RichC
Well-known
It is commonly stated on the internet that a good 35 mm frame of ISO 100 film has a resolution equivalent to 20-25 MP. However, although this figure has some truth to it, it does not reflect what we find when film and digital are compared in real life. Unfortunately, every test I’ve encountered on the web has serious flaws – the fact that the comparisons contradict each other in every detail bears witness to this. So, the answer was to do my own investigation...
I'd be interested if anyone can find any major flaws in my argument.These calculations are of course somewhat crude, but suffice to give ballpark figures.
What none of these web comparisons addresses is the need to consider bottlenecks in equipment that impact the theoretical resolution of film – i.e. lens resolution and scanner resolution (assuming images will be scanned). Consider a typical scan of 35 mm film at a scanner resolution of 2700 ppi: the resulting file has dimensions of about 3800 × 2400 pixels = 9 MP – but this is not equivalent to a 9 MP image from a digital camera because both are theoretical values based solely on the number of pixels, and fail to account for differences in resolving power between the two mediums. Comparing, say, a 20 MP film scan with a 20 MP digital camera image is thus comparing apples with oranges. What we actually need to compare is the resolving power of frame of film and a digital sensor.
Basically, the ad men and marketing executives have co-opted the term "resolution" to mean something far removed from its true meaning, thus sowing confusion. "Resolution" has little directly to do with pixels and file size, and is correctly "the ability of an optical instrument or type of film to separate or distinguish small or closely adjacent objects".
Resolution (in its correct sense of "resolving power") lis governed by the Nyquist–Shannon theorem. This states that the maximum frequency (the Nyquist frequency) that can be resolved without loss of information is half the sampling frequency. Applied to digital photography (both digital cameras and scanned film), the sampling frequency is the resolution of the sensor or film (measured in line pairs per millimetre, lp/mm), and the Nyquist frequency is the resolution in megapixels (MP) needed to resolve all the information recorded by the sensor or film. Thus,
R = (2rh × 2rw)/10^6 = 4r^2hw/10^6
where r is the recorded resolution of the sensor or film (measured by experiment), h is the height of the sensor or film, w is the width of sensor or film, and R is the sensor or scanner resolution in MP needed to resolve the recorded detail.
(Note: I can't type superscripts, so ^ denotes that the following number is a superscript, e.g. 10^6 = 10 to the power 6.)
The real-world resolution of a digital sensor
We need to know how much a typical 35 mm full-frame dSLR can resolve. Rather than consider a top-line camera, let’s consider something more affordable – the Canon 5D Mark II. It’s a 21 MP camera, but this is not a direct measure of its resolution – it’s simply the number of pixels it records. Tests (see DPReview) show that this dSLR has a resolution of about 58 lp/mm. From our equation,
R = (4 × 58^2 × 24× 36)/10^6 ~12 MP
35 mm film
Typical professional 100 ISO colour negative film (Kodak Extar, etc.) has a resolution of about 70 lp/mm (as measured in tests – see film manufacturers’ websites), which is about the same as a dedicated film scanner (not a flat-bed scanner – which destroys resolution – even the best like the Epson V750 cuts this by half, to ~35 lp/mm). The formula gives
R = (4 × 70^2 × 24× 36)/10^6 ~ 17 MP
35 mm colour film thus records a little more detail than most 35mm full frame dSLRs. However, 35 mm colour film has a lot of ‘noise’ (i.e. grain), so that, visually, the smoother-looking dSLR image is preferred by most people, despite having slightly less visible detail overall. (Low-ISO B&W film has much higher resolution and less grain, and can show more detail than medium-format digital backs.)
Medium-format film
The formula gives the following for 645-format film, if we assume medium-format lenses resolve equally to 35 mm lenses:
R = (4 × 70^2 × 45× 60)/10^6 ~ 53 MP
And, for the 6×7 format:
R = (4 × 70^2 × 60 × 70)/10^6 ~ 82 MP
A typical digital back such as the 65 MP Phase One P65+ will resolve about 45 MP.
Large-format film
Turning to 4×5 film, the formula gives a resolution of
R = (4 × 70^2 × 100 × 125)/10^6 ~ 245 MP
Not the whole story...
Lens diffraction and depth of field
We need our photographs from the various formats to appear identical if we are to compare them: this means the same view and the same depth of field. A ‘standard’ (i.e. equivalent to the 50 mm lens used with the 35 mm format) medium-format lens is 80 mm – call it twice the focal length, for convenience. The 150 mm ‘standard’ lens used for large format is three times longer. The depth of field for medium and large format to match that of a 50 mm lens is obtained by multiplying the aperture by the relative increase in the focal length. If we assume the optimum aperture for resolution of f/5.6 for 35 mm film, then the medium- and large-format apertures giving an equivalent depth of field are
2 × f5.6 = f/11 (medium format)
3 × f/5.6 = f/16 (large format)
How does this affect resolution? Lens resolution changes with aperture, being at its optimum at f/5.6 for many lenses. The resolution will fall by 25% at f/11, 35% at f/16 and 50% at f/22 (e.g. see the lens reviews at DPReview). So, the resolution of 4×5 film used at a real-world aperture can thus be as low as ~ 175 MP.
Taking depth of field into account, our film resolutions become
R = 17 MP (35 mm)
R = 53 × 0.75 = 40 MP (645)
R = 82 × 0.75 = 62 MP (6×7)
R = 245 × 0.65 = 160 MP (4×5)
Scanning
Scanning film will reduce the resolution further, from manufacturers’ data. A good drum scan will result in a degradation of about 80%, so the above resolution are now
R = 14 MP (35 mm)
R = 42 × 0.75 = 32 MP (645)
R = 66 × 0.75 = 49 MP (6×7)
R = 196 × 0.65 = 127 MP (4×5)
As mentioned above, flat-bed scanners are awful for scanning film, reducing these resolution by half.
Contrast and grain
Digital photographs look sharper than photographic prints because of their greater edge contrast (which is what you enhance when ‘sharpening’ a digital image) and lack of grain. Let’s knock off an arbitrary 5 MP for ISO 100 digital and film images, to account for the ‘cleaner’ look of digital photographs. So, our final resolutions are now
R = 9 MP (35 mm)
R = 27 MP (645)
R = 44 MP (6×7)
R = 122 MP (4×5)
Note: these values should not be compared with manufacturers' sensor resolutions – those simply tell us how many pixels a sensor has, not how much information is recorded, i.e. the true resolution. They need to be compared with sensor resolutions obtained using the Nyquist–Shannon formula. Here are the "true" resolutions of a few digital cameras (the recorded resolution r is obtained from measurements of test charts by DPReview):
21 MP Canon 5D Mk II and Mk III
r = 58 lp/mm
R = 4r^2hw/10^6
R = (4 × 58^2 × 24× 36)/10^6 ~12 MP
18 MP Leica M9
r = 62 lp/mm
R = (4 × 62^2 × 24× 36)/10^6 ~13 MP
36 MP Nikon D800E
r = 102 lp/mm
R = (4 × 102^2 × 24× 36)/10^6 ~36 MP
(The measured resolution of the Nikon D800E is astonishing, matching the quoted (pixel) resolution of 36 MP, and outresolving DPeview's test chart, which tops out at 83 lp/mm.)
___________________________________________________
In summary, full-frame digital cameras of about 20 MP match the resolution of professional 100 ISO 35 mm colour film scanned on a dedicated film scanner, while cameras using Sony's 36 MP sensor such as the Nikon D800E match the resolution of 645 medium-format film. A 50 MP digital can match 6×7 film, but large-format film still outperforms the best digital camera by a wide margin.
___________________________________________________
Printing
Traditional darkroom prints made from film in an enlarger appear significantly less sharp than digital prints from scanned film: first, limitations from the apparatus – the enlarger and paper must be perfectly parallel, and this becomes more critical the larger the print, and that the lens itself degrades the image; secondly, scans can be made ‘sharper’ by adjusting the edge contrast – which cannot of course be done when printing directly from film.
The sharpest film print is thus digital, despite the loss in resolution from scanning: for best quality, we should scan the film and obtain a C-type or inkjet print – not use an enlarger.
Inkjet prints are slightly sharper and more expensive than C-type prints, but C-types are more robust and are a traditional silver-based photographic medium (if that’s important to you). Also, C-types have a different look to inkjet prints (that's different not better!) – the pigment in C-types sits in the surface, not on the surface, which gives them a subtle depth and three-dimensionality.
So, sharpness vs subtleness of depth – you can only have one!
I'd be interested if anyone can find any major flaws in my argument.These calculations are of course somewhat crude, but suffice to give ballpark figures.
What none of these web comparisons addresses is the need to consider bottlenecks in equipment that impact the theoretical resolution of film – i.e. lens resolution and scanner resolution (assuming images will be scanned). Consider a typical scan of 35 mm film at a scanner resolution of 2700 ppi: the resulting file has dimensions of about 3800 × 2400 pixels = 9 MP – but this is not equivalent to a 9 MP image from a digital camera because both are theoretical values based solely on the number of pixels, and fail to account for differences in resolving power between the two mediums. Comparing, say, a 20 MP film scan with a 20 MP digital camera image is thus comparing apples with oranges. What we actually need to compare is the resolving power of frame of film and a digital sensor.
Basically, the ad men and marketing executives have co-opted the term "resolution" to mean something far removed from its true meaning, thus sowing confusion. "Resolution" has little directly to do with pixels and file size, and is correctly "the ability of an optical instrument or type of film to separate or distinguish small or closely adjacent objects".
Resolution (in its correct sense of "resolving power") lis governed by the Nyquist–Shannon theorem. This states that the maximum frequency (the Nyquist frequency) that can be resolved without loss of information is half the sampling frequency. Applied to digital photography (both digital cameras and scanned film), the sampling frequency is the resolution of the sensor or film (measured in line pairs per millimetre, lp/mm), and the Nyquist frequency is the resolution in megapixels (MP) needed to resolve all the information recorded by the sensor or film. Thus,
R = (2rh × 2rw)/10^6 = 4r^2hw/10^6
where r is the recorded resolution of the sensor or film (measured by experiment), h is the height of the sensor or film, w is the width of sensor or film, and R is the sensor or scanner resolution in MP needed to resolve the recorded detail.
(Note: I can't type superscripts, so ^ denotes that the following number is a superscript, e.g. 10^6 = 10 to the power 6.)
The real-world resolution of a digital sensor
We need to know how much a typical 35 mm full-frame dSLR can resolve. Rather than consider a top-line camera, let’s consider something more affordable – the Canon 5D Mark II. It’s a 21 MP camera, but this is not a direct measure of its resolution – it’s simply the number of pixels it records. Tests (see DPReview) show that this dSLR has a resolution of about 58 lp/mm. From our equation,
R = (4 × 58^2 × 24× 36)/10^6 ~12 MP
35 mm film
Typical professional 100 ISO colour negative film (Kodak Extar, etc.) has a resolution of about 70 lp/mm (as measured in tests – see film manufacturers’ websites), which is about the same as a dedicated film scanner (not a flat-bed scanner – which destroys resolution – even the best like the Epson V750 cuts this by half, to ~35 lp/mm). The formula gives
R = (4 × 70^2 × 24× 36)/10^6 ~ 17 MP
35 mm colour film thus records a little more detail than most 35mm full frame dSLRs. However, 35 mm colour film has a lot of ‘noise’ (i.e. grain), so that, visually, the smoother-looking dSLR image is preferred by most people, despite having slightly less visible detail overall. (Low-ISO B&W film has much higher resolution and less grain, and can show more detail than medium-format digital backs.)
Medium-format film
The formula gives the following for 645-format film, if we assume medium-format lenses resolve equally to 35 mm lenses:
R = (4 × 70^2 × 45× 60)/10^6 ~ 53 MP
And, for the 6×7 format:
R = (4 × 70^2 × 60 × 70)/10^6 ~ 82 MP
A typical digital back such as the 65 MP Phase One P65+ will resolve about 45 MP.
Large-format film
Turning to 4×5 film, the formula gives a resolution of
R = (4 × 70^2 × 100 × 125)/10^6 ~ 245 MP
Not the whole story...
Lens diffraction and depth of field
We need our photographs from the various formats to appear identical if we are to compare them: this means the same view and the same depth of field. A ‘standard’ (i.e. equivalent to the 50 mm lens used with the 35 mm format) medium-format lens is 80 mm – call it twice the focal length, for convenience. The 150 mm ‘standard’ lens used for large format is three times longer. The depth of field for medium and large format to match that of a 50 mm lens is obtained by multiplying the aperture by the relative increase in the focal length. If we assume the optimum aperture for resolution of f/5.6 for 35 mm film, then the medium- and large-format apertures giving an equivalent depth of field are
2 × f5.6 = f/11 (medium format)
3 × f/5.6 = f/16 (large format)
How does this affect resolution? Lens resolution changes with aperture, being at its optimum at f/5.6 for many lenses. The resolution will fall by 25% at f/11, 35% at f/16 and 50% at f/22 (e.g. see the lens reviews at DPReview). So, the resolution of 4×5 film used at a real-world aperture can thus be as low as ~ 175 MP.
Taking depth of field into account, our film resolutions become
R = 17 MP (35 mm)
R = 53 × 0.75 = 40 MP (645)
R = 82 × 0.75 = 62 MP (6×7)
R = 245 × 0.65 = 160 MP (4×5)
Scanning
Scanning film will reduce the resolution further, from manufacturers’ data. A good drum scan will result in a degradation of about 80%, so the above resolution are now
R = 14 MP (35 mm)
R = 42 × 0.75 = 32 MP (645)
R = 66 × 0.75 = 49 MP (6×7)
R = 196 × 0.65 = 127 MP (4×5)
As mentioned above, flat-bed scanners are awful for scanning film, reducing these resolution by half.
Contrast and grain
Digital photographs look sharper than photographic prints because of their greater edge contrast (which is what you enhance when ‘sharpening’ a digital image) and lack of grain. Let’s knock off an arbitrary 5 MP for ISO 100 digital and film images, to account for the ‘cleaner’ look of digital photographs. So, our final resolutions are now
R = 9 MP (35 mm)
R = 27 MP (645)
R = 44 MP (6×7)
R = 122 MP (4×5)
Note: these values should not be compared with manufacturers' sensor resolutions – those simply tell us how many pixels a sensor has, not how much information is recorded, i.e. the true resolution. They need to be compared with sensor resolutions obtained using the Nyquist–Shannon formula. Here are the "true" resolutions of a few digital cameras (the recorded resolution r is obtained from measurements of test charts by DPReview):
21 MP Canon 5D Mk II and Mk III
r = 58 lp/mm
R = 4r^2hw/10^6
R = (4 × 58^2 × 24× 36)/10^6 ~12 MP
18 MP Leica M9
r = 62 lp/mm
R = (4 × 62^2 × 24× 36)/10^6 ~13 MP
36 MP Nikon D800E
r = 102 lp/mm
R = (4 × 102^2 × 24× 36)/10^6 ~36 MP
(The measured resolution of the Nikon D800E is astonishing, matching the quoted (pixel) resolution of 36 MP, and outresolving DPeview's test chart, which tops out at 83 lp/mm.)
___________________________________________________
In summary, full-frame digital cameras of about 20 MP match the resolution of professional 100 ISO 35 mm colour film scanned on a dedicated film scanner, while cameras using Sony's 36 MP sensor such as the Nikon D800E match the resolution of 645 medium-format film. A 50 MP digital can match 6×7 film, but large-format film still outperforms the best digital camera by a wide margin.
___________________________________________________
Printing
Traditional darkroom prints made from film in an enlarger appear significantly less sharp than digital prints from scanned film: first, limitations from the apparatus – the enlarger and paper must be perfectly parallel, and this becomes more critical the larger the print, and that the lens itself degrades the image; secondly, scans can be made ‘sharper’ by adjusting the edge contrast – which cannot of course be done when printing directly from film.
The sharpest film print is thus digital, despite the loss in resolution from scanning: for best quality, we should scan the film and obtain a C-type or inkjet print – not use an enlarger.
Inkjet prints are slightly sharper and more expensive than C-type prints, but C-types are more robust and are a traditional silver-based photographic medium (if that’s important to you). Also, C-types have a different look to inkjet prints (that's different not better!) – the pigment in C-types sits in the surface, not on the surface, which gives them a subtle depth and three-dimensionality.
So, sharpness vs subtleness of depth – you can only have one!
Last edited:
