MTF50 Discrepancy? 70-200 2.8 IS

[canoli]

Unless I'm reading these sites wrong, there is a major difference in the MTF50 results given for the same lens - in this case the 70-200 2.8 IS.


Look at f/2.8 at 200mm:


dpreview shows about 900, and photozone has it close to 1600. They're both in LP/PH, the only difference is DP used a 40D and photozone used a 350D.


I understand DP generally favors Nikon, but the discrepancy shown here is not explained by that. Unless I'm reading the charts incorrectly...?


Is there a reasonable explanation for this? I'd like to be able to trust some of these other sites, not just Bryan's. But if the results vary this much...


Here's the links:


http://www.dpreview.com/lensreviews/widget/Fullscreen.ashx


http://www.photozone.de/canon-eos/199-canon-ef-70-200mm-f28-usm-l-is-test-report--review?start=1

[piiooo]

The dpreview's graph shows line pairs/picture height (900x2=1800) and theone of photozone shows line widths/picture height (1600). I think it isclose enough.

[canoli]

ha! hang it all - I should've caught that. Thank YOU P for catching it and for sharing.


Still...200 units (units?) is a pretty wide margin. I hear ya, "close enough" but ... 200? Of course I can find a 3rd site that'll have it somewhere between I suppose. Maybe those MTF50 tests aren't the most scientific after all?


Anyway, very happy to hear it was my error and not theirs...thanks again!

[Jon Ruyle]

Individual samples of lenses seem to vary. For example, compare Bryan's two samples of the 70-200 f/2.8 IS II. At 200mm f/2.8, I think they look pretty different in the corners. I would not be at all surprised if they are at least 200 points apart.

[piiooo]

Individual samples of lenses seem to vary. For example, compare Bryan's two samples of the 70-200 f/2.8 IS II. At 200mm f/2.8, I think they look pretty different in the corners. I would not be at all surprised if they are at least 200 points apart.

According to this article, MTF paragraph, other variablesthan production samples would belens aperture and object distance. Interesting stuff.

[Jon Ruyle]

Interesting article and point taken. But Bryan's two copies of the 70-200 f/2.8 IS II were compared at the same distance and same aperture.


As for dpreview and photozone, I agree the two 70-200's may have been compared at different distances and this may account for some of the difference (I personally have no idea how much, though).


Plus, nothing against dpreview or photozone in particular, but not everyone is as careful as Bryan. In general, tester error is probably a factor more often than it should be.

[scalesusa]

The values are only valid for the camera body the lens was used on, because that camera and its Bayer filter/AA Filter/sensor/processor are part of the value you see.


The MTF 50 numbers are NOT for the lens itself, they are much higher for just the bare lens, the body drags the MTF down a lot.


Unfortunately, no one posts numbers for just the lens itself, the test equipment is very expensive to test just the lens. The Canon MTF graphs may be from actual barelens tests, but I believe some say they are computer generated.