Comparing FF vs. APS-C

[neuroanatomist]

A long time ago, in a forum very very close.... Jon and I had a lengthy discussion [8-|] about comparing crop vs. FF sensors using lenses that correct for the crop factor (e.g. APS-C with 50/1.2L vs. FF with 85/1.8). The 'sequel' was a discussion with Rick, complete with example images reprocessed by Daniel, comparing the same lens on the 7D vs. 5DII, with the latter cropped to the APS-C angle of view. But, that didn't answer the question of comparing the two cameras with different lenses for taking the 'same' shot.


To satisfy my curiousity, I set up a simple test scene and shot it from the same distance with the 5DII + EF 135mm f/2<span style="color: #ff0000;"]Land the 7D + EF 85mm f/1.2<span style="color: #ff0000;"]LII (the latter having a FF-equivalence of 136mm f/1.92). One rationale for this is that these two combinations are two ways to achieve the same image (same FoV, DoF, and perspective). In our previous discussion (at which time I had the 7D and 85L, but not the 5DII or 135L), Jon's contention was that the FF body and the less expensive lens was the better choice (e.g. at current prices, 5DII+135L is $3600 while 7D+85L is $3850; the difference is more dramatic when comparing the 5DII+85/1.8 vs. the 7D+50L), and the FF body would deliver better IQ. He was right, and ultimately, that discussion was one reason why I added the 5DII to my kit.


Anyway, back to the comparison. Shots were taken wide open and at 2.5 and 5 stops narrower than the max aperture. The 5DII was shot as ISO 400, the 7D at ISO 200 (should have been ISO 160 on the 7D, actually, for the theoretical 1.3-stop difference to compensate for the effect of sensor size on noise - in practice, the 7D is relatively noisier than the 1.3-stop difference in sensor size predicts). Focus was manual with 10x Live View.


Minimal RAW processing was done in DPP to even out slight variations in color temperature and exposure - a custom WB was set using the dropper on a neutral square of theXrite ColorChecker Passport, andexposure was adjusted slightly so the clipping squares (brightest white on the ColorChecker) were all similarly exposed. Picture Style was set to Neutral, and all other corrections (ALO, peripheral illumination, etc.) were turned off.


The first composite shows the full image. The vignetting is evident wide open, and a bit worse with the 5DII + 135L, as you'd expect on a FF body compared to using an EF lens on a crop body.


[img]/cfs-file.ashx/__key/CommunityServer-Components-UserFiles/00-00-00-35-15/5DIIvs7D-Full-Image-Composite.jpg[/img]


The second composite is 100% crops from the center. No scaling was done, so the field of the crop is slightly larger with the 7D, due to the 14% lower megapixel count of the 7D (it's close enough that I think comparisons are possible without upscaling or downscaling both sets of images to the same final resolution). Sharpness and contrast are better across the board with the 5DII+135L. Both lenses improve a bit with stopping down.


[img]/cfs-file.ashx/__key/CommunityServer-Components-UserFiles/00-00-00-35-15/5DIIvs7D-Center-Composite.jpg[/img]


The final composite is100% crops from the upper left corner, with a SpyderCube there to provide a specular highlight and some text to judge sharpness. Wide open, the 7D+85L suffers from softer corners (even on crop) and really noticeable chromatic aberration, which is still evident even stopped down to f/2.8. The 7D's noise is also evident, even at ISO 200.


[img]/cfs-file.ashx/__key/CommunityServer-Components-UserFiles/00-00-00-35-15/5DIIvs7D-Upper-Left-Composite.jpg[/img]


So what have I learned? I've learned that I'm glad I have a 5DII - the IQ is excellent! I do like the flexibility of having the 5DII for portraits and landscapes, and the 7D for more reach and better AF for wildlife and birds. The 135L delivers better IQ than the 85L (consistent with reviews) - still, the 85L on FF is really an amazing and creative tool.


The other thing I've learned is that I'm tired of the New England run of rainy weather! I'd rather be outside shooting the world than indoors shooting test scenes. To that end, there's a break in the rain right now, and I've got my camera in the car, so I'm heading out to take some pictures. []


--John

[NFLD Stephen]

Great comparison John! Thanks for taking the time to set this up and share the results. Unfortunately, this makes me want a FF camera even more!


Stephen

[HDNitehawk]

John


Nice job.


There was really no surprise in your test. It is just as I thought it would be.


I wonder how the 1D IV would compare in a similar test. From hands on experience I think the difference in the5D II and the 1D IVmight be negligible.


Rick

[Jon Ruyle]

Thanks, John, for that. It must have been a lot of work.


The 7D + 85 f/1.2 holds up much better better than I would have guessed. I still stand by everything I said before (the theory part), it's just that the 85 f1.2 is just far more wicked sharp in the center, even wide open, than I knew (I've since purchased an 85 f/1.2, so I'n not as surprised now as I would have been then). I still guess that ff + 135 f/2 is sharper, and if we had 60mp cameras, we would see a bigger difference. But who cares about that.

[Kayaker72]

John,


Thanks for taking the time to perform the comparison. It is very interesting.[Y]


I am trying to decide how much of the quality difference is the lens vs the camera body. Am I right in thinking that much of the blur and CA is the lens but the noise is attributable to the body? I am really impressed by the smoothness of the 5DII, especially between the text. The 7D still looks good, but is noticeably noisier..


Good luck hunting today......it is nice that we've only had one deluge........I am hoping for tomorrow.


Thanks,


Brant

[Daniel Browning]

Thanks for the test, John. If it

[Jon Ruyle]

I am trying to decide how much of the quality difference is the lens vs the camera body.

The body is allowing the 135 to operate at f/2 instead of f/1.2, and making the 135 pixels much bigger.


The CA is the fault of the lens, but using a crop body causes the CA to be magnified as compared to FF. On the other hand, lateral CA will show up more on the edges, which are cropped from the FF.

The 7D still looks good, but is noticeably noisier..

This is strange. In theory all things being equal there should be no difference in noise between FF and crop in this comparison. The 7D has a slightly less noisy sensor (less read noise), so I would have been sure the 7D would come out slightly ahead. Of course John is using ISO 200 instead of ISO160, but I still would have thought the difference between 5D and 7D to be almost zero.


Perhaps the difference has more to do with metering than anything else. To compare photon noise fairly, ISO isn't so important, but both cameras should have the same shutter speed (and ISO should be set to expose reasonably), so perhaps manual mode is best. John- were the shutter speeds similar?

[HDNitehawk]

This is strange. In theory all things being equal there should be no difference in noise between FF and crop in this comparison. The 7D has a slightly less noisy sensor (less read noise), so I would have been sure the 7D would come out slightly ahead. Of course John is using ISO 200 instead of ISO160, but I still would have thought the difference between 5D and 7D to be almost zero.


Perhaps the difference has more to do with metering than anything else. To compare photon noise fairly, ISO isn't so important, but both cameras should have the same shutter speed (and ISO should be set to expose reasonably), so perhaps manual mode is best. John- were the shutter speeds similar?

Jon


I kind of took the side by side test to be a comparison of equal DOF from both lenses and bodies. (F11 vs F6.7 or 6.7 x 1.6 to get to F11). I would think the 7D will have had a higher shutter speed to compensate for the difference. I am sure John will in lighten us.


But as for Noise, in theory it may or may not be true, but I have my own theories. The firmware writers at canon may not work in theory, but marketing. The 7D pictures have so much more room for improvement out of the camera compared to the 5D and 1D it is almost like canon dumbed the firmware down in the 7D so that it wouldn't be as competitive with its big brothers. (this is my conspiracy theory)

[Jon Ruyle]

I would think the 7D will have had a higher shutter speed to compensate for the difference.

The iso was different, too, to compensate.

The firmware writers at canon may not work in theory, but marketing. The 7D pictures have so much more room for improvement out of the camera compared to the 5D and 1D it is almost like canon dumbed the firmware down in the 7D so that it wouldn't be as competitive with its big brothers.

According to DXO Mark's measurements, the 7D has a rating of 854. If you normalize this to full frame, you get 2156, compared with 1815 for the 5DII. Thus the 7D sensor is more sensitive. (And, according to what I've read, it has lower read noise as well.)


I admit, however, that these measured results may differ from real world impressions.

[neuroanatomist]

I would think the 7D will have had a higher shutter speed to compensate for the difference.

The iso was different, too, to compensate.

I chose different ISO's in an attempt to compensate for the effect of sensor size on noise (and chose incorrectly by 1/3 stop). However, to try and keep things on relatively even footing, shutter speeds were kept constant as aperture changed, and were similar on the two cameras - to compensate for the narrower apertures, I increased Speedlite power. Here, too, I goofed...I simply used the C.Fn to set the sync speed to the max, meaning the 5DII shutter speed was 1/200 s throughout and the 7D was 1/250 s throughout - again, a 1/3-stop difference, but unfortunately in the same direction as my ISO choice.


So, in a way, I inadvertently stacked the noise deck in favor of the 5DII by 2/3 of a stop. Still, to me, there is more than 2/3 of a stop worth of extra noise in the 7D images. I'm starting to believe in Rick'sconspiracy theory... Looking at Bryan's noise tests in the 60D review, the 60D appears to have slightly less ISO noise(a subtle difference, but visible),even at lower ISOs, compared to the 7D. This may represent improved technology (despite the fact that the same sensor is used), or may be consistent with Rick'sconspiracy theory - they didn't have to dumb down 60D as much as the 7D in terms of ISO performance, because they dumbed it down in other ways (e.g. no AF microadjustment).