DOF FF vs Crop...

[neuroanatomist]

But, considering how small the DOF is with my crop camera and the fast lenses I have, I have no need for the even smaller DOF a FF sensor would provide me. I don't really see FF as having an advantage in this regard.

I do find the shallower DoF with FF to be a benefit, with no real downside at all. I see your point that the DoF of an f/1.2 or 1.4 lens on FF is often just too thin to be useful, and you'd frequently stop down to f/1.6 or f/2 anyway (that's often the case when I use my 85L II). But...I find the DoF of f/2.8 on FF to be excellent for portraits. Now, I can achieve that same DoF with an 85mm prime (even the 85/1.8 woiuld work), but that means giving up some flexibility - my 70-200mm f/2.8L IS II is a a great portrait lens, but there's no such thing as an f/1.8 zoom lens, is there? If I want even deeper DoF, I can just stop the f/2.8 lens down to f/4.5 on FF, and bump up the ISO to compensate for the lost light, and still have a slightly less noisy image. So...more flexibility, and no downside, except cost.

 

[JonJT]

I'm going to try posting this again. Canon Rumors seems to be deleting my posts.

Anyway, FF sensors will have a shallower DOF for a given perspective and aperture. But, considering how small the DOF is with my crop camera and the fast lenses I have, I have no need for the even smaller DOF a FF sensor would provide me. I don't really see FF as having an advantage in this regard.

I think the DOF on crop sensors work perfectly fine. You can still achieve creamy bokeh FF mainly wins in low light situation. They also win in their ability to collect more packets of photons with their 2x surface area.

Yes, this is where I think FF really makes the extra money worthwhile. High ISO operation and tonal range. Hopefully the new generation of crop sensors will close that gap.

 

[JonJT]

But, considering how small the DOF is with my crop camera and the fast lenses I have, I have no need for the even smaller DOF a FF sensor would provide me. I don't really see FF as having an advantage in this regard.

I do find the shallower DoF with FF to be a benefit, with no real downside at all. I see your point that the DoF of an f/1.2 or 1.4 lens on FF is often just too thin to be useful, and you'd frequently stop down to f/1.6 or f/2 anyway (that's often the case when I use my 85L II). But...I find the DoF of f/2.8 on FF to be excellent for portraits. Now, I can achieve that same DoF with an 85mm prime (even the 85/1.8 woiuld work), but that means giving up some flexibility - my 70-200mm f/2.8L IS II is a a great portrait lens, but there's no such thing as an f/1.8 zoom lens, is there? If I want even deeper DoF, I can just stop the f/2.8 lens down to f/4.5 on FF, and bump up the ISO to compensate for the lost light, and still have a slightly less noisy image. So...more flexibility, and no downside, except cost.

Absolutely true.

 

[arad85]

Teehee..

I've just registered to reply to this thread. In ALL the discussions above, no one has mentioned print size. Why is that important? Because DoF is ONLY relevant when you actually render the image so you can see it. Depending on how big you render the image and how closely you view it, the DoF will change.

Huh? But surely DoFmaster gives precise figures - yes it does, but it does based on assumptions of how big you are printing and how far away you are viewing it. The figures aren't there directly, but they are captured by the CoC value. If you look up how CoC is derived, it is obtained from a standard size print from a standard distance and the thing that matters is how much you are magnifying the sensor image to the final print size. Change the assumptions and the DoF changes.

Don't believe me? Take a look here: http://en.wikipedia.org/wiki/Circle_of_confusion How is CoC defined:

CoC (mm) = viewing distance (cm) / desired final-image resolution (lp/mm) for a 25 cm viewing distance / enlargement / 25

Given an output print size, a desired viewing distance and a resolvability measure, the ONLY thing that dictates CoC is the enlargement. Bigger sensor for a given photo -> less enlargement->larger CoC. That's what defines DoF.

Try this. Print an image from a FF camera at 12x8. Now crop it so that it is the same sensor image area as a crop camera and print the resulting image at 12x8. The depth of field will change because you have altered the zoom factor. You are seeing more of the image magnified. Same photo - different DoF because you are magnifying it differently.

 

[neuroanatomist]

Wait, wait...you mean we can print these digital thingies?

I've gone deeper into the CoC concept in other threads, which often just descend into circular arguments that are confusing.

But you're absolutely correct - since DoF is defined by what's sharp vs. what isn't, print size and viewing distance matter.

 

[arad85]

I've gone deeper into the CoC concept in other threads, which often just descend into circular arguments that are confusing.

Why would they be confusing?

The only really confusing thing is why people never refer to rendering the image when they discuss depth of field.... The heated arguments are always about what DoFmaster says but the calculations there rely on assumptions about how big you are going to magnify the resulting image

 

[arad85]

But you're absolutely correct - since DoF is defined by what's sharp vs. what isn't, print size and viewing distance matter.

Teehee.. you added that as I replied

 

[neuroanatomist]

Sorry about that.

Not sure - for some reason, people are ok with sensor size affecting DoF, but have a hard time grasping that cropping an image in post changes DoF, too.

 

[arad85]

I've noticed that too. I'm convinced it's because people see DoF as something that is frozen when the photo is taken, whereas in reality, it's only evident when the image is rendered so we can actually see it.

 

[JonJT]

Teehee..

I've just registered to reply to this thread. In ALL the discussions above, no one has mentioned print size. Why is that important? Because DoF is ONLY relevant when you actually render the image so you can see it. Depending on how big you render the image and how closely you view it, the DoF will change.

Huh? But surely DoFmaster gives precise figures - yes it does, but it does based on assumptions of how big you are printing and how far away you are viewing it. The figures aren't there directly, but they are captured by the CoC value. If you look up how CoC is derived, it is obtained from a standard size print from a standard distance and the thing that matters is how much you are magnifying the sensor image to the final print size. Change the assumptions and the DoF changes.

Don't believe me? Take a look here: http://en.wikipedia.org/wiki/Circle_of_confusion How is CoC defined:

CoC (mm) = viewing distance (cm) / desired final-image resolution (lp/mm) for a 25 cm viewing distance / enlargement / 25

Given an output print size, a desired viewing distance and a resolvability measure, the ONLY thing that dictates CoC is the enlargement. Bigger sensor for a given photo -> less enlargement->larger CoC. That's what defines DoF.

Try this. Print an image from a FF camera at 12x8. Now crop it so that it is the same sensor image area as a crop camera and print the resulting image at 12x8. The depth of field will change because you have altered the zoom factor. You are seeing more of the image magnified. Same photo - different DoF because you are magnifying it differently.

Indeed. But, for argument sake, sensor size would be irrelevant if the enlargement factor were the same AND the pixel count was the same, as well, right?

 

[ghosh9691]

I've gone deeper into the CoC concept in other threads, which often just descend into circular arguments that are confusing.

Isn't that why it's called "Circle of Confusion"? ;D

 

[TexPhoto]

But wait!, is light particles or waves?

Yup, FF menas shallower DOF vs cropped. Yup.

And I go with Wave–particle duality btw.

 

[Jamesy]

I have posted these once before here - they are far from perfect 'lab' shots but they show the rough difference in bokeh between a 5Dc/85_1.8 and my 40D/50_1.8, both at 1/3200, ISO100, F2. Clearly the 5D has way more bokeh.

 

[Cosk]

I'll add to Jamesy's test... Here are two 'equivalent' images:

#1: 135mm f/4 on a 5D
#2: 85mm f/2.5 still on a 5D, but I cropped the center in post.

The camera-to-subject distance was fixed.

The key thing to note is that the dof/blur appears identical between the two focal lengths.

#3 is the 135mm FF at f/2...

Note that a 135 f2 has far more blur than an 85mm f2.5 cropped.

(Unfortunately I didn't do an f2<>f2 comparison, and I ate the tomato)

 

[Jamesy]

I'll add to Jamesy's test... Here are two 'equivalent' images:

#1: 135mm f/4 on a 5D
#2: 85mm f/2.5 still on a 5D, but I cropped the center in post.

The camera-to-subject distance was fixed.

The key thing to note is that the Bokeh appears identical between the two focal lengths.

#3 is the 135mm FF at f/2...

Note that a 135 f2 has far more bokeh than an 85mm f2.5 cropped.

(Unfortunately I didn't do an f2<>f2 comparison, and I ate the tomato)

Interesting - what made you select 85_F2.5 vs 135_F4 - did you just chimp until the bokeh looked similar or did you bust out the calculator?

 

[Cosk]

Interesting - what made you select 85_F2.5 vs 135_F4 - did you just chimp until the bokeh looked similar or did you bust out the calculator?

I busted out the calculator. I didn't really believe the math at first, so I did the test and the math checks out - perfectly. Even a 1/3 of a stop on either side has visibly different bokeh.

85mm x 1.6 = 136mm (I had those primes)
f/2.5 x 1.6 = f/4 (only a few f/stop combos are 1.6x apart)

I also did the test comparing 50(crop) vs 85(ff), and 35(crop) vs 50mm(ff)... which aren't as closely matched as the 85/135... but it was the same conclusion.

Once again, Math wins.


Here is the article on the math (I aspire to one day reading the whole thing...)
http://www.josephjamesphotography.com/equivalence/

 

[skitron]

...but have a hard time grasping that cropping an image in post changes DoF, too.

I think this is because it comes across as a ruse. Anyone who has ever taken an image and cropped it knows the "out of focus-ness" of the background doesn't magically change just because the photo has been cropped, any more than it changes when the photo is shrunk or blown up. It's the math to describe the result of those manipulations that changes, not the pixels themselves, hence the perception of a ruse. JMO of course.

 

[Meh]

I have posted these once before here - they are far from perfect 'lab' shots but they show the rough difference in bokeh between a 5Dc/85_1.8 and my 40D/50_1.8, both at 1/3200, ISO100, F2. Clearly the 5D has way more bokeh.

You can't really have more bokeh. The OOF areas can be more OOF but that is not bokeh.

Bokeh refers to the quality of the OOF parts of an image... how smooth, creamy, uniform, etc. but there is no formal definition. Bokeh is generally (some may debate) thought to be a function of the lens, for example the shape and number of aperture blades contribute to the look of the OOF areas.