The Sensor Size Debate

[DUCeditor]

Every once in a while there is a discussion of the advantages and disadvantages of the various DSLR sensor sizes and how and why they have influenced members choice of cameras and systems. In such discussions many of us just throw out jargon under the assumption that other members understand what it is we are talking about. In case that is not true for you I thought I'd start a thread to talk about it.

Now this subject, like so many, raises contradictory opinions even among very experienced photographers. That in part is because the subject is not static. Digital imaging technology is constantly changing making yesterday's "truths" today's "myths." More than that because everything in photography and camera system choice is a matter of trade offs what is a very important issue to one photographer can be a relatively insignificant point to another.

Here then is a quick overview of the subject at least as I see it.

Images, be they created on film or via a digital sensor are made up of bits. The more "bits" you have the more detail that image can contain.

In film those bits were exposed and developed silver halides. And the number of them was determined by several things: The size of the silver halides themselves, how tightly they were packed (the emulsion), and the size of the film. Color required three separate layers, one for each of the subtractive or additive colors (Red, green, blue or yellow, cyan, magenta)

In digital imaging the number of bits of information is determined similarly. In this case it is the number of individual light sensitive components each sensitive to one of the three primary colors.

Here are two illustrations of these 'light catchers' seen in enormous maginification:

Silver halides ( image credit: Eastman Kodak)



Digital sensors (in diagrammatic form)

The limits of resolution of an image caught on film was largely determined by film grain. That is why fine art photographers like Ansel Adams created his images using large, sheet film cameras. Sometimes as large as 8 x 10 inches.

In comparison to that the meager film size of a 35mm camera can be seen as a enormous loss. But the trade off of a small, light, and quick too use camera none-the-less made 35mm the format of choice for most serious photographers. And with time film with such fine "grain" was created that 35mm images could be nothing short of superb.

The quality of a digital image is also influenced by digital's equivalent of film size: Sensor size, and there something similar is happening. Which brings us to the point of this thread.

To understand this debate we first need to understand that the final digital image is made up of many individual bits called "pixels." If those bits are small enough and packed close enough together we no longer see the pixels but instead the picture.

Here is an example using a photo of my home created as part of a lens test series.

First the shot as a whole. The yellow arrow points to a detail - a concrete lawn frog that is being used to hold down a tarp. (sorry that it is not something prettier!)




Now here is a blow up of this minute detail where the pixel size just begins to show...




And here is another so greatly magnified that the pixels themselves start to be clearly seen

These images were taken with a Canon T2i "small sensor" camera using an EF-S 17-85 zoom lens. And they reveal something both fascinating and amazing to those of us who are watching the rapid growth in digital imaging technology: That is how far sensors have come! So much so that the quality of the image is often, as it is here, limited, not by the sensor, but by the lens itself. (And the EF-S 17-85 is a good lens!)

How has this been possible? Basically in the same way it was possible to get such good results from "miniature cameras" (as serious large film photographers once mockingly called the early 35mm Leicas and Nikons). I.e., by reducing the size of the "film grain" (the digital pixels) and learning how to pack these fine grains really, really close to one another.

The Canon T21, for instance, has an 18 megapixel sensor. More than that the sensor in that camera is designed so that there is little space between the individual 'light catchers' thus eliminating the "screen door" look that used to mark most digital images.

But it must be pointed out that pixel count and pixel density are not the sole factors that influence the quality of a final image. There is also light sensitivity and digital "noise." In those areas - especially the later - full sized sensors (meaning those that match the size of the traditional 35mm still photography frame) still have the advantage. The question is: For how much longer?

As the image quality of small-sensor digital cameras catch up to their larger sensor counterparts other factors come into play. One of these is the size and weight of the camera and lens.

Here is a photograph of two such cameras from my own collection. The one on the left is a full-sized (although not actually full sensor size) early generation Nikon digital, a once top-of-the-line D1x. The other is a small sensor Canon T2i. Both have single focal length Macro lenses attached. Notice the difference in their size. And know that the difference in their weight is even greater!

If size and weight matter (as they do very much to me) then one can see a real reason to go with a smaller sensor camera. But why then do larger sensor cameras still have so much appeal to many serious (especially professional) photographers?

Well first, as was earlier mentioned, they do have a small but to some still significant enough advantage in picture quality to be worth their size and weight. But there are two other reasons as well.

One is the fact that they are designed to use and take full advantage of the traditional pro-quality lens sets that were first created for their 35mm counterparts. Those lenses are as expensive as they are superb. If one has invested in a large number of such lenses one naturally wants to be able to continue using them. Indeed, if my own large collection of pro-quality Nikkors had not been made obsolete by being manual focus I'd have likely gone the full-sized sensor route myself for just that reason.

Another reason is build quality. That small-sensor cameras are less worthy of hard professional use has nothing to do with their size - old leicas were equally small as were Pentax's wonderful, fully professional, LX series of 35mm cameras.

No, but it was a design decision made by the manufacturers that has so far relegated the small-sensor camera to the 'amateur' market. Small sensor cameras were designed to appeal to serious and beginning amateurs. And such people do not need a camera body designed to withstand the rigors of summer in the Gobi Desert or all day shoots in the icy tundra. Nor do they want to pay what cameras built to withstand such an environment on a day in day out basis simply have to cost.

So that's my primer on the large sensor/small sensor question. I'm sure others have perhaps very different points of view. And if so I'd love to hear them!

-don

[Bill_Anderson]

Don, thanks for the excellent tutorial. I am constantly debating whether or not to upgrade to a 5D Mark??? for sunset/landscape/train shots and relegate the Rebel XSi to motorcycle touring duty. Right now I don't take enough of those types of photos (compared to the number of bird photos I take with the 7D) to justify the $$$$.

[brewmaster]

Nice work DUCeditor and well explained. Thanks for taking the time to post good info like this here as I appreciate it as surely many others do to.

Cheers, Brew

[Austin748]

Very nicely explained, Don. I've gone though the same thought process many times regarding my photo equipment when moving to digital. Although, in the long run, my driving factor in my choice of the Canon 5D (full-size sensor) was not doing the lens focus length conversions when switching between my film and digital bodies.
Back in the film days, your primary investment was always in the lenses, not the body. After all, the body was just a light-tight box to hold your film. It was always your lenses that all of your images pass through. The few pros I knew would use an expensive body (mostly for the reliability) but have an inexpensive body in the trunk as a backup. Never saw any inexpensive lenses as backups.
Now that we've gone digital, we've placed your decision on "What film?", "What speed?", "Who will develop?", ect. into the choice of which camera body we buy.
And now that tips into your discussion subject, sensor size.
One item that I think should also be discussed is the pros/cons of lenses designed for a particular format (i.e. 35mm sensor or film size) when used on another format body (APS-C or "small sensor").
I'm not trying to argue that a larger sensor is automatically better (it's not), but found that we have tended to discuss DSLR photos in 35mm terms when, in many cases, that's not what is really happening.
I have recently seen much higher quality lenses offered specifically designed for APS-C bodies, such as your recent purchase, so perhaps this subject is slowly dying out. But I had a recent experience with my father purchasing a Canon 60D (APS-C sensor size), a nice, general purpose "S" lens, and an EF-L 70-200/f4 for telephoto work. Trying to explain to him that the 70-200 would not provide the same image on his camera body as mine was amazingly tedious. And I was only trying to explain it so that if/when he went looking for other lenses, he understood that any regular or EF, non "S", lenses would't provide the same image as an "S" lens.

On a simplistic level, it is just a matter of cropping. But if looked at with more detail, using 35mm-size lenses on APS-C bodies has you not using a good portion of the lens, and in many cases, the 'expensive' part(!), since lens performance at the edge of the image tends to be a defining factor for a quality lens. I mean, Hey! I paid for that!

I think I'm trying to say that sometimes it's not all about pixel size and density, but about having your various pieces of equipment work well together and to their optimal (and optical) best.

[DUCeditor]

Good thoughts well expressed Austin 748. That "conversion" thing is really a killer in the beginning. I guess it is like starting to speak and new language. At first you are thinking in the old one and translating, but in time you actually start to think in the new.

I went through something similar when I changed over to largely 6x7 (cm.) format film in my commercial work. In the beginning none of the numbers made sense. Than they did. I had learned the 2nd language.

Now let me throw one other loop at you... You point out, quite correctly, that when you use a lens made for 35mm or large sensor on a smaller sensor camera you are "throwing away the best part." And you have a point. Corner to corner sharpness for instance. On the small sensor camera those "corners" are lighting up the mirror box, nothing more. (I never hear anyone speak about what that does to contrast. hmmm....)

But there is more. Since digital sensors read from a relatively narrow arc (in comparison to film, which is almost totally non-directional) a "fast" lens is often no faster than a "slow" one. That because the outer perimeter of the lens - the very thing that makes it a more effective light gatherer -- is sending its light toward the "film plane" (wow, what do we call it now, the "sensor plane"?) at an oblique angle where it has little effect on exposure.

Weirderer and weirderer, eh? At least to us older imagers. To the young it is simply what is.

Lucky them!

(God, I love this stuff!)

-don

[SS904]

Interesting thought about fast lenses and the smaller sensors. Since most, really all, fast lenses have a tendency to vignette at the widest apertures with 35MM format (film or digital), I wonder if it really is effecting the lenses light gathering properties and subsequent EV when used on a smaller sensor. In effect, the smaller sensor is using the sweet spot of the lens: The bright sharp center, while cropping off the corners which have a tendency to be darker and soft wide open. I don’t imagine you have that advantage with the EF S line, or 3rd party DX glass. The image circle is smaller, so you are using more of what’s being projected and are maybe back to a circumstance where you see the vignetting and soft corners. I don’t have any EF S or DX lenses, so I can’t compare.

[DUCeditor]

Originally Posted by SS904

The image circle is smaller, so you are using more of what’s being projected and are maybe back to a circumstance where you see the vignetting and soft corners.

Absolutely correct. My 17-85 has significant fall off at the very widest focal length(s), and then little as it approaches "normal" and moves on to telephoto.

But again technology comes to the rescue. Several available applications, including Canon's own Digital Photo Professional, reads the data recorded with the image and using the RAW image as a base and algorithms that match the lens' measured and known properties, corrects for not only fall off but also geometric distortion and chromatic aberrations.

The effect is nothing less than phenomenal!

-don

[Austin748]

Originally Posted by DUCeditor

Good thoughts well expressed Austin 748. That "conversion" thing is really a killer in the beginning. I guess it is like starting to speak and new language. At first you are thinking in the old one and translating, but in time you actually start to think in the new.

I went through something similar when I changed over to largely 6x7 (cm.) format film in my commercial work. In the beginning none of the numbers made sense. Than they did. I had learned the 2nd language.

Completely agree, with an added dimension... I too have 6x7 camera gear and love (I do mean really, really enjoy) shooting with it. And I was confronted with the fact that the numbers made no immediate sense to me. (Instead of 50mm being "normal" perspective, 90mm was considered normal...If I remember correctly...) But I had an advantage when I started with 6x7. I didn't expect it to relate to 35mm photography. I mean, my image was backwards! And I looked down at it, not towards it!
So, to me, shooting 6x7 was a whole new world and I didn't even attempt to translate.

But holding a DSLR with a lens and a shutter button where I expected it to be... comfortable and familiar controls... my translation was happening regardless. The smaller sensor just didn't work well for me.

[Austin748]

Originally Posted by DUCeditor

Now let me throw one other loop at you... You point out, quite correctly, that when you use a lens made for 35mm or large sensor on a smaller sensor camera you are "throwing away the best part." And you have a point. Corner to corner sharpness for instance. On the small sensor camera those "corners" are lighting up the mirror box, nothing more. (I never hear anyone speak about what that does to contrast. hmmm....)

But there is more. Since digital sensors read from a relatively narrow arc (in comparison to film, which is almost totally non-directional) a "fast" lens is often no faster than a "slow" one. That because the outer perimeter of the lens - the very thing that makes it a more effective light gatherer -- is sending its light toward the "film plane" (wow, what do we call it now, the "sensor plane"?) at an oblique angle where it has little effect on exposure.

Weirderer and weirderer, eh? At least to us older imagers. To the young it is simply what is.

Lucky them!

(God, I love this stuff!)

-don

I really didn't think of 'fast' lenses as providing a 'wider' image. It produces a 'brighter' image, across the entire "image plane" (new term. ) I never bought higher quality (read "faster") lenses to brighten up the edges or be sure to not have an image vignetted. If I had a lens that vignetted the image, at any aperture, I would have dismissed that lens as crap, or at least 'undesirable' due to the caveats . There are a few instances where a high quality lens, the 50mm f1.0L comes to mind, that had the capability to vignette an image. But I never needed to add that type of capability (with noted limitations) to my photo bag.

But really, your discussion, at the point of your comment above, really deals with the characteristics of the film/sensor to a much greater extent than the quality of the lens. I honestly have never really looked at the sensitivity of the film/sensor vs. the angle of incidence of the image. It would be an interesting discussion that encompasses the focal distance from lens to image plane.

I really haven't considered this in my "image quality" thoughts (through omission only).
I wonder what the 'image angle of incidence' is for each of 6x7, 35mm, and APS-C with a wide-angle lens (thinking this as the most extreme situation for each).

Interesting!!

Greg

[DUCeditor]

Originally Posted by Austin748

I honestly have never really looked at the sensitivity of the film/sensor vs. the angle of incidence of the image. It would be an interesting discussion that encompasses the focal distance from lens to image plane.

Greg, I can't take credit for that observation. That has to go to Frederic Guichard at DxO Labs who made the observation and published the findings. I read them in an article by Mason Resnick entitled "50mm Lens Shoot-Out: f/1.8 or f/1.4" which can be found in the Adorama.com Learning Center.

Originally Posted by Austin748

I really didn't think of 'fast' lenses as providing a 'wider' image. It produces a 'brighter' image, across the entire "image plane" (new term. )

That is true. But it is accomplished by the use of larger objectives. To see this just think of the other extreme - the pin-hole camera. There the entire source of light comes from the small orifice centered over the film plane, uh... "image plane." ;-) A very fast lens is just the opposite and there is where angle on incidence comes in.

Well, at least that is how I understand Mr. Guichard's observation.

don