Re: Grain Size of Photographic Film?

Thanks for your question, David - it’s an interesting one. On the one hand 
it could be very simple to answer - I could just give you some numbers and 
leave you to “gauge the difference in resolutions” as you put it, or I 
could do the difficult bit - i.e. help you with the “gauging” bit. I’ll do 
the latter, but this won’t be short!

Before getting into all the complexities, here’s a graph that compares the 
resolutions of different image capturing devices by plotting their 
positions on a curve of comparative image quality as a function of the 
pixel density. The chart shows that to approach photo quality resolution 
with a digital camera, you need a sensor with very high pixel density and 
high dynamic range.  For a 4X print, it takes about a million pixels to 
deliver the kind of image quality you get from a 110 negative.  To 
approach 35mm quality, you need at least 6 million pixels.

 

It helps a lot in all of this, to understand the limitations of our own 
vision, so let us start with brightness, and think about how much we as 
humans can cope with. We can distinguish about 100 levels of brightness 
and very early in our neural pathway, connecting eye to brain, we split 
the signals into colour and brightness (or in jargon - chrominance and 
luminance). We have three chrominance receptors, roughly sensitive to 
blue, red and green light. Our sense of luminance is derived about 60% 
from our green receptors, 30% from the red and only 10% from the blue. We 
can see luminance detail about 0.2mm wide in an image about 350mm or 14 
inches away (normal viewing distance). We cannot see small details in 
chrominance.

Now for some stuff on photography. A typical colour negative film (such as 
consumer 35 mm film) has silver-halide crystals to sense the light. These 
are (depending on the speed of the film - faster = bigger), in the range 
of 0.5 to 3 micro-metres in diameter. (1 micro-metre, is  one millionth of 
a metre). However there is a lot more to the “graininess” (or as we 
measure it in the business - “granularity”) of a colour film than simply 
the size of the light sensitive elements. The image you finally see is 
formed of tiny dye-clouds, centred on the position of a silver-halide 
crystal that was exposed and developed. These dye clouds are formed from 
many very tiny (sub-micron) oil droplets which were coated with the silver 
halide, and each contains a chemical which reacts with oxidised developer 
to form dye. Dye clouds vary in size and can be generated from a number of 
silver halide crystals, depending on how much exposure there was in an 
area; they can overlap with each other, and of course the packing density 
of the silver-halide grains themselves has an impact on the granularity. 
So, the size of image elements is not easily describable in general terms 
for silver-halide colour films. Another complicating factor in this 
comparison is that the grains are randomly (as opposed to regularly) 
distributed in any given area of the film.

Typical consumer electronic or digital cameras are a little easier - at 
present, these seem to be settling down in the one to one and a half 
million pixels per image. This gives you a start on how much information 
is captured. To sense an image each pixel needs to know about colour and 
about brightness, so there is a certain amount of information to collect 
about each pixel. How much information is collected is in itself variable 
from camera to camera. Typically 256 levels of luminance will be detected 
(note this is over twice the number of levels we can distinguish, although 
that comparison isn’t quite right – our perceived levels are spaced 
logarithmically) as well as spatial sampling data and chrominance data. 
All of this is recorded at some pre-determined level of complexity or “bit-
depth”. More bits mean more discrete levels can be recorded - but the 
image file generated gets bigger. Now to reproduce all of this information 
to something we can recognise requires a bit (no pun intended) of 
translation. To do a good job translating the transmittance of a good 
slide film takes as many as 10,000 discrete levels, and would be done in 
an expensive device or scanner and would generate a very large image file.

The comparison can be made on different levels, depending on what you 
want. For a typical snap and 6” x 4” for example, a consumer digital 
camera can yield results which most people can hardly distinguish from a 
conventional print. It’s when you want bigger images that the differences 
start to show up. Whereas with modern film, enlargements continue to look 
good to quite large sizes, you would need much more information to make 
really good enlargements from a typical consumer digital camera. In fact 
you probably need uncompressed file sizes in excess of 30MB to come close 
to being competitive with film when making enlargements. This just 
reflects the rather large amount of information that traditional film can 
capture in an image. So, it depends on whether you want or need all that 
information. One way of getting digital image files is via a scanner, and 
just for comparison purposes the table below shows how much information is 
typical for a number of scanning applications:

Application	Number of pixels	bits per pixel	bytes per image
35mm colour film	6 million	24	18 million
4” x 5” film	24 million	24	72 million
4” x 6” print	2 million	24	6 million
X-ray film	20 million	12	30 million

So, I think you can see that film remains the lowest-cost for highest 
quality method of capturing images, but it all depends on what you want to 
be able to do with the images. Where money isn’t the main issue 
(professional applications, for example) - but speed of transmission of 
images is, there are very expensive, high quality digital cameras 
available. I doubt if they’d make a lot of sense for holiday snaps, though.

By the way, - I think the digital world opens up so many possibilities for 
us to use our precious images, and I’m very excited by it all. For me, and 
I speak only personally, I find scanning ideal as I can use old images 
just as easily as new, and today scanners are cheap and easily available, 
if you have a decent PC. (You probably have one yourself, you certainly 
have access to one as you are reading this!)