Bokeh difference - full frame vs crop??

[Redreflex]

I recall reading somewhere on this forum some comment about how a FF sensor would inherently have better bokeh (or was it shallow depth of field) as compared with a crop. I can no longer find it, and I have since read conflicting opinions elsewhere.

Thus, if an image is captured at the same aperture and full frame equivalent focal lengths on both types of cameras with the same lens, would there be a difference in bokeh and/or depth of field with the 2 different camera types?

If so, is it a theoretical or real difference, and why? Any sample comparison images would be very helpful!

 

[neuroanatomist]

There will be a difference in the depth of field, if you frame the shots identically the FF sensor will give a shallower DoF, by a factor of 1.6 when comparing FF to APS-C. So, the APS-C will give a depth of field equivalent to setting the aperture on the FF camera 1.3 stops narrower (because stops are a base-2 log scale, and the base-2 log of 1.6 = 1.3). Put another way, 85mm f/1.2 on a crop body is approximately equivalent to 135mm f/2 on FF, 135mm f/2 on APS-C is approximately equivalent to 200mm f/2.8, etc. Again, this applies if you frame the shots the same on the two bodies with the same lens.

The reason is distance - because 85mm on APS-C yields a field of view equivalent to 135mm on FF, to get the same framing you must be further away with the APS-C camera. That increased distance results in a deeper DoF.

You can prove it to yourself with the numbers, plug whatever you like into DoFMaster's DoF calculator. They also have a page on the effect of crop on DoF.

As a side note, this isn't bokeh. Technically, bokeh is the quality of the OOF blur, and is a property of the lens and the elements in the picture. A crop sensor affects the quantity of OOF blur.

Well, ok, that's a simplification, since a crop sensor can affect bokeh with some lenses - in some cases, vignetting results in OOF blur spots at the edge of the frame that have a cat's eye shape, which is less pleasing than a round shape, and to the extent that using an EF lens on a crop body reduces vignetting, it will result in rounder OOF blur spots, which would be considered better bokeh.

Hope that helps...

 

[Mt Spokane Photography]

Neuro, thats one of the best explanations I've seen. The subject confuses people. The key is that for the same composition and framing, which means a crop camera is further away. If the cameras are at the exact same location, the depth of fild will be the same, but the framing will be vastly different.

Bokeh tends to depend on the lens design more than sensor size, it is affected by the aperture setting.

 

[Jamesy]

What Neuro said

Here are some examples I shot last year with my 40D and a buddies 5D classic. I was most interested in the 40D/50mm vs. 5D/85mm at F2.0 but I also have an example with the 5D/82/f2.0 and 5D/70-200/F4.0.

They are not framed identically but for my purposes, it illustrated the difference between FF and crop.

 

[Woody]

If you want a full blown thesis on the differences between APS-C, FF, 4/3rd etc, you can also take a look here:

http://www.josephjamesphotography.com/equivalence/

 

[87vr6]

Without those pictures being framed with the camera (not post production cropping) identically, the comparison with them is useless.

 

[Redreflex]

There will be a difference in the depth of field, if you frame the shots identically the FF sensor will give a shallower DoF, by a factor of 1.6 when comparing FF to APS-C. So, the APS-C will give a depth of field equivalent to setting the aperture on the FF camera 1.3 stops narrower (because stops are a base-2 log scale, and the base-2 log of 1.6 = 1.3).

That makes sense. In other words for the same aperture setting and framing, FF will always have 1.3 stops shallower depth of field. That's significant!

As a side note, this isn't bokeh. Technically, bokeh is the quality of the OOF blur, and is a property of the lens and the elements in the picture. A crop sensor affects the quantity of OOF blur.

Well, ok, that's a simplification, since a crop sensor can affect bokeh with some lenses - in some cases, vignetting results in OOF blur spots at the edge of the frame that have a cat's eye shape, which is less pleasing than a round shape, and to the extent that using an EF lens on a crop body reduces vignetting, it will result in rounder OOF blur spots, which would be considered better bokeh.

Hope that helps...

Thanks - I did allude to a difference between bokeh and DOF. Vignetting can be corrected somewhat with post-processing.... closer to the centre of the image, what sort of real difference in QUALITY of bokeh are we talking about?

Hope that helps...

Indeed!

 

[ecka]

You have to understand the simple fact that different focal length/aperture combinations produce different background blur. It doesn't matter if you are using a FF or crop camera. APS-C cameras are called "crop" because the result is exactly the same as if you would take the picture using FF camera (same lens, same distance, wider FoV) and crop it down to 2.5 times smaller picture (1.6x at each side). The bokeh will be the same, it's just a smaller picture.

 

[Jamesy]

Without those pictures being framed with the camera (not post production cropping) identically, the comparison with them is useless.

The two examples I was most interested in seeing for myself, hence the test, was 40D/50mm and 5D/85mm and they are roughly the same framing and clearly the background blur on the 5D/85mm is far creamier. It proved the point to me.

 

[neuroanatomist]

Thanks - I did allude to a difference between bokeh and DOF. Vignetting can be corrected somewhat with post-processing.... closer to the centre of the image, what sort of real difference in QUALITY of bokeh are we talking about?

None.

Another side note, though - vignetting correction in post processing 'fixes' the light loss, but certainly will not alter the shape of the OOF highlight spots.

 

[bycostello]

lens construction i.e. number of blades making the aperture makes a difference, sensor size is irrelevant..

 

[neuroanatomist]

lens construction i.e. number of blades making the aperture makes a difference, sensor size is irrelevant..

Not just number and shape of aperture blades, but also the degree to which the lens is optically corrected. For example, Canon designed the 50mm f/1.2L with intentionally undercorrected spherical aberration to increase the quality of the bokeh (and that slightly decreases sharpness wide open for that lens).

Also, sensor size is relevant for the reason I mentioned above - the effect of optical vignetting on the shape of OOF highlights. Vignetting causes cat's-eye shapes at the edge of the frame, and when you use an EF lens on an APS-C body you are cropping away that portion of the image circle. Check out the photozone.de bokeh comparisons for the 50mm f/1.4 on FF vs. APS-C.

 

[NormanBates]

sorry, neuroanatomist, but your math is wrong: you have to multiply by 1.6, not by 1.3
135mm f/2 on APS-C is exactly equivalent to 216mm f/3.2 on full frame, not f/2.8

I know because I've played with my own DoF/FoV calculator:
http://www.similaar.com/foto/dof/dof.php?f_x=1.6&f_r=1.77&f_mm=135&f_f=2&f_d=6
http://www.similaar.com/foto/dof/dof.php?f_x=1&f_r=1.77&f_mm=216&f_f=3.2&f_d=6

and have actually tested it too:
http://www.similaar.com/foto/doftest/doftest.html


oh, sweet revenge!!!
http://www.canonrumors.com/forum/index.php/topic,1677.msg24362.html#msg24362
;D ;D ;D ;D



also, number of blades is relatively irrelevant for bokeh, unless you're worried about bokeh while significantly closing down the aperture on the lens, which is not the most common scenario
my leitz elmarit-R 50mm f/1.4 has amazing bokeh, and the hexagon iris shape is only an issue from f/4 onwards
it is lens design that matters for bokeh quality, and it is definitely a very, very complex issue
I know because I have run A LOT of bokeh tests:
http://www.similaar.com/foto/lenstests/bokehtests.html

 

[neuroanatomist]

sorry, neuroanatomist, but your math is wrong: you have to multiply by 1.6, not by 1.3
135mm f/2 on APS-C is exactly equivalent to 216mm f/3.2 on full frame, not f/2.8

Yep, you're right. I only added a stop, not 1.3 stops.

also, number of blades is relatively irrelevant for bokeh, unless you're worried about bokeh while significantly closing down the aperture on the lens, which is not the most common scenario

Agreed, although I do occasionally shoot portraits stopped down a fair bit to increase DoF when necessary. Even then, it's not just number of blades, because some lenses have the edges of the aperture blades rounded, which makes a difference. In the TDP reviews of the Canon 50mm primes, Bryan includes a bokeh comparison. All are at f/2.8, but the rounded blades of the 50/1.2L (left) retain a circular shape for the OOF specular highlights, while the 8 aperture blades of the 50/1.4 (middle) and the 5 blades of the 50/1.8 (right) are evident.

I know because I have run A LOT of bokeh tests

I've run across your tests before - great work!

 

[Edwin Herdman]

As a side note, this isn't bokeh. Technically, bokeh is the quality of the OOF blur, and is a property of the lens and the elements in the picture. A crop sensor affects the quantity of OOF blur.

You might as well spell out the next step:

Given an identical lens and identical framing, any out of focus areas will appear to be larger (in comparison to the frame) than they would be on a full frame camera. Put another way - if your background has lots of small objects, there will be fewer of them visible; given a pattern, less of it will be visible. If you can frame identically, you could expect to see something a bit like (though not nearly as dramatic) Bryan Carnathan's background blur comparison.

Of course...this exact same effect can be achieved from full frame (minus the higher density resolution in the center) from a full frame camera, if anybody was wondering.

It strikes me as a somewhat poor single reason to go to APS-C; the other losses that change implies might well overshadow any potential benefit of this effect. In conjunction with focal length (and price!), it makes more sense, and seems to help me get some extra utility out of fast primes.

@ Norman Bates:

Interesting analysis of some classic lenses there. Thanks for sharing!

 

[NormanBates]

^: that's only the case because you're keeping the same lens and using a smaller sensor

if you actually change the lens, so the "equivalent" focal length is the same (i.e. 50mm on fullframe, 30mm on APS-C), then out of focus areas will appear to be THE SAME SIZE (in comparison to the frame) than they would be on a full frame camera, but blurred to a lesser degree (i.e. more in focus)

check this test (90mm lens on 1.6x crop camera vs 35mm lens on 3.9x crop camera):
http://www.similaar.com/foto/doftest/doftest.html