Depth of field of kit lens vs. 2.8 lens

When I read my coauthor's analysis of why he gave up his Canon gear, it got me to thinking about my photography expenses as well.  I don't foresee that I will make a similar move, but I did reexamine my gear to see whether I could trim some fat.  One potential area of savings that I looked into is whether I could replace my Nikkor 24-70 2.8G (blogged here) with the far less expensive Nikkor 28-105 3.5-4.5D (reviewed here).

There are some who would dismiss such a thought outright.  The common wisdom (isn't that an oxymoron?) is that it is foolish to use a cheap lens on a nice body.  I think it's a foolish idea to buy a nice camera (full frame or otherwise) and get a bad lens for it, but a cheap lens is not necessarily a bad lens.  And there is an element of subjectiveness in what constitutes a bad lens.  For example, the bokeh of a lens can be considered bad by many people, but what if the photographer doesn't care about the quality of the bokeh?  Who is to say that that lens is bad for that photographer? 

Getting back to the discussion... among other things, a significant issue with using the 28-105 instead is the depth of field.  The biggest reason I use a full frame camera is to have the option to have a shallow depth of field.  If I couldn't get that shallow DOF because my lens just doesn't have a wide enough aperture, it would defeat the purpose of having a full frame camera in the first place.

It's probably a similar question facing folks who are just moving to full frame for the first time -- whether the kit lens (Nikkor 24-85 3.5-4.5 VR or Canon 24-105 f/4) has a shallow enough DOF compared to a 2.8 lens, or even a 2.8 lens on an APS-C body (such as the Tamron 17-50 VC).  Sure you could use a depth of field calculator but one thing I have learned is that those DOF calculators may be 'technically' accurate but don't give a precise idea of the actual "blurriness" of the background.

In this post, I wanted to put that issue to the test -- how much of a difference in apparent DOF is there between the 24-70 and the 28-105 3.5-4.5D (which has a similar aperture range to the 24-85 VR).  By apparent DOF I mean based on the blurriness of the background, not the range of acceptable focus.  I also tested f/4 for those who are considering the Canon 24-105 f/4 (the kit lens for the Canon 6D) or Nikkor 24-120 f/4G.  I also tested an APS-C body with a 2.8 lens at these focal lengths.




TEST PROTOCOL:
I set up a mannequin head to simulate the subject, and placed a toy block with text about 3 ft. behind it to give us an idea of the blurriness of the background.  I took test shots on a tripod at two distances: about 3 feet from the subject and about 6 feet from the subject.  For the test shots, I used manual exposure and manual flash at constant flash power.  I also manually focused on the near eye.  I adjusted the aperture and for the narrower apertures, I raised the ISO.  In terms of equipment I used a D600 and 24-70 2.8G, and a D300S with the Tokina 11-16 (at 16mm 2.8) Sigma 50-150 (at 50mm 2.8).  With respect to the 24-70, I only adjusted the aperture to see the difference in DOF.  I did not switch to my 28-105.

Here we go:

SIMULATION OF 24-70 2.8G VS. FULL FRAME KIT LENS (3.5-4.5 OR F/4) VS. 2.8 LENS ON APS-C BODY

Part 1: When subject is at about 3 feet.

First, an overview at 24mm:

24mm, f/2.8 on full frame

24mm, f/3.5 on full frame

24mm, f/4 on full frame

16mm (24mm equivalent), f/2.8 on APS-C

Here are magnified versions of the toy block behind the mannequin head:

24mm f/2.8 full frame - cropped view

24mm f/3.5 full frame - cropped view

24mm f/4 full frame - cropped view

16mm (24mm equivalent) f/2.8 APS-C - cropped view

Observations: at an equivalent of 24mm, with the subject at around 3 feet, there is a slight difference in the blurriness of the text on the toy box between f/2.8 and f/3.5.  The blurriness between f/3.5 and f/4 is about the same.  The full frame shots, even at f/4 are noticeably more blurred than f/2.8 on APS-C.

Next, an overview at 70mm at the same 3 foot distance.

70mm full frame f/2.8

70mm full frame f/4.0

70mm full frame f/4.5

50mm APS-C (75mm equivalent) f/2.8

For some reason the 50mm on the APS-C looks much larger than the 70mm on full frame.  50mm is 75mm equivalent so that accounts for part of the difference but there could be other factors such as focus breathing, or possibly the lens' effective focal length is longer than 50mm (FYI I used the Sigma 50-150 on a Nikon D300S).  Because of the dissimilarity of the focal length used, the results for this round are less useful when comparing to APS-C.  Nonetheless, let's proceed.

Here are the magnified versions:

cropped version of 70mm full frame f/2.8

cropped version of 70mm full frame f/4

cropped version of 70mm full frame f/4.5

cropped version of 50mm APS-C f/2.8

Observations:  At this distance, the difference between the backgrounds at the different apertures is much more noticeable at 70mm than at 24mm.  I was surprised that there was a noticeable difference even between f/4 and f/4.5.  The APS-C sample, although flawed, looks about as blurred as the full frame f/4.  I need to confirm this next time with a more similar effective focal length.

Part 2: When the subject is at about 6 feet.

An overview at 24mm:

24mm full frame f/2.8

24mm full frame f/3.5

24mm full frame f/4

16mm APS-C (24mm equiv) f/2.8

Magnified versions of the shots at 24mm:

magnified view of 24mm full frame at f/2.8

magnified view of 24mm full frame at f/4

magnified view of 24mm full frame at f/4.5

magnified view of 16mm APS-C (24mm equivalent) at f/2.8

Observations: there is a slight difference between f/2.8 on full frame and f/4 on full frame.  However, please note these views are very much magnified.  At normal viewing sizes, these differences would probably not be noticeable.  As for the APS-C sample, I screwed up.  It is out of focus, therefore no reliable conclusion can be drawn.

Next, an overview at 70mm:

full frame 70mm at f/2.8

full frame 70mm at f/4

full frame 70mm at f/4.5

APS-C 50mm (75mm equivalent) at f/2.8

As with the samples above at 3 feet, the 50mm sample for APS-C looks like a significantly longer effective focal length, so I would not be able to draw a reliable conclusion.

Magnified views:

Observations:  The results seem similar to those of the 70mm at 3 feet.  At this 6 ft distance, there is a noticeable difference between f/2.8 and f/4, and not much between f/4 and f/4.5.  FWIW, the APS-C sample at f/2.8 looks like the full frame sample at f/4.

CONCLUSION
The difference in apparent DOF between a constant 2.8 lens and a lens with 3.5-4.5 aperture or constant f/4  varies with focal length.  At wider focal lengths the difference is slight.  At longer focal lengths, the difference becomes more noticeable.  With respect to the original question -- whether a kit lens would suffice for shallow DOF -- is a more subjective issue.  For me, the answer is not as clear as I had hoped.  I wanted to replace the 24-70 with the 28-105 and save the money, but I guess I need to think about it some more, although I am leaning toward accepting the 28-105.