Contradicting Results - EF-S Lens Focal Length Different than a Normal Lens?

After some simple tests with various lenses on my new 40D I was surprised
by the result:

a picture taken with an EF-S 17-85 lens at 75 mm focal length shows
a smaller subject than the exact same scene and settings with an
EF-75-300 lens, also at 75 mm. I've retested the setup by comparing the
EF-S 17-85 at 70 mm and a EF 70-300 mm lens also at 70 mm.

From all discussion on the web and your article on this topic I would have
assumed that even though the EF-S lenses are designed only to fit APS-C
type of bodies, they still behave as a 35 mm lens. Even more, Canon states
on its website that the EF-S lenses should also be multiplied with a 1.6 factor.

Is my understanding correct that due to the shorter distance between the back
of the EF-S lenses and the sensor, the 1.6 crop factor is corrected but that the
DOF effects remain applicable to a 1.6 factor magnification.

I'm still quite puzzled by the results and would appreciate it very much if anyone
could shed some light on the results.

Please see below for my thoughts on what could be causing this:

Quote:

---

Originally Posted by anonymous

After some simple tests with various lenses on my new 40D I was surprised
by the result:

a picture taken with an EF-S 17-85 lens at 75 mm focal length shows
a smaller subject than the exact same scene and settings with an
EF-75-300 lens, also at 75 mm. I've retested the setup by comparing the
EF-S 17-85 at 70 mm and a EF 70-300 mm lens also at 70 mm.

---

I am assuming these are both shot on the same EF-S camera. How much smaller? ie, to what focal length would you have to zoom to on the 70-300mm full frame lens in order to get the same angle of view as the 17-85mm lens? Canon and other camera manufacturers, not to one's surprise, often apply generous rounding when listing the focal length. I would not be surprised if this were just a lens spec issue if the difference is not too huge.

An even more likely reason is this: barrel and pincushion distortion. Although this distortion does not really change the angle of view on a full frame camera, when using a full frame lens on a crop camera it can change the angle of view significantly-- especially with the zoom lenses you have listed. Typically a zoom lens will have substantial barrel distortion when zoomed out--enlarging the center of the frame--and substantial pincushion distortion when zoomed in--shrinking the center of the frame. In an odd way, this means that a 75mm zoom lens on a 1.6X crop camera may not actually have an apparent focal length of 1.6*75=120mm, but may in fact be more like 130mm or even greater. This is an example of where the simple 1.6X rule falls apart, because it assumes that the angle of view is mapped without distortion. This effect could account for any remainder in angle of view discrepancy. Let me know.

As you state, in an ideal world a 75mm lens is a 75mm lens is a 75mm lens-- no matter what crop the camera sensor. Both zoom lenses at 75mm ought to display the same field of view as a 120mm lens would on a full frame 35mm sensor.

Thanks for taking the time to provide some feedback on my crop factor problem. Interesting that you bring up the distortion of the lenses, I had not taken that into consideration yet. However, I've played around with PTLens in the past and now with LensFix and although the EFS 17-85 lens has significant distortion at the wide end but at 75 mm the distortion is quite moderate. I've also checked the distortion of both zoom-lenses which are indeed at their maximum zoom-out focal length but the distortion is insufficient to account for the difference in size in my opinion.

Indeed both lenses were mounted on a EF-S camera.

I did not want to start sending JPEGs to show the magnitude of the difference in size but here are some numbers instead: if I crop the photo to let the subject fill the frame in both pictures, I get the following result:

EF-S lens shot requires 1961 pixels
EF lens shot requires 2747 pixels

This comes down to factor of 1.4 which seems rather large for focal length rounding in EXIF, or the lens distortion effects. But on the other hand it is also rather far from the 1.6 factor. I will repeat the test and check at which focal length on the zoom-lenses I get the same subject size as on the EF-S lens.