Wallet full of $100 bills

[Keith B]

I thought the way you explained it first, although at first sounded backwards, was dead on. To evaluate the lenses based on one camera.


The 85 1.2 @ 1.2 on a crop acts like a 135 2.0 (136 1.92) but you can clearly see as great as 85 1.2 is, it is a huge trade off to use it wide open on a crop rather than the 135 2.0 on a FF.

Ah, good. I was starting to think I was going crazy

The only thing I think you could test to SEE what the differences between 5DII and 7D would be, set a tripod up, shoot ISO chart with both bodies with the same lens. Not moving the distance and then examine the frames obviously they won't be the same perspective but I don't thin we are talking about that.


Pardon me if I'm off course, this is a hefty post and I have a short attention span.

[neuroanatomist]

The only thing I think you could test to SEE what the differences between 5DII and 7D would be, set a tripod up, shoot ISO chart with both bodies with the same lens.

Keith, is that an offer to perform such a test? [A] You seem to have the necessary components (85L, 5DII, and 7D)...


Actually, photozone.de's Imatest results (testing the MTF of lenses) come close, although you can't 'see' the results as the testing is quantitative, not qualitative like Bryan's. They tested a EF 85mm f/1.2L II on both a 5DII and a 50D (15 vs. the 18 megapixels of the 7D, but pretty close). The data are presented as line widths per picture height (LW/PH), which is a resolution measurement that's a surrogate for sharpness. The results were:


85mm f/1.2L II @ f/1.2 on 50D: 2400LW/PH at thecenter, 2075LW/PH at the border


85mm f/1.2L II @ f/1.2 on5DII: 3157LW/PH at the center, 2120LW/PH at the border


So, from those data, the lens 'seems' sharper on a FF than on a recent crop body at the center - ~30% sharper. At the border, performance is identical (the numbers are±10% or so). Of course, that's partially a reflection of the fact that the picture height of the 5DII's sensor vs. the 50D - the 5DII's sensor is ~18% taller, meaning the relative sharpness of the lens on the two bodies is not very different (in fact, the difference would fall within the ~10% measurement error, meaning from a statistical standpoint, there's probably no significant difference in those measurements). At the borders, the resolution of the 85L on FF is no better than at the borders of the 1.6x crop sensor.


Although the 50mm f/1.2L was not tested on the 50D, it was tested on the 5DII, as was the 85mm f/1.8, which was also tested on the 50D (the two lenses from our previous comparison), and those results are worth mentioning:


50mm f/1.2L @ f/1.2 on5DII: 2473 LW/PH at the center, 1332 LW/PH at the border


85mm f/1.8 @ f/1.8 on5DII: 3082 LW/PH at the center, 2916 LW/PH at the border


85mm f/1.8 @ f/1.8 on50D: 2267 LW/PH at the center, 2183 LW/PH at the border


So, from these and the above results, I conclude the following:

1. The 85mm f/1.8 is an optically excellent lens - as sharp at f/1.8 at the center as the 85mm f/1.2L @ f/1.2 with either the 5DII or the 50D, and the non-L lens is actually sharper wide open at the borders than the L lens wide open on both bodies(a reasonable penalty for being at f/1.2!)
   
   
2. The 85mm f/1.8 has such good resolution, sharpness is not a reason to get the 85mm f/1.2L II regardless of the body you have - in fact, based on sharpness, I'd argue that getting the 85mm f/1.2L II for a FF body is a waste of money, as much as for getting it for a crop body. But sharpness isn't everything - DoF, an extra stop of light, and color/contrast/bokeh are big advantages.
   
   
3. Both of the 85mm lenses are significancy sharper at their respective maximum apertures than the 50mm f/1.2L @ f/1.2.
   
   
4. 'Simulating' the effect of sensor cropping by comparing resolution of the85mm f/1.8 @ f/1.8 on FF to the 50mm f/1.2 @ f/1.2 on 1.6x is just plain unfair (and invalid) because of the dramatic differences in optical performance between those lenses even on the same body - the 85mm f/1.8 is simply sharper wide open than the 50mm f/1.2 (presumably because Canon chose to design the lens for better bokeh at the expense of sharpness).

I can't say for sure what significance these MTF measurements have in terms of 'real world' performance. But, if you compare those numbers to the visual appearance on Bryan's ISO 12233 charts, they are well aligned (for example, the 85mm f/1.8 @ f/1.8 is sharper in the corners than the 85mm f/1.2L @ f/1.2).


If those numbers are at all accurate as a comparable measure of performance of a given lens on different cameras, it suggests the 85L would not be dramatically better on FF than on a 1.6x crop - at least, in terms of sharpness. Likewise, color and contrast should be equivalent. Obviously, there are other factors - DoF and bokeh being huge ones in terms of image characteristics, and as I stated previously regarding the bodies themselves, sensor differences in terms of noise performance are significant.

[Keith B]

Keith, is that an offer to perform such a test? [img]/emoticons/emotion-13.gif[/img] You seem to have the necessary components (85L, 5DII, and 7D)...

Well I won't have the 85L until Monday but I did compare the 35L on both. The pixel density actually makes the 7D's image larger which actually threw me when I opened them to 100% in PS. Then I notice how poorly the ISO chart printed out of my laser printer. I didn't post them due to the quality and confusion. It still gave me an idea of how much sharper the 5DII's image was. The 35L is mind boggling sharp on the 5DII with just a +2 MA. I'll redo it Monday after the 85 arrives or if you want to see the 35L again since it is comparing the sensors, I can reshoot with a dollar bill or something.


It's funny how confusing it is, with the pixel density the small sensors image is actually larger just cropped. This is too much for my wee non-scienctific brain.

[Fast Glass]

Wow! I have beenaway for about a week and this thread is huge! Oh well, time to start reading!


John.

[Keith B]

Okay. I don't think this will ever help anything. I couldn't think of anything better to shoot. The 7D looks better, the pixel density cause magnification over the 5DII. I probably didn't do this the best either. If anyone has any suggestions on how to do this better, I'm willing.


These are exact pixel representation. I can understand now how the smaller pixels exploit lens deficiencies.


5DII


[img]/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.25.81/5DbarcodeC.jpg[/img]





7D


[img]/cfs-file.ashx/__key/CommunityServer.Components.UserFiles/00.00.00.25.81/7DbarcodeC.jpg[/img]

[neuroanatomist]

Okay. I don't think this will ever help anything. I couldn't think of anything better to shoot. The 7D looks better, the pixel density cause magnification over the 5DII.

Thanks for taking the time to satisfy (perhaps) our collective curiosity!


I don't know that I'd say the 7D looks better - but I can't say that it looksworse, either - as I said above, either way, the difference is certainly not dramatic. I'm not saying that FF doesn't have advantages over crop in terms of IQ - but apparently not in sharpness between these two cameras, which is fine. Keep in mind, the sensor technology improved somewhat between the 5DII and the 7D (e.g. gapless microlenses).


Keith, if you turned down the lights and raised the ISO a bit, you'd see a difference in noise (in fact, the 7D image above looks like it already has slightly more noise than the 5DII).


Jon, I'm not sure if this 'convinces' you that the main differences in terms of images here are angle of view and DoF with a given lens, and noise performance. Clearly, the DoF of f/1.2 on FF is never going to be possible on a 1.6x crop body. As I tried to show above, the IQ differences you were highlighting in your example of FF vs. crop (85mm f/1.8 vs/ 50mm f/1.2L) were principally due to inherent differences in the lenses, not the sensors. Keith's test here, with the same lens on both of the relevant cameras, really supports that assertion.


--John

[Keith B]

These were shot at ISO 100. I'll redo this Monday night after I get the 85L. I'LL use a higher ISO also. I'll post pics of micro-adjustment process too. Since I had never had my 35L on the 7D I had to do MA on it order to make it a fair demo. I was kind of taken how less shallow the DOF of 1.4 was on the 7D.


Yeah after looking at them again, it is just that the 7D image is magnified. When I first mention this test I didn't take in account about pixel density and I was thinking the magnification would be the same if the camera didn't move.


I think somewhere back in this thread it was mentioned about barrel distortion differences. I think the difference between a gapless pixel sensor and one with gaps might have slight effect on that (maybe). It seems when the pixels from are slid together they could exaggerate the effect ever so slightly. On a gapless there is no adjusting.

[Jon Ruyle]

Thanks, Keith. Do I understand correctly that these were both shot with the 35L wide open? If so I'm impressed with the IQ of the 35L.


It's hard for me to tell which is sharper. The 7D shows more detail (as expected, just because the image is bigger) but the 5DII looks sharper (again as expected because the chart is smaller). Ideally, one would frame the chart the same way in both cameras, then
resize the larger (in terms of resolution) image to the smaller.


There seems to be a good deal more CA in the 7D image (again, this is expected: when you crop, you magnify CA along with all the other aberrations).


Once again I'm impressed with the IQ of the 35L... the situation here is probably more pixel limited than I would have guessed.

Jon, I'm not sure if this 'convinces'
you that the main differences in terms of images here are angle
of view and DoF with a given lens, and noise performance.

That's right, it does not[]


While it is an interesting
comparison and I appreciate Keith's posting, I don't see this as
relevant to what I was talking about. To me the relevant comparison is
between two systems that take the same pictures (if we find a 22mm f/0.875, we can
put it on the 7D and compare to the 35 f/1.4 on the 5DII [:P] )

As I tried to show above, the IQ
differences you were highlighting in your example of FF vs. crop (85mm
f/1.8 vs/ 50mm f/1.2L) were principally due to inherent differences in
the lenses, not the sensors.

I don't believe that at all. I
still believe that differences in IQ between two high quality primes
(even between a $400 prime and a $1500 prime) are overwhelmed by the
difference in the sensor size. Don't you think that if we reversed the lenses (ie, stopped down the shorter lens and cropped the longer), the stopped down lens would come out ahead?

[Keith B]

Thanks, Keith. Do I understand correctly that these were both shot with the 35L wide open? If so I'm impressed with the IQ of the 35L.


It's hard for me to tell which is sharper. The 7D shows more detail (as expected, just because the image is bigger) but the 5DII looks sharper (again as expected because the chart is smaller). Ideally, one would frame the chart the same way in both cameras, then
resize the larger (in terms of resolution) image to the smaller.


There seems to be a good deal more CA in the 7D image (again, this is expected: when you crop, you magnify CA along with all the other aberrations).


Once again I'm impressed with the IQ of the 35L... the situation here is probably more pixel limited than I would have guessed.

No problem. Both samples are the 35L opened all the way at 1.4. It is a a great, fun lens. That is why I say don't hold your breath for mark II version. IMO, CA is the lens' only weakness and in some instances can be pretty strong.

[neuroanatomist]

As I tried to show above, the IQ
differences you were highlighting in your example of FF vs. crop (85mm
f/1.8 vs/ 50mm f/1.2L) were principally due to inherent differences in
the lenses, not the sensors.

I don't believe that at all. I
still believe that differences in IQ between two high quality primes
(even between a $400 prime and a $1500 prime) are overwhelmed by the
difference in the sensor size. Don't you think that if we reversed the lenses (ie, stopped down the shorter lens and cropped the longer), the stopped down lens would come out ahead?

In that case, we'll have to agree to disagree, I suppose. Look at the photozone.de MTF numbers for these two lenses -50mm f/1.2L @ f/1.2 on5DII: 2473 LW/PHvs. the85mm f/1.8 @ f/1.8 on5DII: 3082 LW/PH (the 85 1.8 is ~25% sharper by the numbers) on the same body. Bryan's ISO 12233 comparison of those two lenses shows clearly that the 85 f/1.8 is sharperon the same body.


Lenses, even 'high quality primes', are different. Hopefully, you can acknowledge that fact - if not, I'm not sure what else it would take to convince you.


Because of the crop factor, you're going to have differences in angle of view with the same lens on different cameras. There's no way to correct for that by using different lenses to compare the systems - different lenses are different lenses.


I suppose you could use a zoom lens on each body, and keep the subject distance the same, using the zoom to 'simulate' the effect of the crop body apparently changing focal length. You'd stop down the zoom to 'simulate' the apparent effect on aperture. If you used a high quality zoom like the 70-200 II, which doesn't get appreciably better as you stop down from wide open, that would be a fair comparison. I'd bet real money that if you did that, you'd see about the same difference that Keith saw with the 35L - i.e., almost none if shot at ISO 100.


The two systems in question have different sensors - they aren't going to take the same pictures. The only reasonable way to perform that comparison is to use the same lens with a different distance from the subject, and live with the fact that perspective and compression are different. That way, you are comparing bodies. You cannot change both lenses and bodies and expect to compare the two systems. You need to perform two independent comparisons each with one constant and one variable. Keith's comparison came pretty close to a 'real' comparison of the two bodies, and confirmed for me that there's very little difference in sharpness between the 7D and the 5DII. As I've said, ISO noise is a clear differentiator.