Crop factor and f/stop

[Daniel Browning]

for me, shutter speed, itself, is important--not for "camera shake," but for moving subjects.

Agreed. Everything in my post assumed the same shutter speed in all cameras; I wouldn't have it any other way. Shutter speed, like perspective and angle of view, is a critical and fundamental aspect of composition and must be fixed for any comparison to be equivalent.

In that case, the light per unit area per unit time does matter.

Here's why I think it doesn't matter: the larger sensors can be underexposed (increase ISO) and yield the same image as the smaller sensor.


I'm glad you used the S3 in your example. Take a look:





Looks pretty close to me. The 5D is getting much lower light intensity per area, but this is fully compensated by the much higher amount of total area. (Keep in mind that the ISO 80 setting on the S3 would be equivalent to ISO 160 on a 5D, because it only meters for 2.5 stops of highlight headroom. The important thing is that they both have the same shutter speed and DOF.)

If all you photograph are static scenes...

If you use the same iris diameter, then it doesn't matter *what* size sensor or f-number you use: the images will always have the same light, same noise, same DOF, and same diffraction. But there would be no sense in paying more for large sensors if that's how it would be used. So to get any benefit from a larger sensor, one must use a slower shutter speed (that will keep your depth of field, but improve noise, light, etc.), or use thinner DOF.


In ample light, you can shoot the 5D with the same DOF as a digicam, but get more light, less noise. In low light, you have the choice of using thinner DOF than the digicam to get more light.

However, photograph a galloping horse with a 5x7 view camera.

If it was a digital 5x7 view camera, it would be no problem to stop it down to match the DOF of the digicam, use the same shutter speed, and still get the exact same amount of light and noise.

However, in practical terms, a shorter focal length lens gives a smaller image of the horse, so the motion blur isn't as noticeable

I kindly disagree: the motion blur would be the same for all the cameras as long as they have the same distance, same angle of view, and same print size. The smaller focal length on the S3 is only less blurred on the sensor, but once you magnify the sensor 50 times for a large print, the blur becomes visible. The large sensor is magnified much less, so the motion blur comes out equal in the end. (If you print them at different sizes, though, all bets are off, of course.)

As I recall, that's also the reason for the 1/FL guideline, essentially the angular rate as a fraction of the field of view

Yes, agreed.

My confusion is this. Say I go outside on a sunny day, with an
incident light meter, and get an exposure reading of ISO100 at 1/3200
second at f/2.8 And let's say I'm shooting with the three different
bodies: an EOS 1V film camera, an EOS 5D mark II, and an EOS 50D. I'll
say I'm using the 16-35mm f/2.8 L on the full frame bodies, and the
17-55mm f/2.8 IS on the 50D.


My best guess is that if I shoot a photo with each camera, lens wide
open, at 1/3200 second shutter speed and ISO 100, all three would
produce a correct (and identical) exposure on all three cameras,

Exposure is the light intensity (scene luminance, ND filters, shutter speed, f-number, etc.). ISO is not part of the exposure, but it guides you to select the exposure.


By definition, they all have the same exact *exposure*, because they have the same shutter and f-number.


Now, that doesn't mean they all have the same noise. The 5D2 will have less noise than the 50D.If you stop down the 5D2 by 1+1/3 stops (f/4.5), it will have 1.3 stops less exposure than the other cameras. And the brightness on the LCD screen will be less. If you increase the brightness by changing ISO (remember that ISO is not exposure, but it does affect the brightness and noise) to ISO 260, then the 5D2 will again have the same brightness and same noise as the 50D.

despite the variation in field of view, and depth-of-field.

FWIW, I don't think it's very realistic to compare different field of view. If I have a 50D and 50mm lens, then upgrade to a 5D2, I'm not going to stop shooting "telephoto" (80mm-equivalent) focal lengths forever and restrict myself to normal only: I will buy an 80mm lens so I can continue to shoot short tele. What I'm saying is that photographers don't generally let their camera format decide what angle of view they're going to shoot.

Thus, I
would say for the purpose of exposure the lens is an f/2.8 lens
regardless of film or sensor size.

True, but that has nothing to do with the purpose of crop factor. The purpose is to understand what's equivalent.


For the purpose of angle of view, a 50mm is a 50mm regardlness of film or sensor size.


For the purpose of angle of view, a 50mm-on-APS-C is equivalent to 80mm-on-FF35.


For the purpose of depth of field, f/2.8 on APS-C is equivalent to f/4.5 on FF35.


For the purpose of noise, diffraction, total light gathering power, etc., f/2.8 on APS-C is equivalent to f/4.5 on FF35.

Is the aperture adjustment only for depth-of-field calculations?

It's not just depth of field, but also noise, total amount of light, diffraction, approximate lens weight, approximate MTF (in the case of EF-S vs EF), and probably correlated with other factors as well. Size matters to a lot of things.

One other thing: The noise comparisons between different bodies with
the same size sensors (e.g., APS-C, like the 40D vs 50D) but different
pixel densities do show a marked difference in noise for the same
shutter speed, aperture and ISO.

I don't think so. See the recent thread I started:


Myth busted: smaller pixels have more noise, less dynamic range, etc.

A lot depends upon the sensor
technology, as well, i expect, plus the internal processing by the
camera.

JPEG does for sure, but the raw data itself (before conversion) does not vary much with pixel size.

Does this mean that a 16-35/2.8L on a crop body will also have to be
stopped down to F4.5 to get the same exposure as with the EF-S
17-55/2.8 at F2.8 on the same body?

No. The 16-35 f/2.8 does indeed project far more total light than the 17-55 f/2.8, but if you put it on a crop body, all that light will fall on dead spead instead of a sensor, so it is completely wasted.

Wow! I love this forum. It is like
going to school, but for an avocation. I have to go, too. I need to
popsome popcorn and pour some tea! This is fascinating.

I'm enjoying myself too.


The paradigm shift I'm trying to promote is that the *total amount of light* is what really matters, everything else (focal length, f-number, ISO, sensor size, etc.) are just the factors that affect light.


In fact, there is already a very useful shortcut for determining the total amount of light. It used to be called aperture, but the meaning of that word was usurped and lost to most photographers, so now I use a new word: "iris diameter". For a given angle of view and perspective, the iris diameter correlates perfectly to the total amount of light, no matter what the sensor size, f-number, focal-length, or ISO.


Iris diameter also correlates with diffraction, depth of field, and lens weight (loosely).


A 6mm iris diameter on S3 digicam has the same angle of view, focus distance, light gathering power, depth of field, and diffraction as a 6mm iris diameter on fourth thirds, as well as 6mm iris diameter on APS-C, FF35, Medium Format, and even Large Format.


It all comes down to iris diameter.