DOF head scratching

[dem]

A reviewer on petapixel said the following about the 645z. Please let me know your thoughts on that statement / assumption:

The depth of field is beautiful. On smaller sensor systems the DoF drops off near instantly when using fast aperture tele lenses. On medium format you just get this beautiful roll off to the out of focus regions. I just can’t get enough of this look.

The petapixel reviewer is making a general remark that can be either correct or incorrect depending how you interpret it.

He might mean that if you take a telephoto lens attached to an APS-C camera, you won't be able to reproduce the same smooth sharpness roll off at the same framing. That is certainly true, but why are we talking about telephoto lenses? You can take, for example, a 35 mm f/1.4 lens (as many Fuji X shooters will confirm) and have a similar effect on an APS-C camera. Obviously you will have much more latitude on a medium format, but the fact remains that there is an overlap between the two formats where the sharpness roll off will be identical. Maybe the guy is using the medium format at (equivalent) apertures that are simply not accessible on an APS-C camera. Then the "telephoto" comment becomes redundant.

He is comparing apples to fruit. And he really likes apples

[george013]

George, you very intelligently expressed my question and simultaneously explained those focus prism drawings that accompany dof right ups.

Just to clarify in case we have a misunderstanding. Yes, I want to know what does sensor / medium size have to do with the out of focus areas quality and their relationships to the in focus area, if all or most other factors are kept stable (same effective field of view, aperture width, etc...)

I think you can get the answer from the link Den gave. http://howmuchblur.com/#compare-0.5x...m-wide-subject
It took me some time to understand what it meant.
First of all, comparing different sensor sizes gives always the discussion how do you compare them: equal subject distance or equal subject magnification. The link above is doing the latter one.
The horizontal ax shows the range in the diagram after the subject. In front of the subject doesn't play a part in it. The vertical ax gives the theoretical diameter of the blur disk as a percentage of the image. I'm not sure if I understand that right. I think, not sure, on an equal print is meant.
You can fill in the crop factor. Crop factor is "image width ff/image width second sensor". If you take a format of 60x45 then the crop factor would be 35/60=0.58. We forget about the different ratio. So playing with 3 different lenses you can see the differences of the blur disk.
What I'm missing is the coc, or that part what's in focus. When you calculate it to an average size of a A4, 250mm it is 0.1%. Draw a horizontal line through the 0.1 number, and than you have under it the in focus part behind the subject and above the out off focus part behind the subject.
Funny to play with it. Play with different aperture.
http://howmuchblur.com/#compare-0.5x...m-wide-subject
http://howmuchblur.com/#compare-0.5x...m-wide-subject
I hope I didn't make mistakes.

George

[xpatUSA]

Can this "roll-off" be quantified in any way, other than by subjective opinion?

For example, I could imagine shooting a flat slant-edge target at an angle and measuring the edge response or MTF each side of the center but what would that tell me, other than it gets more blurry toward the edge of the shot, and what would make one shot "better" than another?

[NorthernFocus]

Can this "roll-off" be quantified in any way, other than by subjective opinion?...

By definition DOF and particularly quality of bokeh are subjective. However if one considers the widely accepted methods of measuring/quantifying DOF, then there is an argument to be made that the flatter/narrower that DOF is, the less "roll off". But that pre-supposes my own definition of "roll off" as I haven't seen it defined here. However, I must surely be over simplifying the definition else this thread wouldn't be nearly as long and/or indeed the question would never have been posed to begin with

[xpatUSA]

By definition DOF and particularly quality of bokeh are subjective. However if one considers the widely accepted methods of measuring/quantifying DOF, then there is an argument to be made that the flatter/narrower that DOF is, the less "roll off". But that pre-supposes my own definition of "roll off" as I haven't seen it defined here. However, I must surely be over simplifying the definition else this thread wouldn't be nearly as long and/or indeed the question would never have been posed to begin with

Interesting! I can envision (envisage) a diagram where the x-axis is distance (perhaps normalized to the focal plane = 1) and the Y-axis is "sharpness" (perhaps normalized 0-1). Then anything inside the CoC, however determined, is ideally flat within the DOF and outside of the DOF falls toward zero (rolls off) with increasing and reducing distance.

Is, then, the quality of that roll-off determined by it's shape, I wonder? Would a macro shooter just love the roll-off to be a sharp cut to zero, a box filter so to speak? Would a landscape shooter prefer a slight roll-off at first but which then rolls off faster with increasing/decreasing distance? Or, is a linear roll-off all we can get, like it or not??

[tao2]

Take the following scenario and please answer:

Can you distinguish 3 photos taken using the following theoretical setup?

1) A 4/3 sensor / film equipped with 25mm f/1.4 lens and shot wide open.
2) A 35mm full frame sensor / film equipped with a 50mm f/2.8 shot wide open.
3) A 645 medium format film / sensor (assume same size) equipped with 80mm f/4.5 shot wide open.

I hear opinions on the net saying how differently the DOF "rolls over" between formats and how it is smoother on larger formats compared to the smaller ones? They claim the transition from in-focus to out-of-focus areas is more gradual on the larger formats. Is there truth to that? How can we quantify it if is indeed exists at all?

As an example,

On a FF/35mm camera - if ye take a 135mm lens @f4, focused at 10m., the DoF is 9.48 - 10.58m.

on a 6x6 camera, a 135mm lens @f4, focused at 10m, then its DoF is 9.19 - 10.96m. (the maths - Hyperfocal distance and DoF calculations - is available on the web, somewhere. Mine's comes from memory and books)

Given the larger format of the 6x6, it may be arguable that there appears tae be a "smoother", more gradual transition from in-focus tae out-of focus, but we're down tae millimetres here. All DoF calculations depend on an assumption of a perfect lens, a perfect film/sensor since DoF is assumed tae reach infinity at some point.

The (in)famous softness of older Leica lenses was down tae a degree of spherical aberration, chosen by the lens designer at that time. Aberrations like that are enough tae throw calculations awry as they make a difference tae a disk size at a particular point. Small details can't be resolved at longer distances so where do ye begin comparisons? DoF looks greater if the background has large, easily resolved subjects at a greater distance than small objects closer tae the camera. This may account for some folk thinking that there's a greater, smoother transition.

There is no sudden change from sharp tae unsharp.

DoF depends on focused distance and effective aperture....That's a given.

At a given focused distance, DoF is the same for same-sized objects - when the effective aperture is the same....another given.

But then, considering all the methods ye can use tae alter Dof etc and that all these maths calculations are based on theories, often far from the practicalities of photography, so....?

Ah was about half-way tae responding tae the other thread on this subject then decided life was too short.

So Xeliex.....if ye waken up with night sweats in fear about DoF, then just remember that ye can always set Hyperfocal distance, on everything, and get back tae sleep at night.

[xeliex]

Interesting! I can envision (envisage) a diagram where the x-axis is distance (perhaps normalized to the focal plane = 1) and the Y-axis is "sharpness" (perhaps normalized 0-1). Then anything inside the CoC, however determined, is ideally flat within the DOF and outside of the DOF falls toward zero (rolls off) with increasing and reducing distance.

Is, then, the quality of that roll-off determined by it's shape, I wonder? Would a macro shooter just love the roll-off to be a sharp cut to zero, a box filter so to speak? Would a landscape shooter prefer a slight roll-off at first but which then rolls off faster with increasing/decreasing distance? Or, is a linear roll-off all we can get, like it or not??

Valid and smart questions that probably are challenging to answer for two reasons that pop to mind. One, the subjectivity of the matter and two, the introduction of confounding variables.

It also seems to me from what I have learned and thought of so far, that lens build and characteristics determine a lot of those perceived qualities, perhaps more so than anything related to sensor / medium size, as long as the relevant parameters are equalized (ie: angle of view, absolute opening, etc..).

Correct me if I am jumping to conclusions, but it appears that memes like the "medium format look" are somewhat of a myth since the effect can be duplicated with a fast enough lens on smaller formats; especially when it comes to its relationship with DOF.

I also am assuming that with the inherent technology of modern CMOS sensors, even the subtle gradation of tonalities should be close cross format. Distortion wise, software can minimize or cancel out any barrel distortion. A micrro 4/3 with a 25mm f/1.2 should be able to yield a photo that "feels" the same as 645 size medium format sensor with an 80mm f/2.8, if the latter is shot at f/4.5. By writing this last statement, I can see how stopping down the 80mm to f/2.8 can give more DOF artistic capabilities, but that fades off when comparing it to a full frame with a 50mm shot a say f/1.7...

[NorthernFocus]

...Or, is a linear roll-off all we can get, like it or not??

In which case it's not roll off at all but the slope of a line. The term "roll off" seems suggestive of increasing with distance. A curve in your example. Which may be perceived but doesn't appear to be the case. So considering that, isn't the issue simply difference in DOF between sensors with a given lens/aperture combination? Which of course has been thoroughly discussed/documented.

[xpatUSA]

. . . isn't the issue simply difference in DOF between sensors with a given lens/aperture combination? Which of course has been thoroughly discussed/documented.

Indeed, Dan. Time for me to bow out, I reckon.

[William W]

. . . A reviewer on petapixel said the following about the 645z. Please let me know your thoughts on that statement / assumption:

The depth of field is beautiful. On smaller sensor systems the DoF drops off near instantly when using fast aperture tele lenses. On medium format you just get this beautiful roll off to the out of focus regions. I just can’t get enough of this look.

My first comment is that before making a comment it is worthwhile for one to read any selected quote, in its original context: I read the whole article.

Read in the context of the article, the comment is a general reference to the author’s ongoing description of a (general) look that he wants in his images.

The author is a trained and experienced Technician but also a skilled Wedding Photographer: the article therefore has the possibility to exhibit the conflicts and conundrums when there is the meeting of Mathematics and Art, the conflict of mixing quantitative analysis and qualitative interpretation, without strict lines of reference and delineation.

The article is of course A Review and as such and rightly so has opinion, (a subjective element) which would come from the Artist and not the Mathematician.

In the quote that you selected, the author implies that he can only get that look he is after if he uses a medium format camera, but, if one reads the whole article the author repeatedly implies and also repeated states that the look he is after requires a particular lens.

Certainly it is noted that the particular lens is to be paired with the medium format sized sensor (or film) and I would not dismiss the positive value on overall mage quality of having that larger sensor: but, nonetheless the ‘look’ including the Bokeh that the author wanted, was ultimately found when, after trying many lenses, he mounted the Hasselblad 110mm F/2 on his 645Z camera, viz:

“I believe that with the 645Z and the 110mm f/2, I have gotten as close as I can realistically get to the Contax 80mm look that I can get on digital. The 110mm is a bit of a pain to use on the 645Z but well worth it. . .

I’ve never had the look from a lens resonate with me as much as this 110mm does. It’s truly a magic piece of glass and I’m looking forward to using it at upcoming weddings.”

I think that the author is describing something similar to what I described when I mentioned that I could sometimes see when an EF 135/2 lens was used to make a photograph – because the “look” of that lens is so recognizable to my Viewer’s Eye.

WW

[GrahamS]

This may be of some help:

http://www.amateurphotographer.co.uk...pertures-90260