[quote name='curriguy' timestamp='1298449360' post='6289']
Hi there,
I'm trying to make a comparison between the Sigma 10-20mm F3.5 for the Nikon APS-C and the Tokina 11-16mm for the Canon APS-C.
I know its not a good idea to make comparisons between two different camera sensors generally, but no review for the Tokina 11-16mm APS-C exists on the Nikon side -- and I'm making do with what information is available. (Personally, I'm a Nikon APS-C shooter.)
[/quote]
It is not a bad idea at all, as long as you keep in mind that the cameras are not the exact same. At least take into account the difference in resolution. Of course, there are other factors like how strong the AA-filter is, which may have some impact on the resolution tests, and vignetting which can give different results due to different tonal curves (dpreview tests vignetting with the standard JPEG output).
[quote name='curriguy' timestamp='1298449360' post='6289']
One difference I noticed is that the Rating Scales are completely different in the two reviews.
The tokina 11-16mm (which is tested on a 8MP Canon) says that "very good" starts at 1550 and "excellent" starts at 1850.
The sigma 10-20mm F3.5 (which is tested on a Nikon 10MP) says that "very good" starts at 1650 and "excellent" starts at 1950.
[/quote]
You can look at the figures in 3 ways, both giving you enough of an idea.
Way 1:
The measurements are of the number of lines that can be "seen" per image height. The EOS 350D has a vertical resolution of 2304 pixels. The D200 has a vertical resolution of 2592 pixels. To get a rough idea, translate those numbers from the Canon to the Nikon vertical resolution.
Lets take as example the resolution figure for the center at 1mm and f4. 2065.5 line widths per image height.
The 350D has 100 / 2592 * 2304 = 88.89% of the resolution of the D200.
Expect the resolution of the D200 with that lens in the same test and settings to be around: 2065.5 / 88.89 * 100 = 2323 LW/PH.
Here you can see that you will not get exact results, because of differences in for instance AA-filter the Canon 350D seems to resolve more than the Nikon ("per pixel"), and as a result the derived Nikon figure is off the charts.
The 2nd way is a bit more reliable, just look at the bars in the charts, and not at the figures. We can see that for for instance f5.6 at 10mm for the Sigma and 11mm for the Tokina, in the center the bars are more or less the same height. Expect the lenses to be equal in that area, concerning sharpness.
A 3rd way is to look at the figures in relation to the chart scales. The same as way 2 but with numbers:
Tokina at 11mm f5.6: 2080.5. Chart max: 2150. 100 / 2150 * 2080.5 = 96.8%
Sigma at 10mm f5.6.: 2174. Chart max: 2250. 100 / 2250 * 2174 = 96.2%
expect the lenses to be equal, in center resolution at 10/11mm at f5.6.
[quote name='curriguy' timestamp='1298449360' post='6289']
So how can I make an educated guess as to how they would compare if they were BOTH shot on the 10MP Nikon? Can I draw any reasonable conclusions?
[/quote]
Simple, just look at the curves the bars make, or the bars themselves, and disregard the scale/numbers. You might wonder if that is correct, because a higher res. sensor surely outresolves a lens sooner? No, not really, that is not how things work. Let me illustrate with an example. The Canon EOS 350D, 8mp APS-C camera. The Canon 50D, 15mp APS-C camera. Same lens, Canon EF 50mm f1.4.
EOS 350D:
![[Image: mtf.gif]](http://www.opticallimits.com/images/8Reviews/lenses/canon_50_14/mtf.gif)
EOS 50D:
![[Image: mtf.png]](http://www.opticallimits.com/images/8Reviews/lenses/canon_50_14_apsc/mtf.png)
We see basically the same curves in the chart. Only 2 things really stand out: The border measurement wide open is lower for the 50D, and the peak resolution at f4 is also a big higher for the 50D. The first probably is due to when lenses start to resolve quite low, you will see an outresolving effect, the second is probably due to lens sample variation.
Another example, the 60mm f2.8 macro with both cameras:
![[Image: mtf.gif]](http://www.opticallimits.com/images/8Reviews/lenses/canon_60_28/mtf.gif)
![[Image: mtf.png]](http://www.opticallimits.com/images/8Reviews/lenses/canon_60_28_50d/mtf.png)
As you can see, the tends for center resolution, border and extreme corner, ar almost carbon copies. Even though one sensor is only 8mp, and the other a whopping 15mp.
So, just look at the bars, not the numbers. You will get enough of an idea how the lenses preform, regardless of sensor resolution.
[quote name='curriguy' timestamp='1298449360' post='6289']
Furthermore, what assumptions should I make about these lenses if I were to move up to a 16MP sensor like the D7000s? Should I expect that these lenses will all uniformly perform worse / better / the same? Or is it a crapshoot?
[/quote]
The same. See my examples from 8mp and 15mp.
I would disregard the Sigma 10-20mm f3.5, and look at the f4-5.6 version (older one) instead. It has the nicer optics.
The Tokina 11-16mm is a very good lens, also on Nikon. Just know it has quite high CA, which should need a right way of correction via RAW shooting and conversion to get the most out of the lens.
Personally I would also look at the Sigma 8-16mm, it goes very wide and is quite an impressive little lens.
