Film's true DPI?

nzeeman said:
i dont think they are tie. they are totally different things. we cant compare those things.
Click to expand...

In my mind, and yours, this is true. But for the vast majority of people I suspect the comparison is pretty simplistic. It really doesn't matter.

What does matter, is that we use the materials we prefer, in the manner we prefer. DPI and resolution will be for the purists but art is not pure, it will be transcendent.
 
There are some variations of apples and oranges here. One is the way the media records light in its "raw form". For example in a 35mm full frame capture there is:

* Bayern RGB pixel pattern
* Fovecon RGB pixel pattern
* Grain replaced by dye clumps which we call grain on C41 and slide film
* Uniform tabular grain gathered in clumps on modern B/W film
* Mixture of different size grain gathered in clumps on traditional B/W
* Kodachrome

Skip the interneg process of scanning with film and print directly through the neg onto cibachrome, R4 or traditional B/W paper. Then print the same image captured with Bayern digital on a high quality ink jet. Print them all to 11 by 14.

My guess is that the 12 MP digital image will spank those printed from film - if your idea of viewing a print is to do so with a loupe magnifier, - which I'll call pixel peeping.

The traditional B/W print captures a different ascetic. Developed with a Pyro developer it will also allow for a wider range of luminance than digital or traditional C41 color film when printing.

A traditional slow speed in the B/W in the ISO 25 to 100 range will enhance the B/W image ascetically - especially in a 11 by 14 print from a 35mm negative, where one is really exceeding the mediums capability and the 12 MP digital wins - if the elimination of grain is your sole criteria.

Medium and large format are a different kettle of fish - but to answer the question in megapixel form - a 35mm color neg is only in the 8 to 11 megapixel range, if scanned. An 11 by 14 image definitely shows some deterioration of detail when viewed close up and especially with a loupe magnifier.

For scanning purposes I stick to C41 B/W, but when it comes to printing, I'll still print directly through the neg even with C41 B/W.
 
Last edited:
i think 35mm and 6-12mp dslrs are about the same. they begin to show their limits past 8x12, and you would do better using something else, whether it's medium format, large format, 16-21mp dslrs, or mfdb.
 
hahaha!!! *most* things don't matter to the vast majority of people... it is just so much more interesting when you find something that matters to you.

:)

For the record, after being "anti-digital" for years, then "going digital" for 5 years... and now pretty much going back to film again... my take is that a 12MP 5D "kinda/sorta" matches my film work... only *I* prefer *something* from the film.

Regarding traditional prints vs. good inkjet prints (IE. HP dye or Epson's new pigment), only the very very best traditional prints beat them when you consider *everything*. Pure resolution is not what makes for a beautiful photograph.

>>But for the vast majority of people I suspect the comparison is pretty simplistic. It really doesn't matter.
 
Quote:
Originally Posted by ruben
I would say that the best test will be to look side by side a small piece of a negative film converted into an analog print, versus the same small piece with a digital camera and converted into a print too. Print vs print of both small areas can be finally flat bed scaned and shown.

Ruben, problem is that most flatbed scanners suck, and commodity inkjets just as well. Not to mention the processing required in such a chain.
__________________
Eugene


Hi Eugene,
Perhaps I don't follow you quite well, but taken for granted that the final part of the comparizon (the flatbed in our case) "sucks", it will suck the same for the original digital piece as for the original film piece - therefore it will not disturb the comparizon.

Now, in case that you are going wider to the previous part of each process, then the problems should be held as inherent to the original media, by now, and therefore taken into account as ligitimate media limitations.

Cheers,
Ruben
 
off on some of the digital forums, members are freaking out on how to get the best football field sized print from 12MP while the last too museum shows I saw featuring Lee Friedlander (SF and Tucson) featured 6x9" and 8x12" prints.

8MP or scanned 35mm film can do a very good job for these sizes (only I *like* the way the film scan looks better).

>>As I said before. Who prints A3 every day?
 
ps any HP printer that takes a gray cartridge or highend Epson pigment printer will make wonderful b&w prints. Flatbed scanners mostly suck for 35mm and smaller, but are great for moderately sized MF negatives.
 
Dante, I think to a certain extent the delusion, or misnomer, is that there is any "data" in a negative. There is no data. That is a digital term. There is a collection of silver halide particles, assuming we are talking about B&W. Others have brought up the proper terms like apples to oranges and voodoo, etc. They're right.

As much as I love scanning my own negs, I am aware of the limitations of a process that is not really designed for what we do w/ it. Film was designed to be printed in a darkroom w/ an enlarger. I have had a number of really poor negs that have printed optically wonderfully, but scanning was a failure.

I know people are saying that digital is going to bury film, but they have been saying that for a long time. Again, apples to oranges. B&W film has beautiful grain and wonderful tonal range.

All I know is that I can take a tiny 35mm negative and scan it at 4000 dpi, or send it to a good lab to be printed in a darkroom, and I can make 12x18 prints that are just great. People do it all the time. You couldn't do that w/ a 6mp digital image, and even if you could it would look so much worse than a proper B&W film image. The only limits to 35mm film are behind the camera.
 
Last edited:
myoptic3 said:
Dante, I think to a certain extent the delusion, or misnomer, is that there is any "data" in a negative. There is no data. That is a digital term.
Click to expand...

No it isn't! 'Data' in the sense of 'things given [or known or assumed] as fact', at least in its singular, datum, goes back to 1646. A 'data sheet' about a film, camera, a lens is not a digital term either.

I agree that 'information' would be a better and more historically appropriate term, but I really wouldn't go so far as call 'data' a delusion or misnomer.

The fundamental flaw in all attempts at a direct comparison between film and digital is that a digital image is a regular array, while a film or dye image is randomly structured and contains information which may or may not be wanted (or indeed regarded as information), in the form of grain or dye-cloud structure. This is where the example of Art Garfunkel's hair (see post 6 above) is relevant.

Cheers,

R.
 
Last edited:
there is *nothing* in a negative until a human interprets it... and then that goes well beyond a numeric quantity...

Are you saying your 4990 works better than your 5400 for all films or just b&w? I've seen flatbeds doing a better job on b&w since the blur works like an anti-aliasing filter, but for color and C41 b&w grain aliasing isn't such a problem, so the 5400 (and my Coolscan V) should be much much better for 35mm...

>>My Epson 4990 works great for 35mm scans. Better than my highly acclaimed Minolta Elite 5400.
 
Matt(1pt4) said:
Where does one get one of those Kodak F-235 thingies? And do they work with VueScan?
Click to expand...
I Googled it but didn't get to the point of discovery on software compatibility. I've got better things to do with $10 grand, um... and don't have that kind of one-shot disposable cash.
 
Hmm... figures from a company selling drum scans by the Mb and a company trying to stay in business selling film.

The short answers, as far as I can tell are (and this may be subject to revision if I can think of better ways to run the numbers):

1. Most films resolve 50lp/mm, including virtually every color material. Black and white films are generally 50lp/mm too, except for TMX (133 at its best) and TMY (85 if tweaked). Source: manufacturer's sites and Norman Koren's site. This measurement excludes grain.

Theoretically, a 100lp/mm film should provide 34mp of resolution (if you count 200 lines per mm as the "pixels"). A 50lp/mm film gives 8.6mp.

2. In gross terms, a 100lp/mm lens, coupled with a 100lp/mm film will yield about 33lp/mm on a theoretically perfect 5000 dpi scanner (theoretically perfect meaning that it could actually resolve 2500 line pairs per inch, a standard that drum scanners might approach - remember, there is a focusing mechanism in the scanner). This assumes that system resolution is limited to the reciprocal of the sums of the reciprocals of the resolution of each piece in the system.

Following this through to its conclusion, TMX (call it 100/133/5000) provides about 4.5 mp of original subject data.

For comparison purposes, Tri-X (100/50/5000), provides 2.15Mp of subject data.

Looking at the end lp/mm, dropping down to 4000 dpi degrades the enlargeability by only 10% but saves 36% of the file space. This kind of difference (10% in linear resolution) is probably easily taken up by upsampling to print when necessary.

3. Using DPreview's measured extinction resolution numbers for the Kodak 14n and the Leica M8 and converting to lp/mm, both of these outresolve the 100/133/5000 (TMX) combination by a wide margin. This is not surprising, since you are skipping one of the generational losses you get in going from lens to film and film to scanner. Obviously, going up from base ISO is going to affect this.

I could be making the wrong assumption in the resolution limit formula. Does anyone have a different one?

Dante
 
Your assumptions are pretty accurate from my experience. I was easily making extremely good 16x20 prints with a canon D30, a 3mp DSLR, back when it was the current camera. Digital cameras yield results far better than the simple math would suggest.

Of course, if you like the look of film, that fact isn't relevant. The best way to get the look of film is to scan film.
 
Ray Nalley said:
Of course, if you like the look of film, that fact isn't relevant. The best way to get the look of film is to scan film.
Click to expand...

Actually the best way to get the look of film is to print directly through the negative onto photographic paper. The print is your scan.

I use digital for color prints and a traditional enlarger for producing prints from B/W film.
 
DAnte, I'm glad you asked for thoughts, rather than answers--as I do have some thoughts, even if no answers.

I read that Professor Max Berek figured he need to make a lens that would allow for one million discrete date points in the 24 x 36mm Leica negative. Arguably, that ought to equate to one megapixel. Of course, lenses and films are much better now, so let's try another approach.

DOF tables are generally based on a circle of confusion of 0.03mm. That would call for 1,200 points along the 36mm dimension, and 800 points along the 24mm height. And 1200 x 800 is 960,000. Hmm, that is pretty close to Dr. Berek's figure of one million.

Let's try a more demanding approach. Let's try a COC of .01mm. Now we have 3,600 times 2,400, which is 8.64 million. That sounds more in the ballpark, consistent with the observation that we need at least 6MP for excellent IQ. What would happen if we used 0.02mm as the COC? We then have 2.16 million data points, and it's common knowledge that we need more than 2.16MP. So it looks like a figure around 8MP would agree well with the performance data for top-quality lenses and films, using a very tight COC of .01mm.

But I don't think that is the end. I think that to capture all the data in a negative that has eight million data points, we will need more than 8 million pixels. I think the following formula may apply.

Rt = (r1 x r2) / (r1 + r2)

Where Rt is the combined resolution of film and scanner; r1 is the film's resolution; and r2 is the scanner's resolution.

So if you have a negative with 8 million data points, and you scan it with an 8MP scanner, then plugging those figures into the formula, the resulting image will have only 4MP worth of data. To minimize the degrading effect of the scanner on the original image, the scanner should have a resolution much higher than the film image. So let's try Roger's estimate fof 18MP as the scanner's resolution, and see what happens when we scan a negative with 8MP worth of data. We get an effective resolution of 5.538MP. That's in the ballpark for a good picture, but one might wish for better.

So let's try making the scanner resolution higher than the film's by an order of magnitude: 8MP for the film, 80MP for the scanner. Now we have 7.27MP of effective output resolution. That is getting reasonably close to what most of us might consider a top-quality image.

So in practical terms, it would appear to me (if my approach is valid) that we need all the scanner resolution we can get.

Well, Dante, you asked for ideas. Those are my thoughts. Now let's hear from the optical engineers! Is this a valid approach?
 
You must log in or register to reply here.
Back
Top Bottom