2200 isn't too bad out of the box, and there are solutions with four gray inks for other printers.

Petteri
--



Portfolio: [ http://www.seittipaja.fi/index/ ]
Pontification: [ http://www.seittipaja.fi/ ]

My previous digital camera (Canon 1D) was not equal to well exposed and high quality scanned 35mm film. When I purchased the Canon 1DS it changed dramatically. I think the potential quality of images from it are very close to 645 film images. Properly exposed images shot with the Canon 1DS at 100 ISO and properly converted from RAW to TIF and converted to B & W will surpass anything that I have ever been able to produce in the darkroom.

Gil Junger said:

What has a better image quality? The new, full size 12meg chips in
slr's or film shot 35mm and then scanned at a very high
resolution? I am only interested in B&W. Also what is the finest
B&W printing process available?

Click to expand...

Gil Junger said:

What has a better image quality? The new, full size 12meg chips in
slr's or film shot 35mm and then scanned at a very high
resolution? I am only interested in B&W. Also what is the finest
B&W printing process available?

Click to expand...

Canon 1Ds (11 MP) compares well with medium format. See http://luminous-landscape.com/reviews/cameras/1ds/1ds-field.shtml

That should be enough for 35mm. One observation, though. In this forum people seem to confuse the number of megapixels pulled from a scan with the information in the negative. Well, more megapixel from a scan does not equal more picture. Once you start seeing grain, it means the info recorded is more about grain than about the content. It is theoretically impossible to have more than 7 million elements (pixels in the digital world) recorded on the 35mm frame. Real cameras fall far short off that.

Probably, it is safe to assume, that for all practical purposes a good 6 MP DSLR is equivalent to the 35mm.

Juri D said:

Gil Junger said:

What has a better image quality? The new, full size 12meg chips in
slr's or film shot 35mm and then scanned at a very high
resolution? I am only interested in B&W. Also what is the finest
B&W printing process available?

Click to expand...

Canon 1Ds (11 MP) compares well with medium format. See
http://luminous-landscape.com/reviews/cameras/1ds/1ds-field.shtml

That should be enough for 35mm. One observation, though. In this
forum people seem to confuse the number of megapixels pulled from a
scan with the information in the negative. Well, more megapixel
from a scan does not equal more picture. Once you start seeing
grain, it means the info recorded is more about grain than about
the content. It is theoretically impossible to have more than 7
million elements (pixels in the digital world) recorded on the 35mm
frame. Real cameras fall far short off that.

Probably, it is safe to assume, that for all practical purposes a
good 6 MP DSLR is equivalent to the 35mm.

Click to expand...

Juri,

Or, for example, he could look here at an actual pixels crop of 4000 dpi scanned 35mm Fuji Provia 100F exposed at EI 200 (the uncropped full frame is 21.4 megapixels), and maybe realize that while you are doing a great job of quoting the digital photography conventional wisdom, the truth may be just a wee bit different from that.
http://forums.dpreview.com/forums/read.asp?forum=1005&message=4695834

But digital does like its mythology...

Of course, Gil was referring to black and white, and that is one area where even the digitally hard nosed all agree that digital is just not up to the ability of film.

My best,

Ed

--
http://www.blackmallard.com/cal_ls/
California Light and Structure

http://www.blackmallard.com/o_barn/
One Barn

Believe me, Gil, there are so many variables that I cannot think of a "good for everything" reply. What lens are you going to use? What film? An available-light shot with a Leica RF and a Summicron 28 or 35 is hard to beat by any system. I found that Kodak Portra BW 400 scans better than any other film I ever used, but a friend of mine who uses Technical Pan with Hasselblad gets results very near to anything you may ever have dreamt of. And the most important question is: how big are you going to enlarge? Digital "looks" better (it prints smoother, cleaner) until you get to the "critical size" (which depends on the sensor/lens combo); than it falls off quickly, while you can go on enlarging film to bigger sizes. For normal uses (prints up to 11"x17") I think a good digital camera with some Photoshop skill gives cleaner results. Your subject will also influence the reply: you can enlarge a head and shoulder portait of any origin as much as you want, as long ad eylashes and hair are visible a distinguishible to the eye. But for a group portait or a landscape with far away foliage only 6x7 film, or a horribly expensive digital back, may be the answer.

Fabio

Good exposition on what to expect in brief from digital or film. I think the most important limitation for digital is as you said the critical point when enlarging more degrades the digital image very quickly while a high resolution scan on a fine emulsion film can go muuuuuuuuuuuch bigger.

Fabio Amodeo said:

Believe me, Gil, there are so many variables that I cannot think of
a "good for everything" reply. What lens are you going to use? What
film? An available-light shot with a Leica RF and a Summicron 28 or
35 is hard to beat by any system. I found that Kodak Portra BW 400
scans better than any other film I ever used, but a friend of mine
who uses Technical Pan with Hasselblad gets results very near to
anything you may ever have dreamt of. And the most important
question is: how big are you going to enlarge? Digital "looks"
better (it prints smoother, cleaner) until you get to the "critical
size" (which depends on the sensor/lens combo); than it falls off
quickly, while you can go on enlarging film to bigger sizes. For
normal uses (prints up to 11"x17") I think a good digital camera
with some Photoshop skill gives cleaner results. Your subject will
also influence the reply: you can enlarge a head and shoulder
portait of any origin as much as you want, as long ad eylashes and
hair are visible a distinguishible to the eye. But for a group
portait or a landscape with far away foliage only 6x7 film, or a
horribly expensive digital back, may be the answer.

Fabio

Click to expand...

--
Cristian

Gil Junger said:

What has a better image quality? The new, full size 12meg chips in
slr's or film shot 35mm and then scanned at a very high
resolution? I am only interested in B&W. Also what is the finest
B&W printing process available?

Click to expand...

--
thanks for sharing clean
informations!

For good scanning use a good dia scanner, but very expensive.
See scanner by Canon, they have a good dia scanner for low budget.
See link workflow from pro like Caponigro: http://www.johnpaulcaponigro.com/

To print use the piezography (

last i think go digital if you have the money, no more scanning, no more money for film and development. and use piezography system for printing and you have best result.

let me know what you think!

Ed Leys said:

Juri,

Or, for example, he could look here at an actual pixels crop of
4000 dpi scanned 35mm Fuji Provia 100F exposed at EI 200 (the
uncropped full frame is 21.4 megapixels), and maybe realize that
while you are doing a great job of quoting the digital photography
conventional wisdom, the truth may be just a wee bit different from
that.
http://forums.dpreview.com/forums/read.asp?forum=1005&message=4695834

But digital does like its mythology...

Of course, Gil was referring to black and white, and that is one
area where even the digitally hard nosed all agree that digital
is just not up to the ability of film.

My best,

Ed

Click to expand...

Ed,

Your example confirms my point. Pixel-wise, there are not many details in the picture. Perhaps, you could get a similar picture by starting off with a quarter-sized one [one half in linear dimentions], ressing it up 200%, add some noise and blur it.

That should give you an idea about information content. I'm sure film can do better than that. In fact when I said 6 MP is enough, I meant overall impact of the picture. By all measurements, a good 35mm has somewhat more detail than a good 6 MP digital, but these details are weaker. More importantly, details than are important for perception are stronger in digital. So, on the balance, I'd call it a draw.

Cheers,
Juri

my thoughts:

• second that from fabio, Kodak Portra BW 400 is great on 35mm,

or medium format tri-x ,...

• are you the film director ...
• how many B&W movies do see in the movie theater or private shows ?

and how many videos do you rent and watch on TV ?

• do you own a darkroom ?
• if you can afford it, hire a pro.
• a 12MP digital might be sufficient, it is different not better.
• 99% of all digital print services are on color paper

( not counting inkjets, iris,...)
thought you can output on film 4x5, or 8x10, and print from there,....
if you want a real and good B&W print.

• how many scanns have you done lately ?
• oh, the answer,.... NO it is not better, it is different

ok,ok,.... most of us don't earn a living on B&W, so we adapt our existing equipment,

quadtone printing, lightjet output on flex and B&W paper,....do we like it, yes, is it better ?

cheers, Robert Schultz
--
http://www.RobsPhoto.com

Juri D said:

Ed,

Your example confirms my point. Pixel-wise, there are not many
details in the picture. Perhaps, you could get a similar picture by
starting off with a quarter-sized one [one half in linear
dimentions], ressing it up 200%, add some noise and blur it.

That should give you an idea about information content. I'm sure
film can do better than that. In fact when I said 6 MP is enough, I
meant overall impact of the picture. By all measurements, a good
35mm has somewhat more detail than a good 6 MP digital, but these
details are weaker. More importantly, details than are important
for perception are stronger in digital. So, on the balance, I'd
call it a draw.

Cheers,
Juri

Click to expand...

Yep, Juri, I've conducted that experiment, though on a different photograph. First my 4000 ppi scan was printed (suitably sharpened - scanners tend to have that negative effect on a photograph) at 300 ppi on 13 inch x 19 inch paper--no interpolation, of course. Consider it a contact print of scanner output. Then I resampled the file linerarly downward to one half the image length (a quarter of the pixels, overall) then resampled upward by two, back to the original file size. Resampling in both directions was by "nearest neighbor"...the most philosophically appropriate resampling for this kind of exercise. Oh, the work was carried out on the version of the file before unsharpening for printing was introduced.

I unsharpened the file that resulted from the down/up sampling using exactly the same unsharpen mask parameters as before. I did try to be consistant with all this. The final result? Not bad, a digihead might even have thought it was great. But it really did not have the sharpness and detail of the straight to print pass.

Of course, what I was responding to was also your comment about "grain". And how much did you see in the crop? That's shooting and push processing for EI 200 at that (same film, same one stop push as used for my down/up experiment).

And, of course, it's no sweat to swap in a different "CCD", for different purposes...or if a new line of "CCD"s hit the streets.

But, of course, this whole thread was supposed to be about black and white, wasn't it?

My best,

Ed

--
http://www.blackmallard.com/cal_ls/
California Light and Structure

http://www.blackmallard.com/o_barn/
One Barn

Ed Leys said:

Yep, Juri, I've conducted that experiment, though on a different
photograph. First my 4000 ppi scan was printed (suitably sharpened

• scanners tend to have that negative effect on a photograph) at

300 ppi on 13 inch x 19 inch paper--no interpolation, of course.
Consider it a contact print of scanner output. Then I resampled
the file linerarly downward to one half the image length (a quarter
of the pixels, overall) then resampled upward by two, back to the
original file size. Resampling in both directions was by "nearest
neighbor"...the most philosophically appropriate resampling for
this kind of exercise. Oh, the work was carried out on the
version of the file before unsharpening for printing was
introduced.

I unsharpened the file that resulted from the down/up sampling
using exactly the same unsharpen mask parameters as before. I
did try to be consistant with all this. The final result? Not
bad, a digihead might even have thought it was great. But it
really did not have the sharpness and detail of the straight to
print pass.

Of course, what I was responding to was also your comment about
"grain". And how much did you see in the crop? That's shooting
and push processing for EI 200 at that (same film, same one stop
push as used for my down/up experiment).

Click to expand...

Ed, no offense, but the nearest neighbor is not an interpolation. It's just doubling up rows and columns. Besides, there is no need to print something out at 13x19, a small portion of the image would do.

Now, information content is a tricky one. You have probably noticed that in your picture there is not a single detail one pixel wide or tall. That's a good sign that your information content per pixel is well below one. How much is below is difficult to say.

In the same area of a picture this coefficient [information content per pixel] cound be different due to optical effects on different sizes of grain clumps. Say, on average it is one quarter. The key word here is on average. It means it could be, say 2 or 6. Now, if you discard three quarters of info, your details with coefficient 6 will remain in the picture, but with 2 will be completely gone.

It means, that in this particular case, if you resized the picture down to one quarter, and, then upsize it back, and get a similar amount of detail, your picture's information content will be not one quarter, but one sixth of the original.

Now, interpolation. Printers are funny animals. They require native DPI to shine. The best tool for printing would be Qimage, you can rely on it to handle all interpolation [Vector or Lanczos seem to be the best in most situations]. Sharpening can be handled in Qimage too. You can do a similar thing in Photoshop, but it's trickier [see Fred Miranda's tool on that].

I did, by the way, experiment with a small-size picture from your post. I found very little difference before and after. In fact, had I found no difference at all after discarding three quarters of info, this would mean that the info content of your picture would probably be one sixth of 21 MP [see above why]. So, probably, your image has 4 to 5 mil of information points. Which is very good [7 is maximum, but it's very hard to achieve].

It does not mean that all pixels are golden in digital either. In fact, the best 6MP DSLR cannot have more than 4-5 mil of information points. However, as I indicated in my previous post, the detail from 6 MP digital is stronger, that's why many people prefer it over 35mm when printed.

Cheers,
Juri

Fabio Amodeo said:

And the most important
question is: how big are you going to enlarge? Digital "looks"
better (it prints smoother, cleaner) until you get to the "critical
size" (which depends on the sensor/lens combo); than it falls off
quickly, while you can go on enlarging film to bigger sizes. For
normal uses (prints up to 11"x17").

Click to expand...

Fabio,

For best results any image should be printed at native printer resolution (720 DPI for my Epson). So, practically any image (even a 300dpi one from a scan) needs to be interpolated. Qimage seems to be the best tool for that. Using Qimage or Genuine Fractals you can easily go past "critical size".

Juri

Ed Leys said:

Yep, Juri, I've conducted that experiment, though on a different
photograph. First my 4000 ppi scan was printed (suitably sharpened

• scanners tend to have that negative effect on a photograph) at

300 ppi on 13 inch x 19 inch paper--no interpolation, of course.
Consider it a contact print of scanner output. Then I resampled
the file linerarly downward to one half the image length (a quarter
of the pixels, overall) then resampled upward by two, back to the
original file size. Resampling in both directions was by "nearest
neighbor"...the most philosophically appropriate resampling for
this kind of exercise. Oh, the work was carried out on the
version of the file before unsharpening for printing was
introduced.

I unsharpened the file that resulted from the down/up sampling
using exactly the same unsharpen mask parameters as before. I
did try to be consistant with all this. The final result? Not
bad, a digihead might even have thought it was great. But it
really did not have the sharpness and detail of the straight to
print pass.

Click to expand...

Ed, to illustrate my point I am including two images, in no partiicular order: one image is your original pic sharpened, and the other is your image downsized, upsized and sharpened. I wonder if one can guess which is which.

Cheers,
Juri

http://www.pbase.com/image/22535217

http://www.pbase.com/image/22535218

Juri D said:

Ed Leys said:

Yep, Juri, I've conducted that experiment, though on a different
photograph. First my 4000 ppi scan was printed (suitably sharpened

• scanners tend to have that negative effect on a photograph) at

300 ppi on 13 inch x 19 inch paper--no interpolation, of course.
Consider it a contact print of scanner output. Then I resampled
the file linerarly downward to one half the image length (a quarter
of the pixels, overall) then resampled upward by two, back to the
original file size. Resampling in both directions was by "nearest
neighbor"...the most philosophically appropriate resampling for
this kind of exercise. Oh, the work was carried out on the
version of the file before unsharpening for printing was
introduced.

I unsharpened the file that resulted from the down/up sampling
using exactly the same unsharpen mask parameters as before. I
did try to be consistant with all this. The final result? Not
bad, a digihead might even have thought it was great. But it
really did not have the sharpness and detail of the straight to
print pass.

Click to expand...

Ed, to illustrate my point I am including two images, in no
partiicular order: one image is your original pic sharpened, and
the other is your image downsized, upsized and sharpened. I wonder
if one can guess which is which.

Cheers,
Juri

http://www.pbase.com/image/22535217

http://www.pbase.com/image/22535218

Click to expand...

I don't know which is which -- bur 22535218 looks better to me when I opened both (side by side) in Photoshop 7.
Look at the top right side of the images -- 218 is better and has more clarity.
--
Vernon...
http://www.pbase.com/vrain

Hey Gil,

Here are the facts. Maybe film has slightly higher resolution than digital using an expensive drum scanner, but who really has one of those at their disposal. Personally I own the S2 and I have NEVER EVER gotten 13x19" prints as good as I get from a Canon S9000 printer and my S2 Pro. No matter where the images were developed including a pro lab. That is the reality for me. I have many comparisons of the S2 to film and you would be amazed... if you would like to see them e-mail me:

[email protected]

People can argue all day about which is better, digital SLR or 35mm. But for me the proof is undeniable. S2 wins every time. I doubt many people have actually tested the differences. They just say what they 'think' is factual without any actual proof. I will never use 35mm film again.

B&W might be a bit different story. The S2 can handle it, even has a B&W mode, but printers are the weak link. I have yet to find a printer that can reproduce B&W right. Color prints are great though!

Regards,
Sean

Gil Junger said:

What has a better image quality? The new, full size 12meg chips in
slr's or film shot 35mm and then scanned at a very high
resolution? I am only interested in B&W. Also what is the finest
B&W printing process available?

Click to expand...

Sean Keegan said:

B&W might be a bit different story. The S2 can handle it, even has
a B&W mode, but printers are the weak link. I have yet to find a
printer that can reproduce B&W right. Color prints are great though!

Click to expand...

Try ImagePrint software rip with any Epson printer. You will have excellent B&W, both from scanned film and from digicamera. Colour is wonderful too. It orks only with recent systems (Mac OSX or Windows XP) and it is not economical, at about $500, but I recommend it.

Fabio

Sean Keegan said:

People can argue all day about which is better, digital SLR or
35mm. But for me the proof is undeniable. S2 wins every time. I
doubt many people have actually tested the differences. They just
say what they 'think' is factual without any actual proof. I will
never use 35mm film again.

Click to expand...

Check out this, Sean:
http://luminous-landscape.com/reviews/cameras/d30/d30_vs_film.shtml
even 3MP is good enough.

Juri

Juri D said:

Ed Leys said:

Yep, Juri, I've conducted that experiment, though on a different
photograph. First my 4000 ppi scan was printed (suitably sharpened

• scanners tend to have that negative effect on a photograph) at

300 ppi on 13 inch x 19 inch paper--no interpolation, of course.
Consider it a contact print of scanner output. Then I resampled
the file linerarly downward to one half the image length (a quarter
of the pixels, overall) then resampled upward by two, back to the
original file size. Resampling in both directions was by "nearest
neighbor"...the most philosophically appropriate resampling for
this kind of exercise. Oh, the work was carried out on the
version of the file before unsharpening for printing was
introduced.

I unsharpened the file that resulted from the down/up sampling
using exactly the same unsharpen mask parameters as before. I
did try to be consistant with all this. The final result? Not
bad, a digihead might even have thought it was great. But it
really did not have the sharpness and detail of the straight to
print pass.

Click to expand...

Ed, to illustrate my point I am including two images, in no
partiicular order: one image is your original pic sharpened, and
the other is your image downsized, upsized and sharpened. I wonder
if one can guess which is which.

Cheers,
Juri

http://www.pbase.com/image/22535217

http://www.pbase.com/image/22535218

Click to expand...

Well, Juri, your phrasing isn't too tight, since just by reading your words (but without looking at what you did), when you say "your original pic sharpened" it could mean that you just attempted to take my 'unsharpened' example and post it as is, or it could mean that you took my pre-unsharpened photograph and sharpened it in your own methodology--although you admit to down/up rezzing and sharpening the other.

But in actually looking at the photographs... 22535217 is very gross. It is clear that whatever was done to it was a desecration. On the other hand, 22535218 seems to have all the fine detail of my 'unsharpened' version largely intact (but did you perhaps re-jpeg it? ...or maybe pbase did). If that is what your down/up sampliing and sharpening resulted in, I am impressed indeed. I will dash straightaway and wrest the Fuji S2 from my sister's hands never to let her see it again...and kneel before the true altar of digital for the rest of my natural life.

Uh, you're kidding, aren't you??

You messed up 217 big time. Got that? If you cannot see that yourself, you are the perfect candidate for digital photography. Maybe you honestly just don't know what those extra pixels do in a photograph, I suppose that's possible.

Oh, I'm sorry, I was supposed to "guess"...

Of course, you are new to the forums (and maybe new to photography?), some people around here have gotten to know me...

http://forums.dpreview.com/forums/read.asp?forum=1018&message=6411522

Nice talking witcha,

Ed

--
http://www.blackmallard.com/cal_ls/
California Light and Structure

http://www.blackmallard.com/o_barn/
One Barn

Ed Leys said:

22535218 seems to have all the
fine detail of my 'unsharpened' version largely intact (but did you
perhaps re-jpeg it? ...or maybe pbase did).

Click to expand...

Sorry, pal. I made it too easy. I made an equivalent of 2 mln. of information elements out of your "21 MP". So I guess it was easy to guess. Now guess what - try to tell these apart (the prize would be to know that your camera/scanner setup produces more than 4 mln. of information elements .

http://www.pbase.com/image/22543493
http://www.pbase.com/image/22543501

Cheers,
Juri

P.S.: read up on terminology a bit: you do not unsharpen with unsharp masking, you actually sharpen the image. So I did take your original pic and sharpened it a little with unsharp mask (to keep things interesting I will not tell what parameters I used).