22
\$\begingroup\$

I have no particular project in mind so I'm just asking in order to have a better understanding:

Let's say I'm an artist shooting pictures with a 35mm camera and my point would be to make really big prints for an exhibition (let's say 1m x 1,5m).

If I scan the negatives which one of these two answers would apply ?

  1. there's a theoretical maximum scanning resolution above which there is no relevant detail to be gotten, no matter how big the actual print is supposed be (so printing a picture digitally or manually would both give an equal result in terms of details). If this answer is true, which maximum resolution would that be ?

  2. No matter how high the scanning resolution is, there will always be a size above which an analog print will be more detailed than a digital one (unless using some kind of software interpolation maybe?).

I ask because every discussion about this topic ends up with people saying that big digital printing are very fine because people are supposed to see them from a distance anyway. But that's an assumption. It's not necessarily the case. So, what if we look closely? Would we see any difference then?

A part of me thinks that we would eventually see pixels but at the same time, if the pixels are below the size of the film grain, it won't be noticeable anyway. So, I'm really not sure.

\$\endgroup\$
9
  • 8
    \$\begingroup\$ What specific 35mm film? They don't all have the same resolution or grain size. Far from it. \$\endgroup\$
    – Michael C
    Commented Jan 18, 2021 at 6:26
  • 2
    \$\begingroup\$ If your point is to make really large prints for an exhibition, why are you limiting yourself to 35mm film? Why not Medium or Large Format film? \$\endgroup\$
    – Michael C
    Commented Jan 18, 2021 at 6:27
  • \$\begingroup\$ It's a theoretical question so consider the one with the smallest grain. My point is not to exhibit and it's not about a particular format, those are instances used as an illustration. My question is about a hypothetical resolution above which there is no detail to be extracted from a negative anymore, if any. without the example I used people would have digressed about how small the eye can see, which is not my question. That's why I took the smallest size of standards films and large prints as instances. I want to know the limits of resolutions, not to ridiculously enlarge small negatives. \$\endgroup\$ Commented Jan 18, 2021 at 6:37
  • 4
    \$\begingroup\$ Other answers and comments here talk about how much resolution the film itself is able to capture (...grain size, lines per mm...) But I've seen plenty of photographic prints hanging on gallery and museum walls that were enlarged to the point where the film grain itself was plainly visible. Do you want you scans to show off the film grain? or is your goal to hide the grain? \$\endgroup\$ Commented Jan 18, 2021 at 15:07
  • 2
    \$\begingroup\$ Well, clearly if you scan the film with an electron microscope, you will get at least as much detail as any analog enlargement process 😉. Realistically, due to grain size, you'll reach that point without needing to resort to such extremes. \$\endgroup\$
    – Matthew
    Commented Jan 19, 2021 at 13:14

5 Answers 5

35
\$\begingroup\$

According to Ken Rockwell:

Fuji Velvia 50 is rated to resolve 160 lines per millimeter. This is the finest level of detail it can resolve, at which point its MTF just about hits zero.

Each line will require one light and one dark pixel, or two pixels. Thus it will take about 320 pixels per millimeter to represent what's on Velvia 50.

320 pixels x 320 pixels is 0.1MP per square millimeter.

35mm film is 24 x 36mm, or 864 square millimeters.

To scan most of the detail on a 35mm photo, you'll need about 864 x 0.1, or 87 Megapixels.

That's about 8128 DPI. But in practice, consumer lenses and higher ISO films don't record that much detail. Scanning more than about 3048 DPI (12.4 MP) usually just increases the visibility of grain. By the time you see grain, you've captured as much detail as that particular piece of film is capable of containing.

The above numbers assume a perfect scanner. However, scanners are limited by the resolution of their lenses. So even though a sensor may be capable of 6400 DPI, the lens may only be capable of 3200 DPI or less.

\$\endgroup\$
7
  • \$\begingroup\$ Excellent, this is exactly what I wanted to know. Thank you! Yes, I'm talking about the effective resolutions not the announced ones. \$\endgroup\$ Commented Jan 18, 2021 at 14:08
  • \$\begingroup\$ "Each line will require one light and one dark pixel, or two pixels" Err, what? \$\endgroup\$
    – MikeB
    Commented Jan 20, 2021 at 11:29
  • \$\begingroup\$ @MikeBrockington To resolve a series of lines you need at least 2 pixels per line in the direction perpendicular to the lines. One for the line and one for the space. \$\endgroup\$
    – Matt
    Commented Jan 20, 2021 at 14:59
  • 1
    \$\begingroup\$ @MikeBrockington: I would assume that, in context, a "line" refers to a e.g. a black line on a white background. To resolve a bunch of adjacent such lines in a digital image, you will need to sample a pixel both from each line and from each gap between them. (In practice, you actually need more samples to be sure, since if the line and gap centers just happen to align with the edges between adjacent pixels, all you'll get is a flat gray blur. See also the Nyquist–Shannon sampling theorem.) \$\endgroup\$ Commented Jan 20, 2021 at 15:02
  • \$\begingroup\$ Sounds like the quoted author is actually mixing up two different concepts, or discussing a different scenario to this one, but perhaps that is a discussion for a different question. In theory, if the film grains are close enough in size and shape to the digital pixels, then one-to-one is enough for full depiction \$\endgroup\$
    – MikeB
    Commented Jan 21, 2021 at 11:36
6
\$\begingroup\$

Your 1m x 1.5m print has a resolution. If this were dpi, you could have 39.3in x 59.6in, and if that were 300 dpi, then it would be 11790px x 17880px. So, if you wanted to print at 300dpi, you basically need an image of 210,805,200 or 210 megapixels.

300dpi is the quality of most 5x7 or 8x10 photo prints.

Most large prints are not 300dpi. But this gives you an idea of your process. Find what resolution your large print can be printed at, and do some calculations to determine the ideal resolution to scan your film at.

The optimum scan resolution for 35mm may be 3200dpi. You are correct that the film's grain size will cause a limit to the scan resolutions that are reasonable.

\$\endgroup\$
0
3
\$\begingroup\$

The other answers are excellent and give a great explanation of the mechanics of film -> printing pixel resolution.

I wanted to highlight a particular (35mm) contemporary (black and white) film, which has very high resolution (800 line pairs/mm): Adox CMS II 20

From their materials:

Adox CMS II ISO 20 is the one of the sharpest, most fine grain films available. If used in pictorial photography the film achieves ISO 12 to 20 when used with Adotech II developer. The film achieves grain free enlargements of up to 8 feet diagonally. This equals to about 500 megapixels in a digital camera. This is due to the special monodisperse ultra high resolution emulsion of the film.

So that would give you a pretty good idea to the upper limit you could get with 35mm film. It will however depend on the whole system. The quality and size of the lenses of the camera and enlarger (or scanner and printer), will have a large impact to the actual resolution you could capture. A blurry photo, printed 8ft across, is still going to be a blurry picture. You just wont be able to see any grain 'blur'.

Although it is not your question, you could also get a larger print using 120 medium format film (up to 2.3 times) or even better a 8x10 large format camera.


Doing the math:

  • Printing at 300 dpi is equivalent to 12 linepairs/mm. (25.4mm/300)
  • So 800 lp/mm * 24mm (short side of the frame) = 19200
  • Scaling to print resolution 19200/ 12 = 1600mm
  • A 35mm frame is 2:3 so full size that would be 1600mm X 2400mm.
  • Diagonally that is 2884mm or about 9 ft (so with a bit of cropping and rounding 8ft)
\$\endgroup\$
3
  • \$\begingroup\$ The spec sheet says... "up to 800 l/mm"... But it's B&W and ISO 12-20. \$\endgroup\$
    – xiota
    Commented Jan 22, 2021 at 2:41
  • 1
    \$\begingroup\$ The ISO doesn't matter. If you are going to replicate an art work you can spend the time. OP Doesn't say if black and white is important or not. Resolution/lack of grain may be more important to them. Also using 3 lens filters OP, in theory, could create a colored print. My answer aims to give an upper bound, with caveats, that can be achieved with 35mm film. \$\endgroup\$ Commented Jan 22, 2021 at 8:33
  • \$\begingroup\$ I'm not saying this film doesn't deserve mention, just that its use comes with significant tradeoffs. OP says, "artist shooting pictures with a 35mm camera". If moving subjects or less than ideal lighting are involved, ISO does matter. \$\endgroup\$
    – xiota
    Commented Jan 22, 2021 at 12:39
1
\$\begingroup\$

A part of me thinks that [if we enlarged a digital image beyond a certain size] we would eventually see pixels.

I'm guessing that when you say "pixels," you mean, solid-color squares. But that is not actually what "pixel" means in the field of digital image processing. If you blow up a digital image to the point where each image pixel covers many printer dots or many screen pixels, then what you see is the output of a reconstruction filter.

If you see solid-color squares, that's probably the output of a trivial "nearest neighbor" reconstruction filter. But you might also see smooth transitions from each pixel to the next if a more sophisticated filter is used.

I'm not able to comment on either the technical reasons or the aesthetic reasons why somebody would choose one type over another.

\$\endgroup\$
1
\$\begingroup\$

You've got a processing chain :

Object -> lens -> shutter -> film -> scan -> print -> eye

Every section can be the bottleneck in the rendering to your brain, depending on what use case you are in. On top of that, your thinking in terms of resolution is too simple, as the eye does not in fine behaves like a computer but analyses contrasts. Contrast is not a binary concept as pixel can be. In other words, when a black and white grid becomes very small, your eye will see it more and more as an ondulation of light and dark grey. So it is artificial to put a figure on the limit.

In real life, manufacturer build their stuff trying to compromise on the bottlenecks, so you are indeed confronted with the problem to guess where you're limited.

Often, what you will get is :

  • object will not be the limitation except if you 're photographing fuzzy (fog) or very pure subjects (metal, sky)
  • lens will be a limitation in the corners, and more importantly everywhere if you're not on focal plan (which is impossible to be with a non-plan subject. Keep in mind that depth-of-field is a simplification relative to the eye own limitations. In theory there is only one distance which will be sharp. If you are not taking a picture of a vertical wall, you have fuzzy points, maybe smaller than the eye can see on the print, but fuzzy anyway)
  • shutter will bring you diffraction depending on the size of the film (around f/12 and above for a 24x36 digital)
  • film will not be a limitation for small prints, but will be for big, depending on the quality of your film
  • scan will not be, relatively to the film if you are using modern tools
  • print will not be if you are far enough from it. The eye will be.

To give you some figures :

  • A 35 mm color film is usually less defined than a 12 MP digital sensor.
  • A good scanner can go up to 4800 dpi, or 32 MP for a 24x36 mm pic. more than enough to scan your analog film.
  • The average eye is said to have a 1' separation power, which is equivalent to 8MP for a picture viewed at a distance equal to its diagonal (36 cm for a 21-29.7 A4 picture). It will often be the limiting factor.
  • Print is usually 300 dpi or 8 MP for an A4 so usually around the eye limitation at 36 cm, below at a greater distance

To make it short, if your picture is sharp the eye will usually in normal conditions be the limiting factor, then if you get closer the lens in the corners then the film in the center then the printer, then the scanner.

To make it even shorter, in real life, your picture will not be sharp, except on a few points if you're lucky ;-)

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.