Undergrad Factors that influence depth of field

[Oldhouse]

TL;DR

What are the actual factors that influence depth of field in photography?

I came across a youtube video that discusses some aspects of depth of field in photography/videography. Some things that have been said collided with my understanding of DOF. I then had a long discussion with the creator of the video that unfortunately lead nowhere. I'm no scientist, so I lack a proper understanding of the subject matter. If what is said in the video is correct, I would like to know, so I can learn something from it and correct my misconceptions.
The segment in question starts at 8mins 28s and ends at 9mins 10s (just over 30s in total)



He says "the format size, whether it is full frame, super35, any of that, does not change depth of field", he also shows a lens projecting an image on a wall and then says "it doesn't matter how big the wall is, it hasn't changed the depth of field" because "the image has already been formed, as it goes through the lens, before it hits the sensor" (wall in this case).

Are those statements actually accurate?

My understanding of DOF is that you have to take into consideration the properties of the picture as it arrives on our retina. In order for us to consider a part of the image "in focus", the circle of confusion has to be under a certain threshold. Many factors have an influence on the circle of confusion: When recording the image, the main factors are size of entrance pupil of the lens and distance to subject but also the size of the recording format because it influences how much we have to enlarge the image when viewing it on our TV. When we view the picture on a TV, we have to take the distance from the viewer to the TV into account, the size of the TV etc.

We can't really say anything regarding to the DOF by looking at a picture projected to a wall without taking the other factors into acount. The DOF we get in a picture is different if we record it on a small piece of film or on a large piece of film because we need to enlarge the small piece of film more than we enlarge the large piece of film to display it on our TV. Therefore the statement "it doesn't matter how big the wall is, it hasn't changed the depth of field" because "the image has already been formed, as it goes through the lens, before it hits the sensor" seams to be nonsensical.

Am I actually miss-understanding something here?

 

[A.T.]

My understanding of DOF is that you have to take into consideration the properties of the picture as it arrives on our retina.

Maybe for binoculars etc. But for video and photography it's about how it arrives on the film / sensor.
https://en.wikipedia.org/wiki/Depth_of_field

When we view the picture on a TV, we have to take the distance from the viewer to the TV into account, the size of the TV etc.

When you print out a photo, is the depth of field affected by the size of the paper? Nobody uses the term in the way you seem to understand it.

 

[sophiecentaur]

TL;DR Summary: What are the actual factors that influence depth of field in photography?

I came across a youtube video that discusses some aspects of depth of field in photography/videography.

Have you looked at any other videos on the topic? There will be a lot of dross around obn this topic but I think the problem in the video is that he's told you something with no real explanation, although it seems largely ok to me. His comments about sensor format and a wall are correct because the focus only depends on what's happening local to the part of the picture of interest - if that's the way he defines focus and it's what autofocus works on.

Every step in the picture chain will introduce impairments. That doesn't negate the value of assigning quality parameters at the source system. Any serious broadcaster will take into account the later impairments of the broadcast chain. For instance the quality of studio set design and props can be really poor for 405 line TV; It's easy to spot old material in amongst modern studio shots.

TL;DR Summary: What are the actual factors that influence depth of field in photography?

My understanding of DOF is that you have to take into consideration the properties of the picture as it arrives on our retina.

You'd need to give a source for that. In practice, you could say that you need to take into consideration the viewing conditions and display type when deciding whether source picture quality is good enough but that's a massive modification to what I would think of as being Depth of Field of a camera.

Maybe for binoculars etc.

Agreed; your eye can compensate for a lot of bad focus. The plane of the paper / sensor is fixed so the depth of field is easier to define. (you know when it is or isn't in focus)

 

[Drakkith]

He says "the format size, whether it is full frame, super35, any of that, does not change depth of field", he also shows a lens projecting an image on a wall and then says "it doesn't matter how big the wall is, it hasn't changed the depth of field" because "the image has already been formed, as it goes through the lens, before it hits the sensor" (wall in this case).

Are those statements actually accurate?

That's correct. The size of the sensor/film is irrelevant.

When recording the image, the main factors are size of entrance pupil of the lens and distance to subject but also the size of the recording format because it influences how much we have to enlarge the image when viewing it on our TV. When we view the picture on a TV, we have to take the distance from the viewer to the TV into account, the size of the TV etc.

No, the enlargement of an image has nothing to do with depth of field, nor does the distance between the viewer and the tv.

See the following image:

Notice the two colored vertical lines overlapping the sensor. They represent the spread of light rays of an out of focus object when they hit the sensor. The pink line, being larger than the green line, has more blur, and thus the green light has less blur. A camera using the green aperture would have a larger depth of field than a camera using the larger aperture because the objects out of focus have a smaller blur. That is, there is some amount of 'acceptable blurriness' before it actually becomes noticeable to the eye, and the smaller aperture increases this range by simply reducing the blurriness of unfocused objects (everything not near the point/distance that the camera is focused on).

And that's pretty much all depth of field is. Generally, stopping down the aperture (going up in f-numbers) increases your depth of field by reducing the size of the aperture, and vice versa.

 

[Andy Resnick]

TL;DR Summary: What are the actual factors that influence depth of field in photography?

The video is fine, I only partially disagree with one thing presented- the role of image size. It's important to realize that part of lens design includes the image circle size, something not always appreciated. Often, that subtlety gets folded into the concept of 'crop factor' or "35mm equivalent focal length". For example, large-format lenses have much larger image circles than 35mm format lenses, which in turn have much larger image circles than smartphone camera lenses.

Depth of field does indeed depend on the concept of 'circle of confusion' and is inherently a subjective statement, although it can be more-or-less standardized and corresponds to the markings on a lens barrel. Here is where image format kinda-sorta comes in: the size of the circle of confusion does depend on the size of the image circle, due to, as you note, secondary magnification of the image.

Some 'standard' numbers: the CoC for 35-mm format image circles is 30 microns, while large-format lenses have a CoC closer to 200 microns. Don't know what it is for a smartphone.... but, if I could use a large format lens on a smartphone sensor, the image would look totally blurry. Trying to use a smartphone lens on 8x10 film would result in most of the film not being exposed. If I printed a "perfectly sharp" 35mm format image to fit on a billboard, from a great distance the image would look fine.

A good presentation of DoF formulas based on ray optics is here:

https://www.dofmaster.com/equations.html
https://www.dofmaster.com/dofjs.html

Hope this helps....

 

[Oldhouse]

That's correct. The size of the sensor/film is irrelevant.


No, the enlargement of an image has nothing to do with depth of field, nor does the distance between the viewer and the tv.

I think that is exactly where most people go wrong. Please take a look at the two attached screenshots of a DOF calculator. The only parameter that changed is the sensor size. This results in a change of DOF as you can see. So why is that? Simple... The picture recorded with a micro 4/3 sensor needs to enlarged twice as much as the picture recorded with the full frame sensor. Therefore, the permissible circle of confusion for the 4/3 sensor is half the size as the one for the full frame sensor.
Any comments?

 

[Oldhouse]

When you print out a photo, is the depth of field affected by the size of the paper? Nobody uses the term in the way you seem to understand it.

Yes it is. This becomes important when you take pictures for billboards etc. Resolution, display/paper size, distance between viewer to the screen etc. all play a role.

A very simple explanation of this is as follows: Try to manually focus a picture on a small camera screen, it will be very difficult if not impossible. It will look in focus on your camera screen, but the second you look at it on your large TV screen you will notice that it is out of focus.

Another way to illustrate it is the following: Take a slightly out of focus picture, then scale it down to 1/10 of its original size. You will notice that it now looks perfectly in focus. Display size and resolution matter. BTW, here is something that might sound crazy at first, but it is actually a real thing: If you take a picture, crop it, then scale it back to its original size... the DOF changes.

Here are links with some more info on the topic:
https://damienfournier.co/dof-and-circle-of-confusion/

https://photo.stackexchange.com/que...-of-confusion-of-the-human-eye-to-directly-ob

 

[Oldhouse]

You'd need to give a source for that. In practice, you could say that you need to take into consideration the viewing conditions and display type when deciding whether source picture quality is good enough but that's a massive modification to what I would think of as being Depth of Field of a camera.

Not at all... This is actually not even contested. Of course viewing size, distance to screen etc. influences DOF. Here is some info on the topic... jump to the last part for the conclusion: https://damienfournier.co/dof-and-circle-of-confusion/

 

[sophiecentaur]

Of course viewing size, distance to screen etc. influences DOF

No; they affect the perceived depth of focus. That video seems to deny that but, in the same way, he puts very few actual numbers to his statements about lens behaviour. As advice to beginners, the video is fine and could help people to select how best to spend their fortune but I don;t think we can learn any Physics from what he says.

I'm afraid you need to define depth of field formally before you can make statements about it. Take two objects (pairs of points) at different distances. How would you decide that one is 'in focus' and one isn't?. It's no good just eyeballing the images on our ideal wall / screen. There is no camera system that doesn't do something to tinker with apparent sharpness so how does your wall compare with your top of the range intelligent Nikon camera? The camera will do better for the viewer.

There is one absolute (but arbitrary) way and that could involve the Raleigh Criterion. When one pair of points is resolved by the Rayleigh Criterion, that's the best you can do. (Not really but just say it is). Take the other pair of points and move them away (keeping the same angular separation) from the first object distance until they are not resolved. How imperfect would that have to be? Perfect focus is where the dip in the resulting image is at half value so where would you choose the most shallow dip should be for acceptable focus?

So you can't just say it's all down to the lens without using some actual criterion. We can all muck about with our Raw images in Photoshop and make them look sharper but then, we have to deal with a dark area of the image and we see noise suddenly creep in - just when we thought we'd cracked it.

 

[sophiecentaur]

And that's pretty much all depth of field is.

OK as a comparative measure - describing the visible effect rather than making it formal. There's a lot of talk about the importance of pixel count but when the final image is printed / displayed, it's at least as much to do with the image processing on the way through to the 2K, 4K ,super duper resolution of the display. AI can do a brilliant job to convince you that the £nk+ you spent was really worth while and the basic depth of field is swallowed up in the rest of the channel.

 

[Andy Resnick]

Take two objects (pairs of points) at different distances. How would you decide that one is 'in focus' and one isn't?

That's actually fairly easy to do, if your objects are individual pinholes. If you can see any spatial extent to either of the imaged pinholes, you would say that pinhole is not in focus. That's actually how the 30 micron CoC metric was determined, I believe by Polaroid in the 1960s.

 

[sophiecentaur]

That's actually fairly easy to do, if your objects are individual pinholes. If you can see any spatial extent to either of the imaged pinholes, you would say that pinhole is not in focus. That's actually how the 30 micron CoC metric was determined, I believe by Polaroid in the 1960s.

You can 'buy a model star' for checking the colimation of a telescope. My point was that there doesn't seem to be a simple way to measure the DOP of a lens and but there seems little point, bearing in mind all the other links in the video chain. I always used to be haooy with the little marks against the focus ring of my old 35mm cameras which showed you what to expect of the focus scale. Shaking was always a more important factor for me lol.

I think this thread has put things in some sort of realistic order.

 

[Oldhouse]

No; they affect the perceived depth of focus. That video seems to deny that but, in the same way, he puts very few actual numbers to his statements about lens behaviour. As advice to beginners, the video is fine and could help people to select how best to spend their fortune but I don;t think we can learn any Physics from what he says.

I'm afraid you need to define depth of field formally before you can make statements about it.

In that case, what is the difference between DOF and "preceived depth of focus"?
To my understanding, there is no difference. The depth of field is actually the perceived depth of focus. This is why viewing size, distance to screen etc. influences DOF. and is explained in greater detail here: https://damienfournier.co/dof-and-circle-of-confusion/ . I don't even think that there is any point of contention that the mentioned factors influence DOF. Reference to it can also be found it the wikipedia article here: https://en.wikipedia.org/wiki/Circle_of_confusion

As Andy Resnick pointed out, there is actually a way to determine what is in focus or not. To my understanding, it also involves taking the resolution of the eye into account, which then informs how large the CoC can be so we perceive it as a "single point". The closer you are to the image/TV or the larger the image/TV is, the smaller the CoC needs to be for us to still perceive a "single point" as a single point. Of course this is not an absolute because not every eye is the same etc. but there are specific numbers we settled on for the permissible CoC under specific viewing conditions. The recommended permissible CoC therefore differs when shooting a movie for TV vs for the Cinema.

 

[Oldhouse]

That's actually fairly easy to do, if your objects are individual pinholes. If you can see any spatial extent to either of the imaged pinholes, you would say that pinhole is not in focus. That's actually how the 30 micron CoC metric was determined, I believe by Polaroid in the 1960s.

Yes, and I believe the CoC was calculated for the following viewing conditions: 8"by10" photograph viewed at 14in with perfect vision.

 

[Drakkith]

I think that is exactly where most people go wrong. Please take a look at the two attached screenshots of a DOF calculator. The only parameter that changed is the sensor size. This results in a change of DOF as you can see. So why is that? Simple... The picture recorded with a micro 4/3 sensor needs to enlarged twice as much as the picture recorded with the full frame sensor. Therefore, the permissible circle of confusion for the 4/3 sensor is half the size as the one for the full frame sensor.
Any comments?

After looking into this more, the issue boils down to how DOF is defined:

$DOF \approx \frac{2u^2Nc}{f^2}$

Here $u$ is distance to subject, $N$ is the f-number of the system, $c$ is the acceptable circle of confusion, and $f$ is the focal length of the system. The value $c$, the maximum acceptable circle of confusion, is what is important to understand here. It turns out that the circle of confusion DOES change as sensor size changes. Per wikipedia:

Image sensor size affects DOF in counterintuitive ways. Because the circle of confusion is directly tied to the sensor size, decreasing the size of the sensor while holding focal length and aperture constant will decrease the depth of field (by the crop factor). The resulting image however will have a different field of view. If the focal length is altered to maintain the field of view, the change in focal length will counter the decrease of DOF from the smaller sensor and increase the depth of field (also by the crop factor).

I could be mistaken, but my understanding is this: Take an image. To define the maximum acceptable circle of confusion we need to say how big the image is. If the image is very small, such as when displayed on a small screen or a small piece of paper, then the blurring will have been 'shrunk down' very small when compared to displaying the image on a much larger screen or piece of paper. Imagine an image projected onto the side of a building. If you stand very close then no part of the image will be acceptably sharp and you'll need to move waaaaay back make it look good.

So there is an assumption in that calculator that the two images are being displayed at the same size and being viewed at the same distance. This would mean that the cropped version is having its image magnified by 2x, which increases the diameter of the blur by 2x, which reduces the depth of field since parts of the image that used to be acceptably sharp are now not.

So yes, sensor size is part of what determines the DOF of a system, but in a roundabout, counterintuitive way.

 

[Oldhouse]

After looking into this more, the issue boils down to how DOF is defined:

.....

Ok, thank you. I already thought I'm going crazy here

 

[Drakkith]

Ok, thank you. I already thought I'm going crazy here

Indeed. I wouldn't have thought that the final image size had anything to do with how DOF is calculated, but that's because I was only thinking about the physics of the optical system, not how the image is actually displayed.

 

[Drakkith]

So yes, sensor size is part of what determines the DOF of a system, but in a roundabout, counterintuitive way.

For a little more clarity:

Assume two cameras take an image of the same scene with the same optical system at the same distance, the only difference being that the second camera has a sensor that is half the size of the first. There is an assumption that the final images will both be displayed at the same size, even though the second camera took an image that is only half the size of the first.

That assumption is the crux of the issue. If both images were displayed on the same monitor but at their native resolutions then the DOF of both would be the same. But since there's an assumption that the second image will be magnified to be displayed at the same size as the first then we must admit that this will reduce the DOF since it involves magnifying the image and all the blurring inherent in that image.

 

[gmax137]

I have put off contributing so far, but @Drakkith your post 15 summarizes what I recall from making pictures (and reading photography books) 25 or 30 years ago. Format (today called sensor size) and print size (screen size) are both essential to understanding DOF.

 

[A.T.]

Indeed. I wouldn't have thought that the final image size had anything to do with how DOF is calculated, but that's because I was only thinking about the physics of the optical system, not how the image is actually displayed.

Yes, I also apparently misunderstood DoF as based only the parameters that you actually can control when shooting. Especially in the digital age you have very little control on how the footage will be finally viewed: It could be a flat TV way too big for the room, or a tiny mobile phone.

 

[Oldhouse]

To go back to my original question... What is said in the video between 8mins 28s and 9mins 10s is simply wrong, right? Or am I somehow miss-understanding the video?

 

[sophiecentaur]

After looking into this more, the issue boils down to how DOF is defined:

Exactly. I already made the point that a less 'fussy' channel and display will be less fussy about how accurately you will need to focus your expensive objective lens. If it is claimed that the DOF refers to the overall DOF then you get different DOFs for everyone who is looking (on their own systems) will get a different answer. Is that a useful metric?

DOF is a nicely behaved quantity, for a half decent lens, when you define it in terms of an ideal screen / infinite sensor element density. You can predict it pretty well, just on the geometry. I don't remember old film cameras having different DOF markings for different grades of film. What has changed?

 

[A.T.]

To go back to my original question... What is said in the video between 8mins 28s and 9mins 10s is simply wrong, right? Or am I somehow miss-understanding the video?

Unless there is actual disagreement about the physics, it's just a pointless argument about definitions.

 

[Oldhouse]

Unless there is actual disagreement about the physics, it's just a pointless argument about definitions.

I don't see it as pointless at all. If you make up your own definitions (deviating from the established ones), anything you say could be construed as true... whether it is or it isn't. You couldn't agree on anything... I could simply say the sky is green and would be right (if I make up my own definition that green light has a weave length of 460 nm). The definition of what DOF is clearly established. Therefore, unless I completely miss-understand what is said in the video, the explanations on what effects DOF are completely wrong.

 

[A.T.]

I could simply say the sky is green and would be right (if I make up my own definition that green light has a weave length of 460 nm).

As long as you agree on the measurable wave length spectrum the sky has, there is no disagreement about physics, just about the word "green".

 

[Oldhouse]

If it is claimed that the DOF refers to the overall DOF then you get different DOFs for everyone who is looking (on their own systems) will get a different answer. Is that a useful metric?

Not sure what you mean by that. There is nothing other than the "DOF for everyone who is looking at it on their own systems". There is no such thing as "overall DOF" or any sort of other DOF without taking into account the viewing conditions. In order to make any statement about DOF, you need to define a permissible coc. In order to define a permissible coc, you need to decide on a viewing distance, magnification etc.
We established values for the permissible coc for different recording/viewing conditions. If you choose the right value that match with your circumstances, you get a useful metric.

 

[Oldhouse]

As long as you agree on the measurable wave length spectrum the sky has, there is no disagreement about physics, just about the word "green".

We are going in circles here.... My original question is precisely that: Are the statements true if we use the established definition of DOF? If you want to say that the statements are true but only if you use a "made up" definition of the word DOF that doesn't adhere to the established definition, that is fine... although pointless in my opinion.

 

[sophiecentaur]

Not sure what you mean by that. There is nothing other than the "DOF for everyone who is looking at it on their own systems". There is no such thing as "overall DOF" or any sort of other DOF without taking into account the viewing conditions. In order to make any statement about DOF, you need to define a permissible coc. In order to define a permissible coc, you need to decide on a viewing distance, magnification etc.
We established values for the permissible coc for different recording/viewing conditions. If you choose the right value that match with your circumstances, you get a useful metric.

Have you really thought this through? Once you have established the geometry of the initial image formation there's nothing you can do to change the focus (ignoring AI etc.) A 'photographer' will produce a picture that has an impact, often dependent on differential focus (Sharp object in the background with blurry foreground for instance). If the original image has everything in sharp focus then, without Photoshopping and blur control the original effect of creative focus can't be achieved. He will want some control of this differential focus. Adding degradations will completely lose the wanted focal effects. If a viewer sees a degraded picture then how is their perception of apparent depth of any value at all?

Unless there is actual disagreement about the physics, it's just a pointless argument about definitions.

There is the third aspect of the engineering considerations and practicality. It's not pointless to decide on which definition to use, based on practicality. Next time you watch TV, try to spot where differential focus is used to great effect. Watching feature films on a Smart Phone loses a lot of the impact. IMO, discussing that is not pointless.

At least basing the definition of DOF on front end geometry and optics provides a repeatable measure. That could be the safest choice.

 

[Oldhouse]

Have you really thought this through?

Yes, I have thought this through. I actually have a degree in fimproduction with an emphasis in cinematography and also worked professionally as a cinematographher, so I'm quite familiar how DOF is used creatively.
What DOF is, is clearly defined/established. There are not alternative definitions of it as you suggest. Also, I think you still haven't properly understood what DOF is, because it is not possible to base the definition of DOF on the frontend geometry and optics as you suggest. Drakkith even pointed that out in a previous post when he realized that himself.

 

[sophiecentaur]

Yes, I have thought this through. I actually have a degree in fimproduction with an emphasis in cinematography and also worked professionally as a cinematographher, so I'm quite familiar how DOF is used creatively.
What DOF is, is clearly defined/established. There are not alternative definitions of it as you suggest. Also, I think you still haven't properly understood what DOF is, because it is not possible to base the definition of DOF on the frontend geometry and optics as you suggest. Drakkith even pointed that out in a previous post when he realized that himself.

OK. It's good to hear from someone with actual experience. Perhaps you could give me an example of how that definition (or any definition) is actually implemented (objectively) in practice, say when setting up shots of a scene with two different focus settings, moving from one to another pat of the scene with the camera static. What numbers are used? How is 'focus' checked and also the amount of de-focus in another part of the scene. Does the cinematographer actually measure it or does he/she look at the ranges involved? There will be studio monitors which the director can look at and make a decision at the time, using the shots. But how are the choices of apertures made, other than from past experience? What 'calculations' are done?
At a more general level, I do know that TV vision suites use a monitor which shows the resulting output from the studio. This has to involve the sum of artefacts from scene to broadcast signal. I know that, in the days when (analogue) Colour studios were required to produce a Compatible monochrome output, there was also a mono monitor. In sports programmes, the resolution in the compatible mono picture was what determined the field of the colour cameras. When they finally stopped using 405 lines, there were more players in shot. And so on to 4K.

"What DOF is, is clearly defined/established". Could you be more detailed here and tell me an example with definition and numbers - say when assessing when two objects are both sufficiently well focussed AND when the differential focus between two objects is sufficient. As a creative, I would imagine you would have been through that process. Are you old enough to remember using tape measures for film? There must have been many rules of thumb.

Having looked into astrophotography, i am aware that ther Airey Disc is used when deciding on resolution between to point objects - but the situation is limited to the objective lens and the sensor array and we don't go for de-focus very often. Everything is at infinity, too.