Bokeh comparison 24-70 f/2.8 vs 24-105 f/4

[zerotiu]

okay, now I understand, this is what I get :

24-105 with :

[105mm * f/4 *  3 meters ]


 from subject  will give the same OOF as :

24-70 with :

[60mm * f/2.8 * 2 meters]

* So if I want the DOF more shallow , get closer to the subject but like neuro had said I won't get the whole image, maybe just nose or something else.

I've said it on my comment before that I know that f/4 can't beat f/2.8 in making OOF background. I want to know can f/4 make okay OOF background too so the subject can still be distinguished.

FYI, previously I compare the OOF with my old 18-55 f/3.5-5.6.  With f/4 I still get decent OOF background (it is on my t2i. I believe I'll get shallower DOF on FF body). Then I found someone said that 24-105 produce spikey bokeh but all your images shared is good for me. So 24-105 isn't always produce it.

wow, thx again for the OOF background information

[NormanBates]

Using a FF camera (24mm sensor height) shot in landscape orientation framing a 2m height subject and 100m between the camera and background

(1) 400mm @ 5.6: subject distance : 33.73m, subject to background distance 66.27m
      gives 2.39m DOF, and  0.56mm spot diameter on background

(2) 300mm @ 5.6: subject distance : 25.30m, subject to background distance 74.70m
      gives 2.39m DOF, and  0.47mm spot diameter on background

(3) 300mm @ 4.0: subject distance : 25.30m, subject to background distance 74.70m
      gives 1.71m DOF, and  0.66mm spot diameter on background

as you see, bokeh is a complex issue, it depends on depth of field but also on perspective

you could have also included a 28mm lens, and you'd see that moving your subject closer (to see as much from it as with the longer lenses) and keeping the aperture at f/5.6, DoF is unchanged; but the look of the background changes big time: with a 28mm, you'll see a whole building behind your subject, and it will be out of focus but you'll still know it's a building, and recognize its features; with a 400mm lens, with the background at the same distance, all you see behind your subject is a small patch of wall, so it will not be just out of focus, it will be blurred beyond recognition. So depth of field is the same, but the background will be a lot softer on longer lenses.

I learnt about this when playing with my FoV and DoF calculator:
http://www.similaar.com/foto/dof/dof.php
and it lead me to write these notes:
http://www.similaar.com/foto/dof/dof_notes.html
anc reate this simplified DoF and FoV calculator, which I find much more useful than the usual ones:
http://www.similaar.com/foto/dof/dof2.php

I'm currently running bokeh tests on all my lenses (vintage leitz and carl zeiss jena), making them as scientific as I can; I expect them to be online next week

as with the sharpness tests I run some months ago, I'm learning a lot about my lenses in the process; everybody should run sharpness and bokeh tests on their lenses
(for example, I have two lenses whose bokeh is ugly as hell at maximum aperture, but gets much better slightly stopped down; but that's from tests that neutralize both aperture and perspective issues, so I still don't know if "quantity of bokeh" is enough to counteract this "quality of bokeh" effect)

[awinphoto]

okay, now I understand, this is what I get :

24-105 with :

[105mm * f/4 *  3 meters ]


 from subject  will give the same OOF as :

24-70 with :

[60mm * f/2.8 * 2 meters]

* So if I want the DOF more shallow , get closer to the subject but like neuro had said I won't get the whole image, maybe just nose or something else.

I've said it on my comment before that I know that f/4 can't beat f/2.8 in making OOF background. I want to know can f/4 make okay OOF background too so the subject can still be distinguished.

FYI, previously I compare the OOF with my old 18-55 f/3.5-5.6.  With f/4 I still get decent OOF background (it is on my t2i. I believe I'll get shallower DOF on FF body). Then I found someone said that 24-105 produce spikey bokeh but all your images shared is good for me. So 24-105 isn't always produce it.

wow, thx again for the OOF background information

That's what I figured your question was regarding... the answer is yes, you can get decent OOF background if you need it.  The "Spikey" bokeh is purely subjective and if you like what you have seen photo wise, then you wont be disappointed.  Good luck with your purchase and feel free to let us know if and when you make your purchase and start shooting pictures with your new lens(es)

[WarStreet]

as you see, bokeh is a complex issue, it depends on depth of field but also on perspective

you could have also included a 28mm lens, and you'd see that moving your subject closer (to see as much from it as with the longer lenses) and keeping the aperture at f/5.6, DoF is unchanged; but the look of the background changes big time: with a 28mm, you'll see a whole building behind your subject, and it will be out of focus but you'll still know it's a building, and recognize its features; with a 400mm lens, with the background at the same distance, all you see behind your subject is a small patch of wall, so it will not be just out of focus, it will be blurred beyond recognition. So depth of field is the same, but the background will be a lot softer on longer lenses.

NormanBates, seems we came to the same conclusion. But now I am getting contrasting results. I added the 28mm @5.6 example, and the result of DOF is not similar to the other 5.6 lenses. I compared these results with the DOF master and are similar to mine. I might have a small error in my subject distance estimation, but the real reason for this unexpected result can be found from an article by DOF master.

from DOF master article : http://www.dofmaster.com/dof_imagesize.html

Depth of field is the same for all lenses when the image size is constant and the same f-stop is used.
This rule of thumb is approximately true when the focus distance for the shortest lens is less than about 1/4 of the hyperfocal distance for that lens.

For me, the calculation of the spot diameter on the background is more important. I am currently checking photos from flickr to map these values with the background visual results from the photos. It is a big guess to determine the subject and background distance, but my main aim is to get an idea. Maybe I will try to do some tests with my own photos in a rugby pitch, to get better estimates. It's quit an interesting subject. I wish to find some time reading about it and hoping to understand it better.


Using a FF camera (24mm sensor height) shot in landscape orientation framing a 2m height subject and 100m between the camera and background

(1) 400mm @ 5.6: subject distance : 33.73m, subject to background distance 66.27m
      gives 2.39m DOF, and  0.56mm spot diameter on background
      hyperfocal distance = 952.38m

(2) 300mm @ 5.6: subject distance : 25.30m, subject to background distance 74.70m
      gives 2.39m DOF, and  0.47mm spot diameter on background
      hyperfocal distance = 535.71m

(3) 28mm @ 5.6: subject distance : 2.361m, subject to background distance 97.639m
      gives 3.21m DOF, and  0.06mm spot diameter on background
      hyperfocal distance = 4.57m

(4) 300mm @ 4.0: subject distance : 25.30m, subject to background distance 74.70m
      gives 1.71m DOF, and  0.66mm spot diameter on background
      hyperfocal distance = 750m

[NormanBates]

ok, maybe I went too far, 28mm is already wide angle and things start to change a bit there

with a 50mm, it's basically the same

with my DoF calculator (which seems to use a different CoC value):
400mm f/5.6 @ 33.76m --> DoF=2.00m
300mm f/5.6 @ 25.30m --> DoF=2.00m
200mm f/5.6 @ 16.88m --> DoF=2.00m
100mm f/5.6 @ 8.44m --> DoF=2.03m
50mm f/5.6 @ 4.22m --> DoF=2.12m
35mm f/5.6 @ 2.95m --> DoF=2.25m
28mm f/5.6 @ 2.36m --> DoF=2.43m
24mm f/5.6 @ 2.02m --> DoF=2.62m

http://www.similaar.com/foto/dof/dof.php?f_x=1&f_r=1.5&f_mm=400&f_f=5.6&f_d=33.76


edit:
plus: I think you'll like this:
http://prolost.com/blog/2011/1/9/the-shot-you-can-make.html
http://prolost.com/sycm
http://gallery.me.com/prolost

[WarStreet]

ok, maybe I went too far, 28mm is already wide angle and things start to change a bit there

with a 50mm, it's basically the same

with my DoF calculator (which seems to use a different CoC value):
400mm f/5.6 @ 33.76m --> DoF=2.00m
300mm f/5.6 @ 25.30m --> DoF=2.00m
200mm f/5.6 @ 16.88m --> DoF=2.00m
100mm f/5.6 @ 8.44m --> DoF=2.03m
50mm f/5.6 @ 4.22m --> DoF=2.12m
35mm f/5.6 @ 2.95m --> DoF=2.25m
28mm f/5.6 @ 2.36m --> DoF=2.43m
24mm f/5.6 @ 2.02m --> DoF=2.62m

http://www.similaar.com/foto/dof/dof.php?f_x=1&f_r=1.5&f_mm=400&f_f=5.6&f_d=33.76


edit:
plus: I think you'll like this:
http://prolost.com/blog/2011/1/9/the-shot-you-can-make.html
http://prolost.com/sycm
http://gallery.me.com/prolost

thanks for the examples and the interesting links.

[NormanBates]

My bokeh tests are online, here:
http://www.similaar.com/foto/lenstests/bokehtests.html


I ran several scenarios:
* neutralizing both aperture and perspective:
http://www.similaar.com/foto/lenstests/bokehtests1.html
* neutralizing perspective:
http://www.similaar.com/foto/lenstests/bokehtests2.html
* keeping subject-to-background distance constant:
http://www.similaar.com/foto/lenstests/bokehtests3.html
* keeping camera-to-background distance constant:
http://www.similaar.com/foto/lenstests/bokehtests4.html


I learned a lot doing this

for example:

* I always thought my Carl Zeiss Jena Pancolar 50mm f/1.8 and my Carl Zeiss Jena Sonnar 180mm f/2.8 had great bokeh; it turns out, they both have pretty bad bokeh, but they make up for quality with quantity: the 50mm is relatively fast at f/1.8, and the Sonnar, at 180mm f/2.8, is both long and fast

* vintage glass is a lottery with regards to bokeh too: that Sonnar that came out so bad in my tests is considered one of the mythical "best bokeh ever" lenses:
http://www.rickdenney.com/bokeh_test.htm
(the other possibility is that other people also got fooled by the focal length and speed of this lens, but the fact that my other two sonnars, the 135 and the 300, have much better bokeh, seems to point to my 180 being sub-par)

* another thing that points to vintage glass being a lottery: I expected to come out of this with a conclusion to the tune of "how I learned to stop worrying and just buy Leitz glass"; but the 135mm is actually not stellar: at f/2.8 it's not great, and at f/4 the much cheaper sonnar 135 is actually better

* aperture affects both bokeh quantity and bokeh quality: I hardly ever use the pancolar wide open, because it becomes relatively soft in the corners; stopped down to f/2.8, it is both sharp and pretty nice in terms of bokeh

this bokeh thing is a very complex issue...

next step is to test my father's modern glass and see how it does

[epsiloneri]

Great work! Look forward to your "modern" sample.

[WarStreet]

My bokeh tests are online, here:

* keeping camera-to-background distance constant:
http://www.similaar.com/foto/lenstests/bokehtests4.html

Great work. It's really useful for me, who don't have the energy and  time to do such detailed tests. I can't thank you enough for this.

Due to the vast information of your setup, the spot diameter on the background can be calculated and compared with the related visually result. The camera-to-background test is the one I am interested in as this simulates most sport scenarios.

At first I was surprised with the contrasting results of the 90mm @2.8 and 135mm @2.8
I wrongly assumed that you used an FF camera, which would have given the following results :

  90mm @2.8 : spot diameter 1.96mm , spot diameter with an infinite background distance : 2.92mm
135mm @2.8 : spot diameter 2.21mm , spot diameter with an infinite background distance : 4.38mm

With an FF camera, in your scenario, the 135mm would blur the background more than the 90mm


Using a Canon APS-C camera, which I think it is what you have used in the test ( at least I hope so ! ) :

  90mm @2.8 : spot diameter 0.91mm , spot diameter with an infinite background distance : 1.88mm
135mm @2.8 : spot diameter 0.65mm , spot diameter with an infinite background distance : 2.82mm

With APS-C, it's the 90mm which blurs the background more, and match with your visual results.
At enough background distance, the 135mm will always give more blurring.

Visually the spot diameter of 0.91mm produce a much better result than the 0.65mm. From my previous test (very rough estimates), I concluded that a 0.60mm - 0.70mm should give an acceptable blur, but one would still recognise what's in the background. The 0.91mm results in an unrecognised backgound (at least for me)

This shows why the big lenses in sports are so important. The 400 5.6 would produce some acceptable results, but won't achieve the unrecognised background blurring even when used on FF for a 2m hight subject. On the other hand, with a  portrait orientation, or with a tight crop, this could be achieved.

Thanks again for sharing your work.

[NormanBates]

yes, it's an APS-C camera (a 550D; I think it's said in the main page, but not in the pages with results)

[WarStreet]

yes, it's an APS-C camera (a 550D; I think it's said in the main page, but not in the pages with results)

went through your 4 result pages, but I skipped the main page. Still there was a hint in the result page too, since you are using the ef-s lens, and said that it is the 550D kit lens. Don't know why I assumed an FF camera