How to load film in a 35mm Canon camera
This is actually a question I started asking myself back in the 1990s.
I didnu2019t have a digital camera yet, but I was shooting on 35mm film and scanning to digital.
On most color negative film, I could see obvious grain scanning at 10 megapixels.
So I figured, u201cabout 10 megapixelsu201d was a good goal for a digital camera.
On the other hand, my old Kodachrome 25 slides scanned without grain being visibleu2026 so that suggested there was more to be had, with the right film.
And of course, maybe a better lens than I had with my current system, mostly bought in High School with money from my first business, window washing.
,Based on modern numbers, articles Iu2019ve read, etc.
the general consensus seems to be that very sharp 35mm full-frame film ranges between 12 and 20 megapixels of effective resolution.
And while Iu2019ve read that, that was always the macroscopic view.
Weu2019re comparing digitized resolution charts to those captured by digital cameras.
And that certainly is a great way to figure this out, but it doesnu2019t tell you why.
,But I also knew something was up when Canon came out with the Canon EOS 5Ds, a full-frame DSLR with the first 50 megapixel sensor in that format, close to twice the resolution of anything that came before.
And Canon published a list of recommended lenses for this camera (Canon Releases Recommended Lenses List for EOS 5DS and EOS 5DS R).
Keep in mind, unlike many electronic lens systems today, the Canon EOS system was out in the film era.
This was telling me something I had already figured out: no one had ever made a 35mm film comparable to a 50 megapixel camera.
,Dave Did the MathBack in those days, back in the early 1990s, I u201cdid the mathu201d on this, kind of a back-of-the-napkin guesstimate.
You could look up published specifications on film resolution; Fujifilm, as I recall, was pretty good with this, documenting their expected resolution u2014 it does vary a bit depending on the developer used.
,Based on the highest resolution lens I could find (it was made by Leica, I donu2019t recall the specific lens) and the highest resolution ordinary film (not u201cHigh Contrast Copyu201d film or anything weird), I put it, in round numbers, at about 30 megapixels as a theoretical limit.
And that was the best-case answeru2026 not gear I was going to buy for myself.
And I kind of doubt it was ever possible to get there in realityu2026 or if I actually did the convolutions right in my head, or if my number was just too high.
,Roger Clark Did the MathRoger Clark was a photographer working in scientific imaging at the time he came up with this u201cfilm crossover to digitalu201d chart.
,Heu2019s looking at the images produced by cameras as young as the Canon 1Ds Mark II, which came out in 2004.
Cameras today (2019) are dramatically better in noise, and of course, some have more than doubled the 20 megapixel resolution at the top of his chart.
Heu2019s got some of the best films ever made, resolution-wise, at around 16 megapixels in practical terms.
,Dave Does the ScienceBut this got me thinkingu2026 we can measure digital resolution pretty easily: count the pixels.
But for film, whatu2019s a pixel, exactly? Well, it does turn out thereu2019s a reasonable answer.
,Letu2019s start with black and white filmu2026 the fundamental picture element is the silver halide crystal thatu2019s been turned into metallic silver via the development process, right? And it seems that the typical silver halide particle starts out at around 0.
But there are a couple of issues here.
,The first thing to understand is that, on our current digital cameras, each pixel can represent between 4,096 and 16,384 different levels of light in our image u2014 thatu2019s what we get from a raw image on any DSLR or mirrorless camera.
A silver halide crystal is either turned to metallic silver, or itu2019s not.
Itu2019s on or off.
Itu2019s acutally more u201cdigitalu201d in that sense than our camera sensors!,Next, those silver particles, thatu2019s your film grain, right? Those particles are what we see as grain? Well, I had sort of equated the two, but not exactly.
Just as an existence proof: when we pixel peep film scans, we really wouldnu2019t easily see individual 0.
2u03bcm or even 2.
I saw film grain in most of my 10 megapixel scansu2026 for a 35mm frame, letu2019s call that 4,000 pixels across that 36mm horizontal in the scan, about a 2800dpi scan.
You could fit 18,000 2.
0u03bcm dots across that 36mm frame, if arranged as pixels.
So clearly, thatu2019s not what we directly see as grain or discrete picture elements in a 35mm monochrome negative.
,Consider that we often think of film too much in terms of digital these days when making comparisons.
Itu2019s not the same.
The big difference here is that the suspension of silver halide crystals or the developed tiny chunks of silver in film emulsion isnu2019t a perfectly two-dimensional array like a sensor.
Itu2019s an even distribution but random mix of crystals of different sizesu2026most photographers understand this.
But also, itu2019s a three-dimensional layer of randomly placed crystals, suspended in an emulsion between about 9 u03bcm and 25 u03bcm thick.
,Thatu2019s 3D layering of particles, thatu2019s actually how film works.
You have a several layers of random silver halide particles in the space of any u201cpixelu201d that youu2019d have in the digital world.
Any particle that gets about four photon hits will turn black when developed, any particle that does not stays clear.
If youu2019ve ever taken a shot in low light with a small sensor (or really dark with your good camera), you may have seen lots of noise, more than the high ISO might suggest.
Thatu2019s shot noise, based on the random distribution of photons in the photon flux, some sensors getting more, some less.
Thatu2019s what film is going for here: across about 20u201340 particles, even in good light, some will get photon hits, some wonu2019t, and the overall effect is a mix of offs and ons.
Much like what you see from an inkjet printeru2019s u201cdither cellu201du2014 dots are on or off, but there are lots of them, and together they simulate different levels of grey.
,So whatu2019s grain? Itu2019s clusters of developed silver particles, as seen through the emulsion, that we perceive as a single chunk.
The grain is the effect of your seeing multiple, overlapping bits of developed silver at different layers in the in the emulsion as a single entity, grain clumps as Kodak termed it.
Thatu2019s what we see as the smallest picture element in a black and white photo.
They suggest grain clumps range from 10u201320 u03bcm or so, depending on the film and just plain random distribution.
So that suggests a peak of about 8.
But thatu2019s deceptive, because at higher contrasts, the on vs off crystals are being formed right at the edge of the light/dark boundary.
So while they grain clusters can be largish, they are informed by the behavior of the smaller particles.
,And once we get to color film, the actual picture element is no longer a silver particle.
In the color development process, there are multiple color sensitized layers of silver halide in emulsion that are developed, first as a monochrome imageu2026 modern color film has up to nine such layers.
,Once thatu2019s done, dyes are chemically formed in each layer around the silver particles.
Color films use thinner individal silver halide emulsions, as the final image will depend less on formed silver.
The film has a component called a dye coupler, which is transparent and inert when the film is made.
During development, the dye couple reacts with the developed silver and components of the dye itself to form dye clouds.
These are the fundamental picture element in color film, typically 6u201315u03bcm (Kodak) in size.
So thatu2019s the equivalent of 2400 x 1600 (3.
8 megapixels) to 6000 x 4000 (24 megapixels) in effective resolution.
So that actually tracks my 1990s experience with both typical and exceptionally fine grained color films.
,I donu2019t know that Kodachrome 25 was really capable of something approaching 24 megapixels resolution, but I was certain that it had more to offer than 10 megapixels.
And it probably didu2026 but itu2019s also a bit harder to suss gain in transparency films.
After all, we want them to be transparent, eh? So in the developing process, translucent dye clouds form around silver particles, then the silver is completely bleached away.
Dye clouds do not have distinct edges in the way that silver particles do, particularly when theyu2019re translucentu2026 they fade in density toward the edge of the cloud.
So for scanning, thereu2019s the illusion of more resolution than you actually have, because weu2019re looking through all those overlapping layers and there are not true u201cpixelsu201d that deliver any hard picture element boundary.
But they still can only exhibit the resolution inherent in this smallest feature.
,Oh, and That Pesky LensYou do actually have to consider the lens, and more specifically, the effective resolution of the lens, if you want a more accurate effective resolution for film versus digital.
The visible resolution captured is a convolution of the lens and film/sensor resolution.
,Itu2019s become popular to model this using the Modulation Transfer Function (MTF).
Itu2019s possible to construct an MTF for a lens, for the sensor, for other components in an analog system, and use that to measure effective resolution.
Thatu2019s not usually done for a whole system, but any good lens review today will at least mention MTF data.
Most publish it as well.
,Most comparisons avoid the need to calculate an MTF function for every component in a system, simply by normalizing everything else in the system.
So if you want to compare Canons, you can grab my old EOS RT (heck, you can have it!), load up the film, and then test your comparison Canon (or mirrorless, via adapter) with the same lens.
That lets you compare apples directly to other apples, where u201capple = image capture deviceu201d, or oranges to other oranges, where the lens becomes the orange and youu2019re testing with a reference camera body.
,And now weu2019re back at the original premise: that film vs.
digital has been tested based on current gear.
And film is usually judged to deliver 10u201324 megapixels, doing just such an A/B test setup, with everything else in the system being the same between the two shots.
Lots of reviewers use standardized images, resolution charts, etc.
, allowing comparisons done this way to compare new and archived cameras, where it makes sense.
,And That Speed ThingWhile itu2019s not strictly related to resolution, film speed is indirectly related to it.
When youu2019re shooting in low light, youu2019re going to use a faster film emulsion, and get larger grain clusters.
Or youu2019re going to use a camera and crank up the gain/ISO enough to get the shot.
,One factor here is quantum efficiency u2014 what is the chance of a photon impacting your film or sensor actually contributing to the photo? For film, itu2019s usually well below 10%.
For the early days of digital, a good CCD could give you a 75% QE, specialized CCDs even higher.
For CMOS sensors, the older ones probably only managed 60% QE, but todayu2019s better stacked BSI sensors can achieve over 95% quantum efficiency.
If you have ever asked why your DSLR can shoot at ISO3200 or ISO6400 and still give a decent image, where as back in the film days you might have cringed a little at your results from ISO800 film, thatu2019s why.
,ConclusionI was pretty much convinced that digital had won out over film when I had my APS-C Canon 60D at 18 megapixels.
I didnu2019t change my mind moving to a Canon 6D full frame DSLR at 20 megapixels, nor to a few Micro Four Thirds cameras at 16 and 20 megapixels.
I still own three 12 megapixel cameras that are certainly delivering sharper images than much of the Kodacolor I scanned from days gone by.
,And Iu2019ve been recommending 12 megapixels as a minimum.
Thatu2019s more than you need for an 8x10 enlargement, itu2019s certainly more than you need for online postings.
Itu2019s enough to allow that same image sensor to capture HD or 4K video.
I figured this out quite some time ago, but itu2019s interesting that both camera and smartphone manufacturers seem to have reached much the same conclusion.
,For enthusiasts, on a mirrorless, DSLR, or even higher-end 1-inch u201ccompact digitalu201d camera, you have a good argument for more pixels.
Sure, a few folks want that 50 megapixel Canon, 45 megapixel Nikon, 42 megapixel Sony, or even 100 megapixel Hasselblad, but these are not mortal men.
Most of us are really good with less, but find 20u201324 megapixels right now is kind of a sweet spot.
You actually can crop a photo a bit and still have something usable, and still at least as good as the best films you would ever buy.
If youu2019re not getting sharp images at that point, itu2019s probably your lenses.
Or your technique! One thing that convinced me about Micro Four Thirds u2014 I could afford better lenses, overall, than I had for my Canon.
So I was seeing great shots with either system.
,But for tiny sensor cameras, the 1/2.
3u2033 in most P&S or the 1/2.
6u2033 in the average smartphone, you trade off image quality for resolution once youu2019re not in bright sunlight.
All of the best camera phones from 2016 and 2017 used 12 megapixel sensors, down in resolution from the top rated 16 megapixel sensors of 2015, but up in overall quality.
Olympus made the same decision in their TG5 u201cruggedu201d point and shoot camera (I bought one).
My other pocket P&S camera, the Fujifilm X-F1, also sports a 12 megapixel sensoru2026 Fujifilm figured this out, apparently, five years before everyone else.
,And going the other way, regardless of whether its film or digital, getting a larger image means that you have that many more possible picture elements.
So todayu2019s medium format digital cameras start at 50-100 megapixel depending on the vendor and level-up to 400 megapixel in some models.
And they actually use smaller sensors than the 4cm x 6cm medium format cameras of past ages.
Film is still the go-to for 4u2033 x 5u2033 or 8u2033 x 10u2033 view cameras, a place where digital sensors cannot follow, and that 8x10 on your wall is a frickinu2019 contact print! If Iu2019m claming 20 megapixels worth of modern imaging in a 35mm shot, thatu2019s 56 megapixels in that 4cm x 6cm shot, 300 megapixels in the 4x5 shot, and 1,200 megapixels in the 8x10 shot.
I might need a larger hard drive with that camerau2026